Annual Reviews in Control 51 (2021) 172–205

Contents lists available at ScienceDirect

Annual Reviews in Control

journal homepage: www.elsevier.com/locate/arcontrol

Integrated chassis control: Classification, analysis and future trends

Victor Mazzilli a, Stefano De Pinto b, Leonardo Pascali b, Michele Contrino b,
Francesco Bottiglione c, Giacomo Mantriota c, Patrick Gruber a, Aldo Sorniotti a, *
a
Department of Mechanical Engineering Sciences, University of Surrey, GU2 7XH, Guildford, UK
b
McLaren Automotive Limited, McLaren Technology Centre, GU21 4YH Woking, UK
c
Dipartimento di Meccanica, Matematica e Management, Politecnico di Bari, 70126 Bari, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: Integrated Chassis Control (ICC) is one of the most appealing subjects for vehicle dynamics specialists and re­
integrated chassis control searchers, due to the increasing number of chassis actuators of modern human-driven and automated cars. ICC
architecture ensures that the potential of the available actuators is systematically exploited, by overcoming the individual
reference performance
limitations, and solving conflicts and redundancies, which results into enhanced vehicle performance, ride
coordination strategy
comfort and safety. This paper is a literature review on ICC, and focuses on the topics that are left uncovered by
control allocation
control structure the most recent surveys on the subject, or that are dealt with only by old surveys, namely: a) the systematic
vehicle dynamics categorisation of the available ICC architectures, with the critical analysis of their strengths and weaknesses; b)
performance indicator the latest ICC approaches, which are becoming feasible with modern automotive microcontrollers; c) the driving
performance requirements; and d) the procedures to objectively evaluate ICC performance. The manuscript aids
the interested reader in the choice of the most appropriate ICC method for the specific requirements, and con­
cludes with the recent developments and future trends.

sensor and information sharing, which is drawing major attention,

especially for automated driving applications (Schmidt and König,
1. Introduction 2020). ICC can be implemented to cover a single performance domain
(single-objective ICC), or multiple domains (multi-objective ICC).
Passenger cars are equipped with an increasing number of chassis According to Shibahata (2005), the first generation of ICC systems
actuators, e.g., electro-hydraulic or electro-mechanical friction brakes, was implemented in the mid-1980s, with the development of 4WS (also
active and semi-active suspensions, and 4-wheel-steering (4WS), which called rear-wheel-steering, RWS, in the literature), under the initiative
improve vehicle dynamics and active safety performance. In specific of Japanese automotive manufacturers. The second generation of ICC
vehicles, also the powertrain acts as a chassis actuator, as it allows systems was developed in the 1990s, with the introduction of
controllable wheel torque distribution, i.e., the so-called torque- control-by-wire technology, e.g., for braking, traction and steering
vectoring (TV). Modern chassis actuators tend to be packaged into (active front steering, AFS). The third ICC generation started its devel­
complex mechatronic systems, including sensors, dedicated control opment in the early 2000s, and includes cooperative control technolo­
units, and communication interfaces, and therefore can be considered gies for intelligent vehicles, e.g., employing radars, cameras and
‘smart’ actuation systems. external sensors.
Fruechte et al. (1989), Heißing and Ersoy (2011), and Reuss et al. The previous reviews on ICC, e.g. Duffie et al. (1988), Fruechte et al.
(2014) group chassis actuation systems according to their vehicle per­ (1989), Gordon et al. (2003), and Kissai et al. (2017), list the main
formance domain of interest, e.g., longitudinal dynamics, lateral dy­ desirable features, valid also for the upcoming generation of automated
namics, and attitude/ride. The presence of multiple actuators allows vehicles (Schmidt and König, 2020):
their coordination, commonly referred to as integrated chassis control
(ICC), which can: i) compensate the limitations of the individual actu­
ators; ii) exploit synergies; and iii) reduce cost and complexity through

Notes: the notation ‘⋅’ indicates a time derivative, while ‘⋅⋅’ indicates a second order derivative. The subscript ‘i = F,R’ indicates the front or rear axle. The subscript
‘j = L, R’ indicates the left or right side of the vehicle.
* Corresponding author.
E-mail address: [email protected] (A. Sorniotti).

https://doi.org/10.1016/j.arcontrol.2021.01.005
Received 27 October 2020; Received in revised form 24 January 2021; Accepted 26 January 2021
Available online 28 April 2021
1367-5788/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

List of symbols MMij Motor torque

MM,max Maximum motor torque
Aψ̇ Yaw dynamic amplification Mtot z,max,act Maximum actuator torque potential
ahyp Hyperbolic function constant Mz Direct yaw moment
ax Longitudinal acceleration Mcontr
z Controllable yaw moment
ay Lateral acceleration Mtot z Total yaw moment
ay,lim Lateral acceleration rollover threshold Mϕ Anti-roll moment
ay,lin Lateral acceleration at the end of linear range m Vehicle mass
ay,max Maximum lateral acceleration Nc No. of steps of control horizon
ay,stab Lat. acc. threshold Np No. of steps of prediction horizon
az Vertical acceleration Pemij Mechanical power of electric motor ij
B Control effectiveness matrix Pem,max Maximum mechanical power of electric motor
b Control effectiveness function p* Master cylinder pressure
bhyp Hyperbolic function constant pB Brake pressure
CG Vehicle centre of gravity pB,ℓ Brake pressure at previous time step
Cαi Cornering stiffness of axle i Q, R, S, W Weight matrices
dego Distance from ego vehicle Rel Resistance
FActij Suspension actuator force RI1 , RI2 Rollover indices
Fped Driver pedal input s Laplace operator
Fx Longitudinal tyre force sV Slack variable
Fxij,max Max. long. force of tyre ij Ts Sample time
FxF,AWD Long. front tyre force due to AWD ti Track width of axle i
FxFj,VSC Long. front tyre force due to VSC tfay Time delay of TF ay /δ at f Hz
Fy Lateral tyre force tfay ψ̇ Time delay of TF ay /ψ̇ at f Hz
FyF,0 Front lat. tyre force in pure lat. slip tfβ Time delay of TF β/δ at f Hz
Fz Vertical tyre load tfϕay Time delay of TF ϕ/ay at f Hz
f Frequency tf ψ̇ Time delay of TF ψ̇ /δ at f Hz
fd () Equality constraint function U Control input vector
fGay Frequency at 0.9Gay ,max U, U Upper and lower limits for U
fGβ Sideslip eigenfrequency U(⋅) Control sequence
fGψ̇ Yaw eigenfrequency UP Preferred value of U
G() Inequality constraint function Uℓ Previous time step control vector
G0 Steady-state gain uact Actuator effort coefficient
Gay Transfer function gain of ay /δ V Vehicle speed
Gay ψ̇ Transfer function gain of ay /ψ̇ Vslipij Wheel slip speed at corner ij
Gyσx () Fy compensation function w.r.t. σ x Vxij Linear wheel speed at corner ij
GFz ,ARC () Fy compensation function w.r.t. Fz vch Characteristic speed
Gβ Transfer function gain of β/δ vfin Exit speed in obstacle avoidance
Gψ̇ Transfer function gain of ψ̇ /δ vin,max Max. entry speed in obstacle avoidance
Gfay Gay at f Hz vx ,vy ,vz Vehicle speed components
Gfβ Gβ at f Hz wβ Weight coefficient for ψ̇ ref,SS
Gf ψ̇ Gψ̇ at f Hz X State vector
g Gravitational acceleration X, X Upper and lower limits for X
gβ85% Sideslip gradient at 0.85ay,max Xd Desired state vector
gβlin Sideslip gradient at 0.4g Xin Initial value of the state vector
gβSO Sideslip snap oversteer XCG, YCG Global CG coordinates
gδlin Steering gradient at 0.4g xCOP,i COP coordinate of axle i w.r.t. CG
gδ85% Steering gradient at 0.85ay,max Xref , Yref Global reference coordinates
gϕlin Roll gradient at 0.4 g of lat. acc. ZV Output vector
Hc Control horizon ZV,d Desired output vector
Hp Prediction horizon zs Vertical displacement of sprung mass
Istatus Understeer/oversteer index zu Vertical displacement of unsprung mass
Izz Yaw moment of inertia αij Slip angle of tyre ij
Kδ Understeer coefficient αij,peak αij at peak lateral force
Ki Tuning coefficient αij,peak,0 αij at peak lateral force, in pure slip
k Discretization step αij,ref Reference tyre slip angle
J Cost function αFFW FFW slip angle of axle i
i
L Wheelbase αSTABLE Stable tyre slip angle value
lF , lR Front/rear axle distance to CG αUNSTABLE Unstable tyre slip angle value
ld Look-ahead distance β Vehicle sideslip angle
ℓ() Stage cost β85% Sideslip angle at 0.85ay,max
ℓNp () Terminal cost βOS Sideslip angle overshoot
MB Braking torque

173
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

βlin β at the end of linear range ADS Active differential system

βmax Maximum sideslip angle ANN Artificial neural network
βref Reference sideslip angle ARC Active roll control system
βref,SS Steady-state ref. sideslip angle ASS Active suspension system
ΔFz,ARC Load transfer due to ARC AWD All-wheel-drive system
ΔMM,max Maximum torque variation Aut Automotive
ΔMMij Torque variation at corner ij BStep Backstepping controller
ΔU Change in control action vector CA Control allocation
ΔyCG Lateral displacement error CAN Controller area network
ΔyCOP,i Lat. displ. err. at COP of axle i CoC Cooperative coexistence
Δyld Look-ahead error CDC Continuous damping control
COP Centre of percussion
ΔzCG Path tracking error vector
CeC Centralised coordination
Δδ Steering angle variation
DAS Dual axis steering
Δδmax Max. steering angle variation
DT Double track
Δψ CG Heading angle error
DYC Direct yaw moment control
δ Front wheel steering angle
EBD Electronic brake distribution
δmax Maximum steering angle
ECU Electronic control unit
δmin Minimum steering angle
EPS Electric power steering
ε Brake pressure model coefficient
ESC Electronic stability control
η Safety coefficient
ESP Electronic stability program
ηij Motor efficiency at corner ij
FB Feedback
θ Body pitch angle
FFW Feedforward
κ Trajectory curvature
FL Fuzzy logic
μ Tyre-road friction coefficient
FLC Fuzzy logic coordination
σxij Wheel slip ratio at corner ij
ICC Integrated chassis control
σx,max Stability threshold for σx
Kin Kinematic
σx,peak σx value at the peak of Fx KPI Key performance indicator
σx,ref Reference tyre slip ratio LIN Local interconnect network
τc Virtual control vector LMA Largest modulus activation
τI 1st order TF time constant LPV Linear parameter varying
τII 2nd order TF time constant LQ Linear quadratic controller
ϕ Body roll angle Lg Longitudinal
ϕlim Roll angle threshold MuC Multi-layer coordination
ψ Body yaw angle MPC Model predictive control
ψ̇ OS Yaw rate overshoot NL Nonlinear
ψ ref Reference heading angle PeC Peaceful coexistence
ψ̇ ref Reference yaw rate PID Proportional integral derivative
ψ̇ ref,SS Steady-state reference yaw rate PS Pure subsumption
ψ̇ ref,δ,SS Steady-state ref. yaw rate based on δ RB Rule-based
ψ̇ stab,β β-based stability yaw rate RWS Rear-wheel-steering
ψ̇ stab,μ μ-based stability yaw rate SMC Sliding mode controller
ST Single track
ωij Angular wheel speed at corner ij
SuC Supervisory coordination
ωn Natural frequency
TCS Traction control system
ζ Damping ratio
TF Transfer function
Acronyms TPC Tyre pressure control
AAS Active aerodynamic system TV Torque-vectoring
ABS Anti-lock braking system VDC Vehicle dynamics control
ACC Adaptive cruise controller VSC Vehicle stability control
ACS Active camber system 4WS 4-wheel-steering
AFS Active front steering

• Adaptability, i.e., the capability of dealing with different operating • Low computational power requirements.
and environmental scenarios.
• Fault-tolerance, i.e., the capability of providing well-defined and safe Despite the many publications on ICC, to the best of the authors’
behaviour in case of malfunctions. knowledge, the literature misses a thorough review with: a) the sys­
• Dynamic re-configurability, i.e., the property of ensuring soft tematic categorisation of the available ICC architectures, with the
switching during vehicle operation. analysis of their strengths and weaknesses; b) the most recent ap­
• Modularity, i.e., the capability of easily inserting or removing sub­ proaches, which are becoming feasible with modern automotive
systems without redesigning the whole architecture. This feature can microcontrollers; c) the reference driving performance requirements;
also ensure “openness,” namely the compatibility of systems from and d) the procedures to objectively evaluate ICC performance. In
different suppliers, while respecting the appropriate intellectual particular, c) and d) are highlighted as important gaps by Ivanov and
property rights. Savitski (2015), and Kissai et al. (2017).

174
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

This review covers a)-d) by considering references (papers and pat­ The car maker mainly needs to define which chassis actuators are suit­
ents) in English language only, and is organised as follows: Section 2 able for the specific vehicle, and verify their compatibility beforehand.
classifies and describes the ICC architectures; Section 3 discusses the
typical reference variables; Section 4 presents the main coordination 2.1.1. Peaceful coexistence
strategies, and highlights recent examples; Section 5 summarises the key In the PeC architecture (Figure 1) all chassis actuation systems are
performance indicators (KPIs) for objective ICC assessment; Section 6 developed independently, and each of them includes both software
outlines future trends, which is followed by the conclusions in Section 7. (control layer) and hardware (physical layer). The aim is to enhance
With respect to automated driving applications, given the several recent performance through simultaneous operation, without interferences.
surveys on path tracking control (Paden et al., 2016; Sorniotti et al., For this reason, Duffie et al. (1988), Gordon et al. (2003), and Yu et al.
2017), this study will focus on references in which the automated (2008) refer to PeC as a decentralised control or heterarchical archi­
driving system replacing the human driver, and deciding the front tecture. Each chassis system conveys its control input independently,
steering input and total wheel torque demand, is coupled with a further and has its own electronic control unit (ECU), which is ideal for the
chassis actuation system, e.g., direct yaw moment control based on the traditional organisation of car makers. The only interaction among the
actuation of the friction brakes and/or electric powertrains, or 4WS. systems is in the form of information sharing, e.g., the systems can
communicate through networks (CAN, LIN, FlexRay, etc.), and appro­
2. Integrated chassis control architectures priate interfaces must be provided by the relevant suppliers (Yu et al.,
2008). This set-up allows a mild coordination of the control actions,
In Kissai et al. (2017), ICC architectures are classified into two cat­ while the conflicts between the reference vehicle responses set by each
egories, which can be further divided into sub-categories according to controller can be prevented through calibration of the individual algo­
Gordon et al. (2003), Heißing and Ersoy (2011), and Reuss et al. (2014): rithms. In general, PeC reduces the number of sensors and improves
robustness with respect to the lack of any coordination, and can result in
• Downstream architectures: peaceful coexistence (PeC) and cooper­ lower computational effort than more complex architectures (Duffie
ative coexistence (CoC), covered in Section 2.1. et al., 1988). Thanks to its reduced development time, PeC is normally
• Upstream architectures: centralised coordination (CeC), supervisory cost-effective when implemented for single-objective ICC, but can imply
coordination (SuC), and multi-layer coordination (MuC), covered in higher costs and less favourable packaging with respect to centralised
Section 2.2. solutions in case of multi-objective ICC (Vivas-Lopez et al., 2015).
Fruechte et al. (1989), and Gordon et al. (2003) consider PeC as the
least suitable architecture for active safety control, since there is not any
actual supervision nor coordination. For the same reasons, Heißing and
2.1. Downstream architectures Ersoy (2011) state that multi-objective ICC is not appropriate for PeC
architectures. Hence, the number of the actuators managed through PeC
Downstream architectures have a parallel structure in which the is usually small. PeC was adopted in the first generation of ICC systems
actuation systems work separately, and some form of integration is (Shibahata, 2005). For example, Mitsubishi and Toyota used PeC for
achieved at the lowest level of the control hierarchy, i.e., close to the 4WS coordination with anti-lock braking systems (ABS) and traction
actuators. For this reason, He et al. (2006) also refer to downstream control systems (TCS), e.g., see Mitamura et al. (1988). The 1991 Toyota
coordination as a bottom-up architecture, which tends to enhance the Soarer included an example of PeC-based coordination of 4WS and
capability of containment and recovery in case of malfunctions, as the active suspensions to enhance cornering response and prevent high
dysfunction of a subsystem does not compromise the performance of the transient roll (Sato and Inoue, 1993); the suspension system had an
remaining actuators. The decentralised structure favours system emergency logic to give priority to the braking actuation when needed.
modularity, modifiability and extendibility, because the introduction of Some researchers have used PeC to coordinate suspension control with a
a new chassis actuation system does not necessarily require major highly integrated set of other chassis actuators, e.g., Cho et al. (2008),
redesign, nor additional high-level controllers. The algorithms of the and Fergani et al. (2017) adopt PeC to combine suspension system
individual systems can be purchased in bulk from traditional suppliers.

Fig. 1. Example of peaceful coexistence architecture, adapted from Reuss et al. (2014).

175
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

control with active steering and braking actuation coordinated through driving conditions (Kissai et al., 2017). According to Heißing and Ersoy
another ICC architecture. Due to their relatively limited contribution to (2011), and Reuss et al. (2014), CoC removes the typical loops of PeC in
vehicle handling, the active aerodynamic systems (AAS), recently the solution of negative interactions, and provides prompter in­
adopted by automotive manufacturers such as Toyota (Mares, 2014), terventions. However, Heißing and Ersoy (2011), and Kissai et al. (2017)
Ferrari (e.g., for the SF90), and Pagani (e.g., for the Huayra), can be highlight that:
integrated with the other chassis actuation systems through PeC. For
example, Ahangarnejad (2018), and Ahangarnejad et al. (2019) use PeC • CoC shows difficulties in adapting to random situations, because the
for integrating an AAS with an upstream coordination architecture for testing scenarios of the a-posteriori coordination cannot consider all
yaw rate control. possible driving conditions.
• Each subsystem has a control law likely to be based on different
2.1.2. Cooperative coexistence models, e.g., each algorithm internally calculates a different refer­
According to Heißing and Ersoy (2011), and Reuss et al. (2014), the ence vehicle behaviour. The different parameters and models for
CoC architecture maintains independent controllers for each actuator; reference calculations as well as the different locations of the various
however, in comparison with PeC, CoC adds a coordination layer subsystems within the system bundle can undermine the coordina­
(Figure 2) that receives the outputs from the individual controllers, and tion strategy reliability.
provides corrected values of the same variables back to the multiple
actuators. The requirement of a coordination layer followed the intro­ Hence, CoC is more suitable for single-objective ICC, and, similarly to
duction of vehicle stability control (VSC, several other acronyms, such as PeC, is not recommended for the complexity of automated driving
ESC, ESP and VDC, are available in the literature for the same system) applications.
and AFS in the mid-1990s, as well as the design of new actuators, e.g.,
the steer-by-wire system of Daimler-Benz (Shibahata, 2005).
Although CoC solves the absence of coordination rules of PeC, it 2.2. Upstream architectures
cannot prevent conflicts at the origin, but only provides a-posteriori
remedies. Therefore, CoC was soon overcome by upstream architectures He et al. (2006) defines upstream architectures as top-down, since a
for vehicle dynamics research (while production vehicles tend to be high-level multivariable controller is placed between the sensor/state
conservative in terms of ICC architectures). The benefit of CoC is that car estimation layer and the chassis actuation systems. The high-level
makers can implement it without adding any high-level coordination controller coordinates the subsystems and prevents conflicts, through
and master controller, and without substantial modifications of the in­ a control allocation (CA) algorithm (Johansen and Fossen, 2013), often
dividual systems, other than compatible interfaces to be developed based on optimisation.
together with the coordination module. Hence, from an industrial Upstream architectures can easily account for many aspects, e.g.,
perspective, CoC is financially attractive because it can exploit the driving conditions, driver intentions, actuator response times and power
economy of scale of conventional systems (Kissai et al., 2017). Wang consumption, as all relevant parameters are stored in the multivariable
et al. (2009) highlights that the car maker should be responsible for the controller. The ability of managing multi-objective ICC makes upstream
coordination unit that sets priorities, detects potential conflicts, and coordination a good solution for future vehicle dynamics controllers,
limits interferences. According to Heißing and Ersoy (2011), the various where the chassis systems must be able to handle the whole spectrum of
systems can also request and provide assistance to one another, e.g., the driving conditions (Kissai et al., 2017), and concurrently increase
VSC can request assistance from the controlled dampers in the form of vehicle capabilities (Schmidt and König, 2020). These architectures are
optimised tyre loads to help stabilise the vehicle in critical scenarios. also suitable for the next generation of connected vehicles (Montanaro
The coordination strategy (see Section 4) is often implemented et al., 2018), benefitting from car-to-x information exchange, e.g.,
through rule-based algorithms derived from desired settings for selected regarding obstacles, accidents, weather and road conditions (Schmidt
and König, 2020). Although Kissai et al. (2017) deems top-down

Fig. 2. Example of cooperative coexistence architecture, adapted from Reuss et al. (2014).

176
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

coordination exaggerated for a single purpose, upstream architectures computational requirements, and system cost. Alternatively, the
have also been successfully implemented for single-objective ICC, for fail-safe redundancy of microcontrollers can be a solution.
example in Boada et al. (2006), and Falcone, Tseng, et al. (2007). The
main downside is the relatively high computational effort, which made For the previous reasons, Ahangarnejad (2018) states that CeC might
these architectures unattractive for the first two ICC generations. not be the best choice for real-vehicle implementation of multiple
chassis actuators, but considers this architecture a reasonable choice for
2.2.1. Centralised coordination related or coupled systems, e.g., for steering and braking coordination,
The top layer of the CeC architecture (Figure 3) collects the data from as presented in studies by General Motors (Salman et al., 1992), Toyota
the sensors and state estimators, and conveys them to the second layer, (Hirano et al., 1993), and Audi (Mihailescu et al., 2019). The recent
including the multivariable master controller, which provides the con­ study in Tang and Khajepour (2020) confirms the flaws of CeC and – as
trol actions to the actuators, located in the bottom layer, and the ref­ an effective alternative – proposes a modular architecture based on
erences to the so-called slave controllers (Wang et al., 2009), fulfilling distributed optimal control.
specific tasks not involved in the integration, e.g., feedback wheel slip
control. The integrated controller has complete authority on all sub­ 2.2.2. Supervisory coordination
systems, and thus CeC can provide enhanced performance by simulta­ The SuC architecture adds a local control layer to CeC, or an up­
neously manipulating all modules. Moreover, this solution allows stream coordination layer to PeC. Figure 4 reports the three main layers
systematic and formal consideration of system stability since the initial that identify the SuC architecture (Kissai et al., 2017), i.e., the super­
design stages. vision layer, the control layer, and the physical layer.
However, the CeC architecture has drawbacks: Starting from the vehicle signals and human or automated driver
commands, the supervision strategy classifies the current driving situ­
• Lack of modularity, which requires the car maker to develop the ation, detects the occurrence of specific instabilities or system failures,
master controller together with its suppliers, to allow them to and determines the working area of each chassis actuation system by:
appropriately modify the actuation algorithms of their subsystems.
• The substantial complexity associated with a single multivariable • Monitoring the control actions (Doumiati et al., 2011, 2013; Fan and
controller. Li et al. (2007, 2008), and Shen and Yu (2006b, 2007) Zhao, 2019; Gáspár et al., 2009; Hwang et al., 2007; Kim et al.,
reduce this difficulty by dividing the system into a main control loop, 2013), and allowing the intervention of the actuators according to
which produces the main control inputs, and a servo control loop, pre-defined priorities, e.g., through direct yaw moment control
which distributes the related control actions. The system keeps the activation when steering control cannot provide any further yaw
multivariable control structure, although the tasks are appropriately moment (Mousavinejad et al., 2017).
decoupled. This architecture, called main/servo-loop partition, is the • Monitoring the reference states (Cho et al., 2011, 2012; Her, Suh,
first step towards supervisory and multi-layer architectures. et al., 2015; Yoon et al., 2008), where the system detects the driving
• Lack of flexibility with respect to the inclusion of additional actua­ conditions basing on indices, e.g., see the split-μ identification al­
tors or functionalities. gorithm by Bedner et al. (2004), and then modifies the reference
• Difficulty in coping with central unit failures. Unlike the downstream model, e.g., to produce different target yaw rates (Cho et al., 2012).
architectures, CeC in itself does not provide fault tolerance; however, • Combining the two previous strategies (Burgio and Zegelaar, 2006;
it is possible to identify faults and design additional control logics Gáspár and Németh, 2016).
that help system recovery from malfunctions (Duffie et al., 1988).
Andreasson and Bunte (2006), and Schiebahn et al. (2010) recom­ The control layer can be further divided into high-level controller
mend the presence of a supervisory layer providing automatic (master controller) and low-level controllers (Gordon et al., 2003). The
on-board reconfiguration in case of failures. However, these solu­ high-level controller typically consists of a multivariable controller that
tions further increase complexity, development effort, converts the references generated by the supervision layer into control

Fig. 3. Example of centralised coordination architecture, adapted from Gordon et al. (2003).

177
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Fig. 4. Example of supervisory coordination architecture, adapted from Gordon et al. (2003).

actions for each local controller. The low-level controllers are the local an overall system dysfunction.
controllers of the chassis actuation systems, which can be separately A feature of SuC is its modular nature, which allows the vehicle
designed and validated by the respective suppliers. manufacturer (typically responsible for the supervision strategy and
The layering of the architecture implies some degree of fault- master controller) and its suppliers to independently develop comple­
tolerance. In particular, if the control layer is not completely depen­ mentary controllers, provided that consistent interfaces are in place.
dent on the supervision layer, i.e., when the essential vehicle sensor and This feature makes this hierarchical framework rather common in the
state estimation outputs are also routed to the control layer, the struc­ automotive industry (see Figure 19 in Section 6). Examples of SuC
ture ensures minimum functionality to keep the vehicle safe. Otherwise, integrate braking system control with AFS, see the implementations by
if the vehicle state information is only used by the top layer, the archi­ Bosch (Lohner et al., 2007) and Delphi (Hac et al., 2002); adaptive cruise
tecture remains centralised, and the loss of communication would cause control (ACC) with steering control, see the implementation by General

Fig. 5. Example of multi-layer architecture, adapted from Gordon et al. (2003).

178
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Motors in (Lee et al., 2012); and direct yaw moment control with simplicity of the algorithm.
steering actuation (Cheng et al., 2019). Savitski et al. (2015) uses SuC
for an off-road vehicle concept including brake, suspension, and tyre 2.3. Summary and implementation examples
pressure control. A peculiarity of SuC is the possibility of extending the
architecture horizontally by adding further actuators, e.g., see the su­ As a summary of the discussion in Sections 2.1-2.2, Table 1 quali­
pervisory strategies patented by Delphi (Chandy, 2003) and General tatively evaluates the architectures with respect to three group of at­
Motors (Majersik et al., 2010). tributes, related to the: i) ease of industrial implementation; ii) resulting
ICC performance; and iii) suitability to different categories of passenger
2.2.3. Multi-layer coordination cars. The ratings are based on the comments from the literature and the
Similarly to SuC, MuC presents the essential features of system experience of the authors. Moreover, Table 2 shows which active sys­
modularity and design flexibility, by separating the functional re­ tems have been integrated within the different ICC architectures, i.e.,
quirements while coordinating and monitoring each module. Figure 5 each row refers to a category of chassis actuation system, and reports the
displays the typical MuC layers: studies integrating the specific actuation method according to the ICC
topologies indicated in the columns. The column ‘Hybrid’ refers to ICC
• The supervision strategy, which decides the appropriate control systems based on the simultaneous adoption of multiple ICC architec­
mode and computes the references according to the dynamic states tures, or with features at the boundary among different architectures.
and inputs, e.g., by defining a state machine; In practice, the boundaries are often rather loose, and many ICC
• The high-level controller, which calculates the global control inputs, implementations have features that could be categorised under more
e.g., in terms of total yaw moments, to track the selected reference than one architecture. For example, the relevant distributed architecture
signals; in Tang and Khajepour (2020) formulates a model predictive control
• The coordination strategy, which selects the chassis control systems problem for each vehicle corner, characterised by independent wheel
according to the mode defined by the supervision layer; torque and steering actuation. In the internal model of the formulation
• The CA strategy, which distributes the control inputs among the of each control module, the disturbance term observed by the module
chassis actuation systems; includes the external contributions from the neighbouring modules,
• The individual actuator controllers, which track the CA references, which are shared as generalised forces, and integrated into agent
convey the local control actions to the hardware, and can include models. This weak coupling enables each agent to interpret their
slave controllers not involved in the ICC integration; neighbours’ impact on vehicle dynamics without knowing the detailed
• The physical layer, which executes the various operations by means configurations of the other corners. A distributed controller is devel­
of smart actuators. oped, where in each module an iterative algorithm: a) calculates the
optimal solution for the corner module based on the generalised forces
The car maker would be typically responsible for the vehicle dy­ received from the other modules; b) updates the generalised force
namics supervision down to the CA layer, whereas the relevant suppliers contribution of the module based on the solution in a); and c) broadcasts
would independently design the individual actuation systems. its updated generalised forces and receives the updated generalised
This hierarchical structure was formally introduced by Fruechte forces from the other modules. The process in a)-c) is iterated for all
et al. (1989) for the coordination of braking and steering systems. The vehicle corners until: i) either a consensus is reached, i.e., the norm of
review of Gordon et al. (2003) presents MuC as a layout that decomposes the variation of the generalised forces of the last two iterations is less
a complex chassis control problem into smaller and more manageable than a specified threshold for all modules; or ii) the specified maximum
subparts, wherein the actuators are grouped according to the respective number of iterations is reached. This set-up can be considered rather
domain of interest, see also Ivanov et al. (2010). Zhao et al. (2019), and close to that of a PeC architecture, as the only coordination is in the form
Zhao et al. (2017) describe this architecture as a mean to easily outline of communication among modules, but in a much more advanced form
and design the optimal operating regions of the actuators and prevent than in conventional PeC implementations. At the same time, the co­
coupling issues. ordination of the control efforts within each module is managed in a
MuC allows extending the layering upwards to include further centralised way within the module. The overall resulting performance at
vehicle behaviours and system interactions. Fruechte et al. (1989) ex­ the vehicle level tends to that of the corresponding CeC, provided that a
plains that each time the problem is decomposed into modules, e.g., sufficiently high number of iterations is allowed, with the additional
performance and safety, controlled dynamics, and fault detection, an flexibility and computational efficiency deriving from the distributed
additional coordination among them is required. Each module is controller configuration.
implementable with the current automotive microprocessor technology.
However, the implementation cost and complexity of this architecture 3. Reference behaviours
are not justified at the industrial level, and such typology is not widely
adopted for actual ICC yet. Chang and Gordon (2008), and Song et al. In their literature review, Ivanov and Savitski (2015) indicates that
(2015) associate MuC to full drive-by-wire systems, which are not the previous chassis control studies focus on the individual or concurrent
broadly applied in the automotive industry yet. However, MuC has control of the following domains: i) braking and traction, i.e., longitu­
already attracted the attention of companies as Bosch (Trachtler, 2004), dinal vehicle dynamics; ii) handling and lateral stability, i.e., lateral
Hyundai-Kia (Her, Suh, et al., 2015), and General Motors (Deng, 2012). vehicle dynamics, which is the core of most of the available ICC papers;
In general, among the upstream architectures, MuC seems the most and iii) attitude control, i.e., the control of the vehicle body motion
suitable one for automated driving applications (Kissai et al., 2017), e.g., caused by the longitudinal and lateral accelerations, as well as road ir­
see the example in Chang and Gordon (2008). Chatzikomis et al. (2018) regularities. Within each domain, the controllers track appropriate
compares MuC – a similar hierarchical strategy is reproduced in Ren reference variables, which can indirectly influence the dynamics of the
et al. (2018) – and CeC for the integration of steering angle control and other domains. The main reference variables selected in the literature
TV for path tracking in an automated race vehicle. The modularity of are discussed in the following sections.
MuC allows considering the variations of tyre-road friction conditions
within the TV controller placed in a lower layer than the path tracking 3.1. Longitudinal dynamics
controller. Conversely, the specific CeC implementation needs either a
different tuning as a function of the identified friction conditions, or In modern vehicles, the total reference longitudinal traction/braking
must include further stability considerations, which undermines the force or wheel torque are either determined by the driver, through the

179
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Table 1
ICC architecture categorisation and evaluation at a glance.

accelerator/brake pedal input, or an automated driving system (Lin the torque applied to the individual wheels, including generation of
et al., 2019; Ni, 2017; Ren et al., 2018), which tracks – typically through direct yaw moments and total torque variations with respect to the
a feedback controller – the reference vehicle speed defined by the path driver demand, e.g., see the trail braking effect in Velenis et al. (2008),
planning layer. and Zarkadis et al. (2018). In some of the ICC implementations from the
Nowadays the traction torque demand is always conveyed by-wire to literature, the wheel slip control modules are used as “slave” controllers,
the propulsion unit, which enables automated driving, and is calculated tracking the reference slip ratios defined by the control blocks located in
through drivability maps, typically functions of accelerator pedal posi­ the higher layers of the ICC hierarchy (Wang et al., 2009). In industrial
tion, vehicle speed, and further inputs, e.g., the selected driving mode implementations, the reference slip ratios and wheel deceleration
(He et al., 2013; Nishio and Shen, 2019; Shen et al., 2017). In electric thresholds for ABS and TCS are usually determined through heuristics, e.
vehicles, the drivability map also includes the computation of a regen­ g., based on the wheel dynamics measured during the previous ABS
erative braking contribution for the initial part of the accelerator pedal cycle (Chen et al., 2012; Reif, 2014; Robert Bosch Gmbh, 2007). In ac­
travel (Reif, 2014; Robert Bosch Gmbh, 2007). In cars equipped with ademic papers, different extremum seeking methods have been pro­
conventional braking systems with brake booster, tandem master cyl­ posed for the robust generation of the reference slip ratio (Drakunov
inder and VSC unit, during base brake operation the friction braking et al., 1995; Morrison and Cebon, 2017).
torque is determined by the brake pedal force and the resulting tandem The powertrain unit is normally equipped with an anti-jerk
master cylinder pressure, i.e., without any form of control. In controller, which targets the reduction of the torsional drivetrain dy­
brake-by-wire systems, either including electro-hydraulic or namics and related longitudinal acceleration oscillations, to enhance
electro-mechanical actuation (Jonner et al., 1996; Cheon, 2010; Yu and passenger comfort during swift torque demand variations. Scamarcio
Liu, 2016), the reference friction braking torque originates from maps, et al. (2020) categorises anti-jerk controllers based on the variables used
mainly based on the position of the brake pedal, which is connected to a for the computation of the anti-jerk control action, and discusses the
brake pedal force emulator, giving the driver desirable force feedback. respective reference values. Traditional anti-jerk controllers are deac­
In some brake-by-wire implementations (Cho et al., 2012; Joa et al., tivated during TCS interventions; however, recent implementations
2015), the reference at the vehicle level is directly expressed in terms of include integration of the anti-jerk and TCS modules (Batra et al., 2018;
longitudinal acceleration, ax . This is generated through appropriate De Pinto et al., 2017; Scamarcio et al., 2020).
feedback controllers based on the longitudinal acceleration measured by The actuation of the powertrain/s and friction brakes, based on
the on-board inertial measurement unit (IMU), which is typically longitudinal slips and longitudinal tyre forces, can be supported by
affected by signal noise, and needs compensation strategies for the ef­ further chassis actuation systems, e.g.: i) active or semi-active differ­
fects of longitudinal road gradient and bank angle (Grip et al., 2009). All entials (Scalzi and Marino, 2008), to enhance the traction capability and
brake-by-wire systems are obviously ideal for the implementation of cornering response (see the relevant reference variables in Section 3.2);
driving automation, and for the flexible blending of the friction braking and ii) semi-active or active suspension systems, which can compensate
torque and regenerative braking torque contributions in electrified ve­ the pitch dynamics in traction and braking (Alleyne, 1997; Ting and Lin,
hicles (Lv et al., 2018). 2004), and reduce the dynamic tyre load variation as well as stopping
ABS and VSC are mandatory on passenger vehicles, and are normally distance during ABS activation (Shao et al., 2007; Shen and Yu, 2006a),
implemented – together with the traction control system (De Pinto et al., see Section 3.3.
2017; Hrovat et al., 1998) and the electronic brake distribution (EBD)
module – as a single system packaged within an integrated mechatronic
3.2. Lateral dynamics
unit (Van Zanten et al., 1996). Their role is to prevent excessive longi­
tudinal slip of the individual wheels, thus facilitating the directional
Recent lateral dynamics control systems, e.g., based on TV control,
control of the vehicle, and to reduce the yaw rate error and sideslip angle
tend to continuously or frequently intervene, e.g., they seamlessly
(see Section 3.2 for the discussion on the lateral dynamics), by altering
enhance the cornering response not only during limit handling

180
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Table 2
Actuation systems involved in the different ICC architectures
Actuator PeC CoC CeC SuC MuC Hybrid

ACS, AFS, Lee (2002); Mastinu Cao, Zheng (2019); Ahangarnejad (2018); Bedner, Chen (2004); Abe, Mokhiamar Ahangarnejad
RWS, et al. (1994); Plochl, Chowdhri et al. (2021); Ahangarnejad et al. Burgio, Zegelaar (2006); (2007); Alberding et al. (2018);
4WS Lugner (1996); Sato, Elhefnawy et al. (2017); (2019); Ando, Fujimoto Cho et al. (2012); Chokor (2014); Chang, Gordon Ahangarnejad et al.
Inoue (1993); Scalzi, He et al. (2006); Kou (2010); Andreasson, et al. (2019); Doumiati (2008); Chatzikomis (2019); Cho et al.
Marino (2008); et al. (2004); March et al. Bunte (2006); Boada et al. (2013); Feng et al. et al. (2018); Chen et al. (2008); Fergani et al.
Taheri, Law (1990) (2007); Rahimi, Naraghi et al. (2006); Brennan, (2020); Gáspár, Németh (2019); Fruechte et al. (2017); Tang,
(2018); Wang et al. Alleyne (2001); (2016); Hou et al. (2008); (1989); Hajiloo et al. Khajepour (2020)
(2009); Xie et al. (2018) Chatzikomis et al. (2018); Hwang et al. (2007); Joa, (2020); Her, Suh, et al.
Chen et al. (2006); Cho Park, et al. (2018); Kim (2015); Ivanov et al.
et al. (2008); Ding, et al. (2013); Liang et al. (2010); Kissai et al.
Taheri (2010); Falcone, (2020); Mirzaei, (2018); Ono et al.
Tseng, et al. (2007); Mirzaeinejad (2017); (2006); Ren et al.
Falcone et al. (2008); Fu Mirzaeinejad et al. (2018); (2018); Shyrokau et al.
et al. (2017); Guo et al. Mousavinejad et al. (2013, 2015);
(2017); Hang, Chen (2017); Németh et al. Shyrokau, Wang,
(2019); Hirano et al. (2017); Selby (2003); (2012); Song et al.
(1993); Hou et al. (2008); Shen, Yu (2006a); Wu (2015); Trachtler
Li et al.(2008); Li, Yu et al. (2020); Xiujian et al. (2004); Xia et al.
(2007); Li, Arat (2016); (2009); Yim (2018); (2020); Zhang, Li
Matsumoto, Tomizuka Zhang et al. (2018); (2019); Zhao et al.
(1992); Nagai et al. Zheng, Shyrokau (2019) (2019); Zhao et al.
(1997, 1998, 2002); Ono (2017)
et al. (1994); Reinold,
Traechtler (2013);
Saikia, Pathak (2019);
Salehpour et al. (2015);
Salman et al. (1992);
Schiebahn et al. (2010);
Shen, Yu (2006b,2007);
Shuai et al. (2013); Wang
et al. (2018); Warth et al.
(2020); Yu, Moskwa
(1994); Zhao et al.
(2015); Zhu et al. (2014)

AAS - - Ahangarnejad (2018); - - -

Ahangarnejad et al.
(2019)

ABS Alleyne (1997); Kou et al. (2004) Hou et al. (2008) - -

Kawakami et al.
(1992); Lee (2002);
Mastinu et al. (1994);
Mitamura et al.
(1988); Plochl,
Lugner (1996); Sato,
Inoue (1993); Shao
et al. (2007); Taheri,
Law (1990); Tchamna
et al. (2014); Ting, Lin
(2004); Wang et al.
(2009)

ACC - - - Cheng et al. (2019) - -

ADS, TV Scalzi, Marino (2008) - Ahangarnejad (2018); Feng et al. (2020); Chatzikomis et al. Tang. Khajepour
Ahangarnejad et al. Gáspár, Németh (2016); (2018); Chen et al. (2020)
(2019); Chatzikomis Her et al. (2016); Her, (2019); Edrén et al.
et al. (2018); Hirano Koh, et al. (2015); Hwang (2019); Hajiloo et al.
et al. (1993); Li et al. et al. (2007); Joa, Park, (2020); Xia et al. (2020)
(2008); Li, Yu (2007); et al. (2018); Joa, Yi, et al.
Schiebahn et al. (2010); (2018); Kim et al. (2013);
Shen, Yu (2006b, 2007); Liang et al. (2020);
Warth et al. (2020); Yu, Németh et al. (2017); Wu
Moskwa (1994); Zhu et al. (2020)
et al. (2014)

ARC, Alleyne (1997); Cao, Zheng (2019); Ahangarnejad (2018); Chokor et al. (2019); Her, Suh, et al. (2015); Ahangarnejad
ASS, Kawakami et al. Elhefnawy et al. (2017); Ahangarnejad et al. Gáspár et al. (2009); Her, Song et al. (2015); (2018);
CDC (1992); Kou et al. Kou et al. (2004); March (2019); Chen et al. Koh, et al. (2015); Her Trachtler (2004); Zhao Ahangarnejad et al.
(2004); Lee (2002); et al. (2007); Rahimi, (2006); Li et al. (2008); et al. (2016); Joa, Yi, et al. et al. (2019); Zhao et al. (2019); Cho et al.
Lou et al. (2010); Naraghi (2018); Sun et al. (2019); Wang (2018); Poussot-Vassal (2017) (2008); Fergani et al.
Mitamura et al. Velardocchia, Vigliani et al. (2018); Xiao et al. et al. (2011); Savitski (2017)
(1988); Ricciardi (2013) (2011); Yim (2012) et al. (2015); Shen, Yu
et al. (2019); Shao (2006a); Xia et al. (2019);
et al. (2007); Soltani Yoon et al., (2008)

(continued on next page)

181
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Table 2 (continued )
Actuator PeC CoC CeC SuC MuC Hybrid

et al. (2018);
Tchamna et al. (2014);
Ting, Lin (2004);
Valášek et al. (2004)

DYC, Ricciardi et al. (2019); Chowdhri et al. (2021) Andreasson, Bunte Bedner, Chen (2004); Abe, Mokhiamar Ahangarnejad
ESC, Soltani et al. (2018); Elhefnawy et al. (2017); (2006); Boada et al. Burgio, Zegelaar (2006); (2007); Alberding et al. (2018);
ESP, Valášek et al. (2004) He et al. (2006); Rahimi, (2006); Brennan, Alleyne Cheng et al. (2019); Cho (2014); Chang, Gordon Ahangarnejad et al.
VDC, Naraghi (2018); (2001); Chen et al. et al. (2012); Chokor et al. (2008); Fruechte et al. (2019); Cho et al.
VSC Velardocchia, Vigliani (2006); Cho et al. (2008); (2019); Doumiati et al. (1989); Hajiloo et al. (2008); Fergani et al.
(2013); Wang et al. Ding, Taheri (2010); (2013); Gáspár et al. (2020); Her, Suh, et al. (2017)
(2009); Xie et al. (2018) Falcone, Tseng, et al. (2009); Her, Koh, et al. (2015); Ivanov et al.
(2007); Falcone et al. (2015); Her et al. (2016); (2010); Kissai et al.
(2008); Fu et al. (2017); Hou et al. (2008); Hwang (2018); Ono et al.
Guo et al. (2017); Hang, et al. (2007); Joa, Park, (2006); Ren et al.
Chen (2019); Li, Arat et al. (2018); Joa, Yi, et al. (2018); Shyrokau,
(2016); Lin et al. (2019); (2018); Mirzaei, Wang (2012, 2013,
Matsumoto, Tomizuka Mirzaeinejad (2017); 2015); Song et al.
(1992); Nagai et al. Mirzaeinejad et al. (2018); (2015); Trachtler
(1997, 1998, 2002); Mousavinejad et al. (2004); Zhang, Li
Reinold, Traechtler (2017); Poussot-Vassal (2019); Zhao et al.
(2013); Saikia, Pathak et al. (2011); Savitski (2019); Zhao et al.
(2019); Salehpour et al. et al. (2015); Selby (2017)
(2015); Salman et al. (2003); Wu et al. (2020);
(1992); Shuai et al. Xia et al. (2019); Xiujian
(2013); Sun et al. (2019); et al. (2009); Yim (2018);
Xiao et al. (2011); Yim Yoon et al. (2008); Zhang
(2012); Zhao et al. (2015) et al. (2018); Zheng,
Shyrokau (2019)

TCS Kawakami et al. He et al. (2006); Rahimi, Ando, Fujimoto (2010); Hou et al. (2008); Zhang Abe, Mokhiamar -
(1992); Sato, Inoue Naraghi (2018); Hou et al. (2008); Ono et al. (2018) (2007); Her, Suh, et al.
(1993); Wang et al. Velardocchia, Vigliani et al. (1994) (2015); Ivanov et al.
(2009) (2013); Wang et al. (2010); Kissai et al.
(2009) (2018); Ono et al.
(2006); Ren et al.
(2018); Shyrokau,
Wang (2012, 2013,
2015); Song et al.
(2015); Trachtler
(2004); Zhao et al.
(2019); Zhao et al.
(2017)
TPC - - - Savitski et al. (2015) Shyrokau et al. (2015) -

scenarios, but during normal driving as well (Doumiati et al., 2013). Yaw rate tracking is the focus of most lateral dynamics controllers.
This is also reflected in the last generation of VSC systems (König et al., The literature – among many others, see Cheng et al. (2020), Tang and
2018, 2019), which operate more frequently and progressively than Khajepour (2020), and Zhu et al. (2019) – often considers formulations
their initial implementations (Tseng et al., 1999; Van Zanten et al., in which the steady-state value of the reference yaw rate, ψ̇ ref,δ,SS , is
1995). based on the steering angle δ according to:

Fig. 6. (a) Example of reference yaw rate map as a function of vehicle speed and steering angle, adapted from Ricco et al. (2020); and (b) Stable and unstable regions
in the β − β̇ phase-plane, adapted from Mousavinejad et al. (2017).

182
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

V Alternatively, Chatrath (2019), Chatrath et al. (2020), and Heißing

ψ̇ ref ,δ,SS = δ (1)
L + Kδ V 2 and Ersoy (2011) obtain the dynamic reference yaw rate through the
second order transfer function of the single track model:
which derives from the well-known linearised single track vehicle model
(Gillespie, 1992; Milliken and Milliken, 1995). For a given speed, (1) ψ̇ ref 1 + τII s
(s) =
brings a linear dependency of the reference yaw rate on steering angle. ψ̇ ref ,SS 2ζ
1 + s + 2 s2
1
This is appropriate for systems intervening only rarely, in case of sig­ ωn ωn
nificant yaw rate errors, but could be non-ideal for continuously active mVlF
systems (Lu et al., 2016), as it is desirable for human drivers to feel the τII =
Cα R L
progressive transition from the linear region to the condition of terminal (6)
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
understeer, and become aware of the approaching cornering limit. CαR lR − CαF lF CαF CαR L2
Therefore, in recent TV strategies (Chatzikomis et al., 2017, 2018; De ωn = +
Izz Izz mV 2
Novellis et al., 2015; Lenzo et al., 2020; Lu et al., 2016; Scalzi and
( )
Marino, 2008), ψ̇ ref,δ,SS is expressed in the form of nonlinear maps, e.g., 1 CαF + CαR CαF l2F + CαR l2R
ζ= +
as functions of steering angle, vehicle speed as well as additional vari­ 2ωn mV Izz V
ables, such as the torque demand and estimated tyre-road friction
parameter, μ, see Figure 6(a). Optimisation approaches based on where τII is the yaw time constant, ωn is the natural frequency of the yaw
nonlinear quasi-static or dynamic vehicle models have been formulated motion, and ζ is the yaw damping ratio, which are functions of the
and implemented to generate the reference yaw rate maps, minimising vehicle parameters, as reported in the formulation. Further formulations
criteria such as tyre slip power losses or total power losses (De Novellis are available in the literature; for example, Joa, Park, et al. (2018)
et al., 2013, 2014; Parra et al., 2020); however, to the best of our calculates ψ̇ ref through the following dynamic equation, deriving from
knowledge, although implemented on vehicles with TV control or active the single track vehicle model:
suspensions, these methodologies have not been applied to vehicles with LCαF CαR
(
L
)
LCαF CαR
ICC systems yet. ψ̈ ref = − Kδ V + ψ̇ + δ (7)
Izz (CαF + CαR ) V ref ,δ,SS Izz (CαF + CαR )
To provide stable behaviour in variable tyre-road friction conditions,
many authors, e.g., Cho et al. (2008, 2012), Doumiati et al. (2013), Some authors e.g., Fan and Zhao (2019), Mousavinejad et al. (2017),
Khosravani et al. (2018), Tang and Khajepour (2020), Xiujian et al. Salehpour et al. (2015), and Zhang and Li (2019), calculate a reference
(2009), Zhang and Li (2019), Zhu et al. (2014), and Zhu et al.(2019), sideslip angle, βref , as the filtered version of the following steady-state
obtain the steady-state reference yaw rate, ψ̇ ref,SS , by saturating |ψ̇ ref,δ,SS | value, deriving from the single track vehicle model:
with the maximum yaw rate compatible with the available μ: V 2 lF m
lR −
ημg βref ,SS = 2CαR L
δ (8)
ψ̇ stab,μ = (2) L+ Kδ V 2
V
βref , either given by (8) or simply set to zero, can be adopted for direct
where η is a safety factor. Moreover, large values of |β| (or alternatively
rear axle slip angle) imply significant reductions of the yaw moment that yaw moment control, in the context of a multi-variable controller using
can be achieved through the variation of δ, see Shibahata et al. (1994) reference values for both yaw rate and sideslip angle, even if direct yaw
and Van Zanten (2000). This means that at high |β| the driver cannot moment control on its own cannot simultaneously track independent
control the vehicle through the steering input, and provides the theo­ yaw rate and sideslip angle references (Kaiser, 2015; Lu et al., 2016);
retical justification to the limitation of |β| carried out by stability control however, this can be achieved by adding a 4WS system (Bedner and
systems, see Andreasson and Bunte (2006). The sideslip angle limitation Chen, 2004; Matsumoto and Tomizuka, 1992).
can be directly achieved through a specific constraint, if this is allowed The maximum safe value of sideslip angle, βmax , depends on μ, i.e.,
by the selected control structure, or alternatively through control for­ the larger is μ, the larger is βmax . For this reason, Ahangarnejad (2018),
mulations based only on yaw rate tracking, e.g., by modifying ψ̇ ref,δ,SS Chatzikomis et al. (2018), Funke et al. (2015), and Zhang and Li (2019)
define variable β thresholds based on the estimated μ. An empirical
according to a weighting function wβ , which increases with |β|, see
formula relating βmax to μ is used in many ICC references, e.g., Cheng
Chatzikomis et al. (2018), and Lenzo et al. (2020):
et al.(2019), Fan and Zhao (2019), Mousavinejad et al. (2017), Wang
( )
et al. (2009), Xiujian et al. (2009), and Zhang and Li (2019):
ψ̇ ref ,SS = ψ̇ ref ,δ,SS − wβ ψ̇ ref ,δ,SS − ψ̇ stab,β (3)
βmax = tan− 1 (0.02μg) (9)
where ψ̇ stab,β is a yaw rate compatible with the available μ, and can be
Instead of β, Burgio and Zegelaar (2006) limit the lateral slip speed,
expressed as a function of the current lateral acceleration of the vehicle,
which is equivalent. Doumiati et al. (2011), Mirzaei and Mirzaeinejad
where the term Δay provides conservativeness (Lenzo et al., 2020):
(2017), Mirzaeinejad et al. (2018), Selby (2003), and Xie et al. (2018)
( ) implement controllers that try to keep the vehicle in the stable region of
ay − sign ay Δay
ψ̇ stab,β = (4) the β − β̇ phase-plane, see Figure 6(b), which can be defined as:
V
The formulations in (2) and (3)-(4) can be applied either individually |K1 β̇ + K2 β| < 1 (10)
or concurrently.
The reference yaw rate, ψ̇ ref , to be tracked by the feedback controller The β − β̇ phase-plane allows conditions in which |β| is small and |β̇|
is large (the signs are also important to determine divergence and
is usually based on the low-pass filtering of ψ̇ ref,SS , to provide realistic
convergence), or conditions where large |β| values are considered stable
and desirable dynamic response, compatible with the natural behaviour
for small |β̇| (Smakman, 2000), and is popular among the authors
of the vehicle, see Mirzaeinejad et al. (2018), Shen and Yu (2007), Wang
because of: i) its direct connection to sideslip motion, and thus vehicle
et al. (2009), and Zhang and Li (2019):
stability (Doumiati et al., 2013; Selby, 2003); ii) its relative indepen­
ψ̇ ref 1 dence from the road conditions when the stability margins are appro­
(s) = (5)
ψ̇ ref ,SS 1 + τI s priately selected, as discussed in Smakman, (2000), see Figure 6(b),
unlike other phase-planes, e.g., αR − ψ̇ and β − ψ̇ (Abe, 1992; Selby,

183
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

2003); and iii) its ease of interpretation, e.g., if a state and its time de­ ( )
rivative have the same sign, they are increasing in magnitude and Δyld = ΔyCG + ld (Δψ CG + βSS ) = ΔyCG + ld Δψ CG + αFFW
f + lR κ (14)
diverging from the stable area, vice versa if their signs differ (Selby,
2003). where βSS is the steady-state sideslip angle value, which can be
Beal and Gerdes (2012), and Funke et al. (2016) respectively choose expressed through αFFW
f , i.e., the front slip angle value predicted through
β − ψ̇ and vy − ψ̇ phase-planes to overcome the vehicle speed and a feedforward algorithm, and κ, i.e., the trajectory curvature. (14) tends
parameter dependencies that are distinctive of the β − β̇ phase-plane to bring the lateral displacement error at the centre of gravity to zero,
(Selby, 2003). Abe (1992) deems the β − ψ̇ phase-plane the best and make the vehicle operate with Δψ CG = − β, see Fig. 7(b).
method to assess stability during aggressive combined manoeuvres The ICC implementation including path tracking through front and
because it is largely invariant with respect to vx . Joa, Park, et al. (2018), rear steering in Hiraoka et al. (2009) is based on the control of the lateral
and Salehpour et al. (2015) prefer controlling αR , and calculate its displacement errors at the front and rear centres of percussion, respec­
maximum allowed value as: tively ΔyCOP,F and ΔyCOP,R , which are located on the symmetry plane of
3μ|FzR | the vehicle, at longitudinal positions with respect to the centre of gravity
αR,peak = (11) defined by:
CαR
⃒ ⃒ ⃒ ⃒
Some implementations tend to relax the limits of the desirable ⃒xCOP,F ⃒ = Izz ⃒xCOP,R ⃒ = Izz (15)
sideslip angle region; for example, Beal and Gerdes (2012) designs a mlR mlF
controller that can stabilise the vehicle by reaching high β regions, Very interestingly, the resulting state-space formulation of the single
bringing the tyres to operate beyond the slip angle values corresponding track vehicle model indicates that the path tracking problems at the
to the maximum lateral forces, and thus permitting experienced drivers front and rear centres of percussion are decoupled. This means that the
to make the car slide to regain stability. Accordingly, the autonomous lateral displacement dynamics of the front centre of percussion is in­
driving studies of Funke et al. (2015, 2016) relax the rear slip angle dependent from the lateral force of the rear tyres, while the lateral
stability constraints to prioritise collision avoidance and path tracking displacement dynamics of the rear centre of percussion is independent
over stability and handling. from the lateral force of the front tyres, which is confirmed also in the
Rahimi and Naraghi (2018), and Wang et al. (2009), on top of the front steering implementation for path tracking in Kritayakirana and
reference yaw rate, consider a reference lateral acceleration, while they Gerdes (2012). Hence, each centre of percussion path deviation can be
do not use reference values for β or vy ; Li and Arat (2016) directly track independently controlled by the front and rear steering angles, and the
reference front and rear axle lateral forces, calculated through a bicycle control laws for the front and rear axles can be separately designed. A
vehicle model. different selection of the longitudinal position of the control points
Automated driving controllers (Gao et al., 2010; Kim et al., 2014; would imply the design of a multi-variable controller for a 4WS path
Yin et al., 2015; Zhou et al., 2005) track a reference path, which is tracking implementation. Many references on path tracking control with
calculated through methodologies that are beyond the scope of this preview based on model predictive control, e.g., Falcone, Borrelli et al.
paper. The path tracking errors (Chatzikomis et al., 2018) are often (2007), tend to use error variables, ΔzCG , considering the predicted
expressed in terms of lateral displacement and heading angle errors at profiles of the heading angle and lateral displacement errors, calculated
the centre of gravity: through the internal vehicle model of the MPC, based on the knowledge
( ) ( ) of the future reference trajectory and the predicted actual trajectory of
ΔyCG = YCG − Yref cosψ ref − XCG − Xref sinψ ref
(12) the vehicle:
Δψ CG = ψ − ψ ref
ΔzCG = [Δψ CG ΔyCG ]T (16)
In Kapania and Gerdes (2015), dealing with stand-alone path
tracking control, the lateral displacement error is projected at a In some cases, the error variables also include the time derivatives of
look-ahead distance ld in front of the current position of the vehicle, see the errors defined in (16), or other vehicle states, such as yaw rate,
Fig. 7(a): sideslip angle, or vehicle speed, see also the formulations in Chatziko­
mis et al. (2018), Chen et al. (2019), Hajiloo et al. (2020), and Wu et al.
Δyld = ΔyCG + ld Δψ CG (13)
(2020). Fig. 7(c) is a graphical representation of the lateral displacement
A similar formulation is presented in Xia et al. (2020). However, the errors along the preview distance, according to the linear quadratic
vehicle cornering behaviour resulting from a feedback controller based formulation with preview in Chatzikomis et al. (2018). Many other
on (13) gives origin to a non-zero lateral displacement error at the centre formulations are available for the computation of the references for path
of gravity (see Fig. 7(a)), which can be compensated through the tracking control in automated vehicles. For example, Lee (2002) uses the
following modified formulations (Kapania and Gerdes, 2015): curvature of the trajectory as reference, while Cheng et al. (2019),
Chowdhri et al. (2021) also includes the desired distance between the

Fig. 7. Examples of path tracking control set-ups: (a) based on look-ahead error, (b) based on modified look-ahead error, and (c) with preview, adapted from
Kapania and Gerdes (2015) and Chatzikomis et al. (2018).

184
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

ego vehicle and the preceding one, which is typical of the literature degradation of the heave response, which, for example, could be
including collision avoidance conditions. addressed through a CeC architecture.
The discussion of this and the following sections does not deal with
the details of electric power steering systems (EPS), as they only have an 3.4. Summary on reference vehicle models and variables
effect on the steering wheel torque and subjective driver experience, but
they do not exert a direct influence on the vehicle motion, for a given As a summary of the discussions in Sections 3.1-3.3, Fig. 8 graphi­
steering wheel angle profile. cally shows the main reference variables and control actions (see the
nomenclature as well) adopted in the considered ICC implementations
from the literature, while Table 3 also reports the models used for
3.3. Attitude control reference state generation in a representative set of ICC studies.

The continuous damping control (CDC) systems, active roll control 4. Coordination strategies
systems (ARC), and active suspension systems (ASS) installed on pro­
duction cars are mainly designed to improve ride quality and comfort by This section describes the characteristics of the main ICC coordina­
reducing the roll, pitch and heave motions, caused either by road ir­ tion strategies, which, in accordance with the categorisation in Section
regularities or the longitudinal and lateral accelerations of the vehicle. 2, are divided based on the downstream or upstream configuration of the
Typical suspension controllers for ride comfort improvement set zero respective architectures.
reference values for the relevant body dynamics variables, e.g., the
heave, pitch and roll accelerations and displacements (Lou et al., 2010; 4.1. Coordination strategies for downstream architectures
Shen and Yu, 2006a). Other authors include consideration of the road
holding aspects, by setting zero reference values for tyre deflection PeC does not require any coordination strategy, as it fulfils the
(Tchamna et al., 2014), or, equivalently, dynamic tyre load (Chen et al., integration via shared information (Sato and Inoue, 1993, Lou et al.,
2006). The ICC strategies in Fergani et al. (2017), and Vivas-Lopez et al. 2010, Taheri and Law, 1990), or by positively influencing the control
(2015) adopt rule-based suspension controllers to prioritise road hold­ actions of the other actuators, which is still quite effective and often
ing or comfort, starting from the suspension states. Based on the expe­ prevents global failures. In the historical example by Mitsubishi (Mita­
rience of the authors, industrially implemented vehicle body controllers mura et al., 1988), the coordination is achieved through the hydraulic
using active suspension actuators compensate the vehicle body motion connection between the subsystems, without any data sharing via net­
caused by the longitudinal and lateral vehicle dynamics through refer­ worked communication. Alleyne (1997), and Ting and Lin (2004)
ence anti-roll and anti-pitch moments, typically calculated from the calculate the suspension control action by using information from the
measured longitudinal and lateral vehicle accelerations (Li et al., 2008; active braking system. Scalzi and Marino (2008) defines the steering
Rahimi and Naraghi, 2018). Her, Suh, et al. (2015) compute the desired actuation while considering the semi-active differential control action.
roll angle for the active suspension system from the lateral acceleration, In Tang and Khajepour (2020) optimisation iterations in the distributed
and integrate anti-roll moment and direct yaw moment actuations, to vehicle corner controllers are implemented (see Section 2.3) based on
reduce roll angle and roll rate during stability control activations. the information exchange on the generalised actuation forces at the
ARC and ASS can support the rollover prevention function of VSC vehicle corners, until convergence is reached in all controllers.
systems. For instance, the cost function in Yim (2012) considers terms to In CoC architectures, the coordination strategies are usually rule-
minimise roll angle, roll rate and roll acceleration. Rahimi and Naraghi based, and consider the effect of the specific actuator on the vehicle
(2018), similarly to Yoon et al. (2008), compute a rollover index, RI1 , response. Gordon et al. (2003), and Kissai et al. (2017) mention that
which considers the roll and lateral motions of the sprung masses: arbitration strategies for ICC based on CoC frequently use (see also
⎧ ⃒ ⃒
⎪ |ϕ|ϕ̇lim +|ϕ̇|ϕlim ⃒ay ⃒ ϕ Fig. 9): (a) artificial neural networks (ANN); (b) fuzzy logic control
⎪
⎪
⎪
⎪
K1 +K2
a
+(1− K1 − K2 ) √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅, ϕ(ϕ̇+K3 ϕ)〉0 (FLC); (c) pure subsumption (PS); and (d) the largest modulus activation
⎨ ϕlim ϕ̇lim y,lim 2 2
RI1 =
ϕ + ϕ̇ (LMA), which are discussed in the following subsections.
⎪
⎪ ⃒ ⃒
⎪
⎪ ⃒ay ⃒
⎪
⎩ K2 , ϕ(ϕ̇+K3 ϕ)≤0 4.1.1. Artificial neural networks and fuzzy logic coordination
ay,lim
ANN can coordinate decoupled actuators by adapting to unknown or
(17)
uncertain parameters, and have the potential advantage of using
Gáspár et al. (2009), and Tchamna et al. (2014) provide an alter­ nonlinear interpolation functions preventing actuator saturation and
native rollover index formulation, which can also serve as a wheel lift-off allowing smooth transitions between different coordination modes. On
detection variable: the downside, the number of nodes can affect complexity, turning the
ANN into a complex nonlinear multivariable controller, sometimes
|(FzFL + FzRL ) − (FzFR + FzRR )|
RI2 = (18) resulting in poor robustness and interference among actuators (Gordon
(FzFL + FzRL ) + (FzFR + FzRR )
et al., 2003). However, although mentioned in the previous ICC surveys,
Its value should range from 0 to 1; if RI2 > 1, lift-off has already the considered literature does not include ANN examples for ICC
occurred, hence Soltani et al. (2018) deems the vehicle behaviour as coordination.
hazardous when the factor reaches the critical threshold of 0.8. FLC can coordinate the stand-alone chassis actuation systems of CoC
Katsuyama (2013) highlights that vehicles equipped with in-wheel architectures for a limited number of coupled or uncoupled control
motors, which are part of the unsprung mass, generate a vertical sus­ objectives, and, unlike ANN, gives origin to easily predictable decisions.
pension reaction force during driving. This phenomenon permits to The challenge of FLC design is in the definition of the most suitable
independently control the roll and pitch motions, while ensuring the membership functions. For example, Cao and Zheng (2019), Elhefnawy
required longitudinal and yaw dynamics. Accordingly, the PeC imple­ et al. (2017), March et al. (2007), and Shao et al. (2007) use membership
mentation in Ricciardi et al. (2019) embeds a pitch controller into the functions that only depend on the tracking error, e.g., to coordinate a
EBD algorithm to reduce the control effort of the ASS. The target of the controlled suspension system with another chassis actuator. Wang et al.
ICC implementation is the reduction of the pitch motion and the pro­ (2009) defines the weights of each subsystem contribution by means of
vision of the expected acceleration/braking performance; in particular, membership functions depending on the yaw rate error, the magnitude
the vehicle body control module “acts to minimise the acceleration of the of the total tangential vehicle acceleration, and sideslip angle; the
top mount positions” of the suspension systems. However, the price is a resulting ICC system presents good handling and stability performance

185
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Fig. 8. Main reference variables (in magenta) and control actions (in red) in the lateral (a), rear (b) and top (c) views of the vehicle.

while preserving desirable longitudinal dynamics. The previous strategy 4.2. Coordination strategies for upstream architectures
is extended in Rahimi and Naraghi (2018) to include roll control
contribution via the rollover index in (17). He et al. (2006) adopts a Within upstream architectures, the coordination is usually achieved
membership function based on the distance of the current conditions of through the so-called control allocation algorithms, which are often
the car from the desirable region in the β − β̇ phase-plane (see Section 3), supported by supervisory decision strategies. The supervision layer is
to ensure smooth transition from AFS to VSC in critical situations. especially useful in MuC and SuC, e.g., to simplify the optimisation
problem, reduce the computational effort, and facilitate real-time
4.1.2. Pure subsumption and largest modulus activation implementation of the CA algorithms. In some cases, it is rather diffi­
Gordon et al. (2003) defines PS and LMA as coordination strategies cult to determine the boundary between supervisory layer and CA
based on conflict resolution between the controlled behaviours. PS as­ strategy; nevertheless, the following sub-sections discuss the algorithms
signs rankings to the different actuators, and sequentially sends the from the literature according to this categorisation.
commands based on the ranking, i.e., it uses only the top-ranked actu­
ator until its saturation, after which it activates the second one as well, 4.2.1. Supervisory decision strategies
and so on. Velardocchia and Vigliani (2013) implements this technique The supervisory layer makes strategic decisions to enhance vehicle
to prioritise ARC over the other actuators until this system reaches its performance while avoiding any “synthetic” driving feeling, which
limit, point at which TV intervenes, followed by the VSC. As the actuator could arise without smooth transitions among actuators. In most su­
potential depends on tyre force nonlinearity, the strategy also needs an pervisory layers, appropriate indices identify the driving situation, and,
adaptive algorithm that detects the operating condition of the tyres. based on the prior knowledge of the system, switch among control
LMA activates the actuators by prioritising those that can compensate modes or the available actuators. In other cases, the control modes
the highest endangering behaviour, e.g., defined based on the largest selected by the supervisory layer modify the reference values for lateral
normalised error between the reference and actual states. For example, dynamics control (Cho et al., 2012).
the LMA architecture in Xie et al. (2018) sets the priorities according to Bedner and Chen (2004), Burgio and Zegelaar (2006), Doumiati
the risk of stability loss in the β − β̇ phase-plane. et al. (2013), Hou et al. (2008), and Yim and Jo (2019) prioritise the yaw
In general, ANN and FLC are preferable over PS and LMA because moment contribution of the steering actuation, based on 4WS or AFS,
they do not involve abrupt changes in control action when the coordi­ over the direct yaw moment of the braking system and powertrain, to
nator switches between modes, thus preventing undesirable transients. prevent any influence on the longitudinal vehicle dynamics until the
Nevertheless, PS and LMA are still worth being considered since they do yaw moment generation capability of the steering system saturates. In
not require any additional rule in case of failure, as the chassis systems some examples, each control mode formalised in the supervision layer is
work in parallel. Moreover, Gordon et al. (2003) suggests the possibility related to the activation or deactivation of actuators, e.g., in Xia et al.
of low-pass filtering the activated outputs, or including ramped transi­ (2019) DYC and ASS are alternatively activated. In several supervisory
tions to cope with abrupt variations of the control actions. layer implementations, the control modes depend on the tracking error
and/or the magnitude of a vehicle state (Cheng et al., 2019; Cho et al.,
2012; Her, Suh, et al., 2015; Hwang et al., 2007; Kim et al., 2013;
Poussot-Vassal et al., 2011; Yoon et al., 2008; Vivas-Lopez et al. 2015).
In many supervisory strategies, the measured ay and ψ̇ are manipulated

186
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Table 3
Overview of the selected reference models, reference variables, and control actions in the considered ICC studies.
Target behaviour Ref. model/assumptions Ref. states Control References examples
action

Lateral dynamics αi,ref = αi,peak,0 αF , αR Mz Joa, Yi, et al., (2018)

ST Bernardini et al. (2009); Di Cairano et al. (2013)
NL-DT ψ̇ Alberding et al. (2014)
ST Joa, Park, et al. (2018); Khosravani et al. (2018); Kim et al. (2013); Nah, Yim
(2019); Xia et al. (2019); Yim, Jo (2019); Zhang et al. (2018)
δ, FAct March et al. (2007)
δ, TB Chen et al. (2012); Hirano et al. (1993); Ono et al. (1994)
δ, Mz Doumiati et al. (2011); Guo et al. (2017); Kissai et al. (2018); Mirzaei,
Mirzaeinejad (2017); Németh et al. (2017); Saikia, Pathak (2019)
NL-ST; NL tyre ψ̈ , β̇, δ̇ Warth et al. (2020)
ST Fyij Li, Arat (2016)
ψ̇ , β Cheng et al. (2020); Fan, Zhao (2019); Fu et al. (2017); Mousavinejad et al.
(2017); Nagai et al. (2002); Peters, Stadelmayer (2019); Xiujian et al. (2009)
ST; β-β̇ He et al. (2006)
ST; β = 0 Boada et al. (2006); Nagai et al. (2002); Shuai et al. (2013); Xie et al. (2018);
Zhao et al. (2015); Zhao et al. (2017); Zhu et al. (2014)
ψ̇ , vy Mz , F x , δ Shen, Yu (2006b)
δ, TM Tang, Khajepour (2020)
ST ψ̇ , β δ, Mz , Fz Abe, Mokhiamar (2007); Ahangarnejad (2018); Ahangarnejad et al. (2019);
Shyrokau, Wang (2012)
ST ψ̇ , vy δ, TB Burgio, Zegelaar (2006); Ding, Taheri (2010);
ψ̇ , ϕ, ti δ, Mz , Mϕ Gáspár, Németh (2016)
Lateral dynamics (incl. path ST; Kin; β = 0 vx , Y, ψ , ψ̇ δ, Fx , Mz Ren et al. (2018)
tracking) Kin δ, TM Xia et al. (2020)
δ, TB Falcone et al. (2008)
Y, ψ , ψ̇ , dego Chowdhri et al. (2021)
vx , Y, ψ , ψ̇ δ, Mz Wu et al. (2020)
Kin; ψ̇ = map(ax , V, δ) vx , X, Y, ψ̇ Chatzikomis et al. (2018)
ST; Kin vy , Y, ψ , ψ̇ Hang, Chen (2019)
ST; Kin vx , X, Y, ψ̇ Lin et al. (2019)
Y, ψ , ψ̇ , β Guo et al. (2018); Hang et al. (2019); He et al. (2018)
vx , vy , Y, ψ Chen et al. (2019); Hajiloo et al. (2020)
Lateral dynamics and attitude ST; ϕ = 0 ψ̇ , ϕ δ, Mϕ Wang et al. (2018)
control ST; ϕ = map(ay ) M z , Mϕ Her, Suh, et al. (2015)
ψ̇ , β, ϕ δ, Mz , Mϕ , Mθ Chokor et al. (2019)
ST; ϕ = 0 δ, Mz , Mϕ Elhefnawy et al. (2017)
ST ψ̇ FAct , Mz Yim (2012)
ST; zs = z̈u = ϕ̈ = θ̈ = ψ̇ ,β,zs , z̈u , ϕ̈, θ̈ Tchamna et al. (2014)
β =0
ST; β = 0 ψ̇ , β, żs Soltani et al. (2018)
ST; β = θ̈ = zs = z̈s = ψ̇ , β, zs , z̈s , δ, FAct Shen, Yu (2006a)
0 θ̈
Lateral dynamics (incl. path ST; Lg-Kin; β = 0 ψ̇ , vx , vy Mz , Mϕ , Fx , δ Li et al. (2008)
tracking) and attitude control ST; Lg-Kin; β = vz = θ̇ = ψ̇ , θ̇, ϕ̇, vx , vy , Mz , Mϕ , M θ , Zhao et al. (2019)
ϕ̇ = 0 vz Fz , Fx , δ
Longitudinal dynamics and σx,ref = σx,peak σx FAct , TB Alleyne (1997)
attitude control z̈s = θ̇ = 0; σx,ref = σx , ax , z̈s , θ̇ Shao et al. (2007)
σx,peak
σx,ref = σx,peak ; zs = żs = zs , żs , σx Lou et al. (2010)
0
Longitudinal and lateral dynamics ST; σx,ref = σx,peak ; β = 0 ψ̇ , β, σx δ, TB Plochl, Lugner (1996)
ST; ax = map(Fped ); β-β̇ ψ̇ , β, ax Fx , δ, Mz Shyrokau et al. (2015)
ST; σx,ref = σx,peak ; Lg- ψ̇ , β, ay , σx , δ, Mz TB , TM Wang et al. (2009)
DT-L ax
Longitudinal (incl. vx tracking) ST; Lg-Kin; ax = ψ̇ , vx , ax Fx Joa et al. (2015)
and lateral dynamics map(Fped )
ST; Lg- Kin; β = 0 ψ̇ , vx , β Mz , F x , δ Li, Yu, (2007); Shen, Yu (2007); Song et al. (2015); Zhu et al. (2019)
ST&NL-DT; Lg- Kin; β=0 F x , Mz Cho et al. (2011, 2012)
Kin; β = 0; vx δ, σx Chang, Gordon (2008); Kirli et al. (2019)
ST; Kin ψ̇ , vx , ax , β, Fx Cheng et al. (2019)
dego
Longitudinal and lateral dynamics, ST; σx,ref = σx,peak ; Lg- ψ̇ , β, ay , σx , δ, Mz , Mϕ , TB , Rahimi, Naraghi (2018)
attitude control DT-L; RI ax TM

(continued on next page)

187
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Table 3 (continued )
Target behaviour Ref. model/assumptions Ref. states Control References examples
action

ST; Lg-Kin; RI ψ̇ , ax , ϕ Fx , Mz , Mϕ Yoon et al. (2008)

ST; ϕ = map(ay ); ax = Her et al. (2016)
map(Fped )

to determine stability indices and thresholds. For example, Her, Koh, possibility of adapting the vehicle behaviour to the desire of the driver,
et al. (2015), Her, Suh, et al. (2015), and Joa et al. (2016) define a su­ following a personalisation approach. A custom mode is included, in
pervisory state machine based on an index, Istatus , and lateral accelera­ addition to passive, drift, sport and safe modes. The custom mode allows
tion, ay , as described in Fig. 10. Istatus depends on the yaw rate tracking the driver to tune the reference states for lateral stability, understeer
error, and assesses whether – and how much – the vehicle is oversteering characteristics and active steering responsiveness, to promote or restrict
or understeering: the control interventions. Furthermore, the modes allow tuning how
much the driver will notice the ICC interventions, acoustically or
ψ̇ ref − ψ̇
Istatus = (19) palpably, to build confidence in the vehicle (Schmidt and König, 2020).
ψ̇ stab,μ Strategies along the lines of the one in König et al. (2014) are used in
Istatus and ay define three control regions, namely agility, manoeu­ high-performance passenger cars, such as the Porsche Taycan, Lam­
vrability and stability control, which are used to prioritise 4WS, VSC and borghini Huracan, Pagani Huayra, Honda NSX, and Ferrari SF90. For
ASS. As soon as the vehicle is experiencing an understeer/oversteer example, the hybrid electric Ferrari SF90 allows selecting among eDrive,
tendency and/or risks losing stability, the control priorities are modi­ Hybrid, Performance and Qualify modes, where Qualify uses the
fied. This approach is very effective from the implementation viewpoint, maximum potential of the batteries and electric motors, prioritising
but attention is required in the definition of the thresholds and in state performance over efficiency and durability (BBC Top Gear Magazine,
transitions, with possible abrupt changes compromising the smoothness 2019).
of the control action. The supervisory strategies can eventually overcome the driver
The control modes can also depend on the driver input. For instance, command to preserve safe operation. In this respect, Chang and Gordon
König et al. (2014) provides a human-machine interface that grants the (2008), and Gao and Gordon (2019) include an emergency switching

Fig. 9. Main coordination strategies for downstream architectures, namely (a) ANN; (b) FLC; (c) PS; and (d) LMA, to be included in the coordination block in (e),
adapted from Kissai et al. (2017).

Fig. 10. Schematic of an index based supervisory strategy, adapted from Her, Suh, et al. (2015).

188
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

strategy between passive and assisted driving, where the assisted mode understeering; and Her, Suh, et al. (2015) uses Istatus and the lateral ac­
takes over the driver commands when the system assesses incorrect celeration to allocate the required direct yaw moment between the front
judgement by the human driver. and rear axles.
An important requirement is to prevent abrupt control action vari­ Gordon (1996), Nah and Yim (2019), Schiebahn et al. (2010),
ations that can be unpleasant for the driver and passengers; this can be Shyrokau et al. (2015), and Yim et al. (2012, 2016) use pseudo-inverse
managed through smooth weight scheduling in the switching logics, formulations for ICC CA, under the assumption of system linearity, e.g.,
which accounts for the transition between two driving conditions and see Durham et al. (2017), and Johansen and Fossen (2013):
gradually assigns more weight to one control action over another. For ( )− 1
instance, Cho et al. (2012), Chokor et al. (2019), Mousavinejad et al. U = W − 1 BT BW − 1 BT τc (20)
(2017), and Zhang et al. (2018) smoothly scale weights that depend on
where τc is the virtual control vector, i.e., the vector with the vehicle-
the deviation from the desirable region in the β − β̇ phase-plane. Joa, Yi,
level objectives to be achieved, for example the total yaw moment; B
et al. (2018) introduces a smooth control input switch based on the α −
is the control effectiveness matrix, which expresses the link between the
Fy phase-plane of the rear axle in Fig. 11, where the peak line is the
individual control actions, included in the vector U, and the vehicle-
locus of the maximum lateral axle force at different μ values. This su­
level effect, i.e., τc = BU; and W is a weighting matrix that prioritises
pervisory method defines whether the lateral behaviour of the vehicle is
the different actuators.
unstable, namely when α is beyond the threshold in (11). The smooth
Yim et al. (2012) defines a fault-tolerant CA implementation based
scheduling can also account for physical constraints or actuation effec­
on the daisy chaining method, where the effectiveness matrix is split into
tiveness to softly vary the control action allocation. Gáspár and Németh
two parts, i.e., B = [B1 B2 ], where the actuators related to B2 are used
(2016), and Németh et al. (2017) adopt a smooth switching strategy
only when the actuators related to B1 cannot provide τc on their own.
considering the adhesion potential and the maximum steering angle as
Most of the recent CA implementations adopt on- or off-line opti­
possible thresholds. Gáspár et al. (2009) uses RI2 to assess when ASS is
misation algorithms, considering a cost function and a set of equality
not able to supply enough anti-roll moment to counteract an incipient
and inequality constraints, according to the following formulation
rollover condition. In this case, a weight scheduling strategy demands
(Alberding et al., 2014; Her et al., 2016; Joa et al., 2016; Joa, Park, et al.,
heavier VSC interventions. Importantly, smooth switching logics, if
2018; and Johansen and Fossen, 2013):
supported by fault detecting and identification algorithms, can cope
with active system failures by gradually deactivating an actuator and min ‖ QsV − J(X, U, t) ‖
U∈Rp ,sV ∈Rm
allocating higher efforts to the remaining ones, without degrading s.t.
driving performance and stability (Gáspár et al., 2009; Németh et al., τc − b(X, U, t) = sV , U ∈ U (21)
2017; Savitski et al., 2015). Gáspár and Németh (2016) presents a U = Uℓ + ΔU
reconfigurable fault-tolerant strategy; in case of fault, the yaw moment ΔU ∈ C
contributions of friction brakes, AFS, and ASS are changed by modifying G(X, U, t) ≤ 0
the scheduling weights with respect to the relevant variables. Fergani
et al. (2017) use an ay -based index to switch the suspension actuation where the notation ‖ ⋅ ‖ indicates a norm; Qis a weight matrix that
prioritises the requirements that should be met if the virtual control
from comfort to roll control, whenever the other chassis systems fail to
vector cannot be achieved; sV is a slack variable; J is the secondary cost
prevent rollover conditions.
function, which can be introduced and minimised as dimU > dimτC ,
namely the system is over-actuated, and, hence, the solution is not
4.2.2. Control allocation
unique; b is the control effectiveness function; X is the state vector; G is
The purpose of CA algorithms is to distribute the control action
the function expressing the inequality constraints; and t is time. Limits
among multiple redundant actuators.
for ΔU, that is the change in control action with respect to its value Uℓ at
The simplest CA strategies are rule-based. For example, Zhu et al.
the last sampling point, are often introduced in CA formulations.
(2019) uses rules for deciding the torque distribution between the
Optimisation-based CA can mathematically account for many factors, to
electric powertrain and friction brake of each vehicle corner; Falcone
cope with changing conditions, dynamics interactions, actuator limits,
et al. (2008) allocates the direct yaw moment among the four vehicle
and system stability. For this reason, these allocation methods are
corners according to the difference between the front and rear slip an­
particularly suitable for multi-objective ICC. Common examples of sec­
gles, to take into account whether the vehicle is oversteering or

Fig. 11. Stable and unstable regions in the lateral tyre force characteristics, adapted from Joa, Yi, et al. (2018).

189
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

ondary cost function formulations (Johansen and Fossen, 2013) are: uncertainties on actuator dynamics causes a delay in matching the vir­
tual control vector, and proposes the combination of an MPC based CA
1
J(X, U, t) = (U − UP )T W(U − UP ) implementation with an online adaptive estimation of the parameters
2 (22)
defining the dynamics of the specific steering and braking actuators. In
J(X, U, t) = ‖ WU ‖
general, adaptive parameter estimation is a rather common feature that
is combined with MPC implementations.
where W ∈ Rp×p is a positive definite weighting matrix that prioritises
For the analysed ICC literature, Table 4 shows examples of:
the actuators; and UP is the preferred value of U, e.g., corresponding to
zero effort of the actuators.
• Variables used in τc , e.g., the longitudinal force, lateral force, roll
Appropriate methods, e.g., active set and interior point, are used for
moment, total yaw moment, and direct yaw moment, calculated at
the solution of the resulting optimisation problem (Nocedal and Wright,
the vehicle level; or the longitudinal tyre forces for each corner.
2006). In the on-line implementations, the optimisation is solved in the
• Cost functions, typically related to tyre workload or tyre slip power
vehicle control unit, by using simplified models of the vehicle system,
losses, which require advanced state estimators of the tangential tyre
with reduced number of degrees of freedoms to limit the computational
forces.
effort. In the off-line cases, the solution of the CA problem, e.g., gener­
• Constraint formulations, which usually limit: a) control effort and
ated through off-line optimisation routines using quasi-static nonlinear
effort rate of the powertrain, brake, suspension and steering actua­
vehicle models (De Novellis et al., 2013), is stored in look-up tables in
tors; b) vehicle variables such as slip ratios, angular wheel speeds,
the vehicle control unit. The selection between on-line and off-line CA
and slip angles; c) tyre operation, to remain within the available
depends on the trade-off between accuracy and complexity of the model
friction limits; and/or d) system dynamics, e.g., the yaw moment.
used in the optimisation problem, and the computational power and
flash memory capabilities of the available control units. With the
The literature includes several examples of CA based on fuzzy logic
enhanced performance of recent automotive control hardware, which
control, see Fig. 12, thanks to the possibility of transforming different
allows the real-time execution of optimisation routines (Her et al.,
vehicle dynamics scenarios into linguistic variables with membership
2016), the on-line approaches are becoming more common, as they
functions ranging between 0 and 1. For example, Zhao et al. (2017)
allow easier adaptation to the operating scenario, and more agile
adopts fuzzy logic for the allocation strategy in the intermediate ICC
parameter tuning while the vehicle is tested.
layer, where a particle swarm algorithm modifies the shape of the
A clear trend in the recent literature, e.g., see Hajiloo et al. (2020),
membership functions. Ahangarnejad (2018) applies fuzzy logic control
Kirli et al. (2019), Lin et al. (2019), Wu et al. (2020), and Xiang et al.
as coordination strategy among rear wheel steering and TV actuations.
(2020), is to use model predictive control (MPC) for CA, e.g., the control
The membership functions output weights depending on steering angle,
action is split among the redundant actuators by considering the pre­
speed, yaw rate and sideslip angle. To define the weights, an optimisa­
dicted future behaviour of the system in the optimisation problem. A
tion algorithm minimises the weighted sum of the normalised yaw rate
typical optimal control problem formulation is:
and sideslip angle overshoots. With respect to neural network imple­
( ( )) N∑
c− 1 N
∑ p− 1 mentations, Wang et al. (2018) presents a rare example of ANN-based
min J(X(0),U(⋅)):= ℓNp X Np +
U
ℓ(X(k),U(k))+ ℓ(X(k),U(Nc − 1)) CA algorithm, calculating the individual control actions starting from
the vehicle error variables.
k=0 k=Nc
s.t.
Trachtler (2004) highlights that CA algorithms must be capable of
X(0)=Xin
managing partial shut-down events of faulty chassis systems, and the
X(k+1)= fd (X(k), U(k))
re-distribution of the control action among the functioning ones. For
X ≤X(k)≤X
( ) example, Alberding et al. (2014) defines constraints on the roll dynamics
X ≤X Np ≤X to prevent rollover when the suspension actuators are malfunctioning.
U ≤U(k)≤U Moreover, many ICC implementations include the combination of mul­
G(X(k),U(k))≤0 tiple techniques, e.g., supervisory decision strategies and CA, or on-line
(23) and off-line optimisations. For example, the MuCs in Lin et al. (2019),
and Xiang et al. (2020) use MPC to compute the steering angle for an
where the notation U(⋅) indicates the control sequence; Xin is the initial AFS system, and the direct yaw moment for controllable friction brakes
value of the state vector, obtained from the available sensors and state and multiple electric motors, where the individual torque values are
estimators; Np is the number of steps of the prediction horizon HP , i.e., output by a further CA algorithm. In the intermediate layer of their ICC
Hp = Np Ts , with Ts being the discretization time; k indicates the dis­ system, Mousavinejad et al. (2017), and Zhao et al. (2017) adopt
cretization step along the prediction horizon; X and X are the lower and weighting coefficients, respectively based on the β − β̇ phase-plane and a
upper limits for X; U and U are the lower and upper limits for U; X(k +1) fuzzy logic, to allocate the yaw moment contributions to be generated
= fd (X(k), U(k)) is the discretized model of the system; ℓ(X(k), U(k)) is through AFS and DYC, while a further CA layer determines the indi­
the stage cost function associated to each time step; ℓNp (X(Np )) is the vidual wheel torque levels. In some papers, such as Khosravani et al.
terminal cost; and Nc is the number of steps of the control horizon Hc , i. (2018), the CA weights are fixed, i.e., they do not change with the
e., Hc = Nc Ts . A typical stage cost formulation is: vehicle parameters and driving conditions. To achieve adaptation to the
operating scenario without significantly increasing complexity, other
ℓ(X(k), U(k))=‖ZV (k) − ZV,d (k)‖2S + ‖ U(k) ‖2R (24)
ICC examples, e.g., Németh et al. (2017), Shyrokau, Wang (2012, 2015),
and Yim (2012), propose scheduling of the CA weight matrices, and
where ZV (k) and ZV,d (k) are the output vector and its desired value; and S
reduce the computational effort by including conditions to limit the
and R are weight matrices, which respectively prioritize the tracking
number of iterations within the CA algorithm. Schiebahn et al. (2010)
performance of the different references, and penalize the control effort
analyses off-line the maximum total yaw moment potential of each
of the individual actuators. A linear quadratic (LQ) controller can be
considered as a particular case of (23)-(24). actuator, see Figure 13. In the on-line CA algorithm, U =
Many studies based on MPC for CA neglect actuator dynamics, e.g., [uact1 ; uact2 ; …; uactn ]T consists of nondimensional values, according to:
see Borrelli et al. (2005), which can affect performance. Other re­ Mztot = Mz,max,act1
tot tot
uact1 + Mz,max,act2 tot
uact2 + … + Mz,max,actn uactn (25)
searchers, e.g., Kim et al. (2014), assume that the actuator dynamics are
known and time-invariant. Chatrath (2019) highlights that ignoring the where Mtot tot
z,max,act1 ,…, Mz,max,actn are the maximum total yaw moments,

190
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Table 4
Overview of the most frequent formulations adopted for CA in ICC systems.
Examples of variables included in τc
Variable name and symbol References

Anti-roll moment, Mϕ Her et al. (2016); Her, Koh, et al. (2015)

Direct yaw moment, Mz Alberding et al. (2014); Her et al. (2016); Her, Koh, et al. (2015); Joa et al. (2016); Joa, Park, et al. (2018); Li, Yu (2007); Ren et al. (2018);
Shyrokau et al. (2013); Song et al. (2015)
Individual longitudinal tyre forces, Fx,ij Joa et al. (2016)
Total longitudinal force, Fx Cho et al. (2012); Her et al. (2016); Her, Koh, et al. (2015); Joa, Park, et al. (2018); Kissai et al. (2018); Li et al. (2007); Li, Yu (2007); Song
et al. (2015)
Total lateral force, Fy Li et al. (2007); Li,Yu (2007); Shyrokau et al. (2013); Song et al. (2015)
Total yaw moment, Mtot
z Cho et al. (2012); Kissai et al. (2018); Li et al. (2007); Nah, Yim (2019); Yim et al. (2012, 2016)

Examples of terms included in the cost functions

Definition Formulation example References

Maximisation of front lateral tyre forces Gyσx (FxF,AWD , FxFj,VSC )GFz ,ARC (ΔFzF,ARC )FyF,0
2 Her et al. (2016)
∫ ∑ ∑
Minimisation of electric and mechanical ( Vxij Fxij + Rel F2xij )dt Edrén et al. (2019)
energy consumption
∑
Minimisation of lateral tyre dissipation (|Vxij Fxij | + |Vyij Fyij |) Joa, Park, et al. (2018)
power on each wheel
∑
Maximisation of power consumption max(0, Pemij ) Shyrokau et al. (2013, 2015)
1−
coefficient of electric motors 4Pem,max
∑
Maximisation of power recuperation of min(0, Pemij ) Shyrokau et al. (2013, 2015)
1−
electric motors 4Pem,max
∑
Maximisation of tyre energy efficiency |Fxij Vslipij | Shyrokau et al. (2013, 2015)
1− ∑
during Fx generation |Fxij Vxij |
Maximisation of tyre energy efficiency Shyrokau et al. (2013, 2015)
during Fy generation 1−
∑
|Fyij Vyij |
∑ ⃒ ∑
0.5(CαF |VxFj ⃒ + CαR VxRj )αpeak,0 tanαpeak,0
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Minimisation of tyre saturation penalty P2ij + b2hyp − Pij Guo et al. (2017); Joa et al. (2016); Joa, Park, et al. (2018); Khosravani et al. (2018);
∑
function Reinold, Traechtler (2013)
μFzij
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Tyre workload minimisation F2xij + F2yij Abe, Mokhiamar (2007); Ando, Fujimoto (2010); Ivanov, Savitski (2015); Ono et al.
∑
(2006); Reinold, Traechtler (2013); Song et al. (2015); Yim et al. (2012)
∑ μFzij
Minimisation of motor inefficiency MMij /ηij Jalaliyazdi (2016)
∑
MMij
Examples of constraints
Definition Formulation example References

Free rolling wheel Fxij ≤ 0 Cho et al. (2008, 2012); Yoon et al. (2008)
Friction ellipse F2xij + F2yij ≤ (μFzij )2 Cho et al. (2008, 2012); Her, Suh, et al. (2015); Her et al. (2016); Joa, Yi, et al. (2018); Li et al.
(2008); Shen, Yu (2006b); Song et al. (2015); Zhang, Li (2019); Zhao et al. (2019)
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Maximum pB pB ≤ pB,ℓ + ε Ts p* − pB,ℓ Zhu et al. (2019)
Maximum MM MMij ≤ MM,max Borrelli et al. (2005); Falcone, Tseng, et al. (2007); Guo et al. (2018); Hang, Chen (2019); Joa,
Park, et al. (2018); Khosravani et al. (2018); Ren et al. (2018)
Maximum ΔMM ΔMMij ≤ ΔMM,max Edrén et al. (2019); Falcone, Tseng, et al. (2007); Rengaraj,Crolla (2011)
Maximum Δδ Δδ ≤ Δδmax Edrén et al. (2019); Falcone, Tseng, et al. (2007); Rengaraj, Crolla (2011); Shen, Yu (2006b)
Maximum δ δmin ≤ δ + Δδ < δmax Borrelli et al. (2005); Edrén et al. (2019); Falcone, Tseng, et al. (2007); Guo et al. (2017); Li, Yu
(2007); Li,Arat (2016); Zhang, Li (2019);
√̅̅̅̅̅̅̅̅
Minimum pB pB ≥ max(pB,ℓ + ε Ts pB,ℓ , 0) Zhu et al. (2019)
Rollover constraint ϕ ≤ ϕlim Alberding et al. (2014)
( ) (
Total controllable yaw tf FxRL,max tf Cho et al. (2008, 2012); Yim et al. (2016); Yoon et al. (2008)
− FxFL 1 + | | + FxFR 1 +
moment constraint 2 FxFL,max 2
) ( )
FxRR,max FzFR
| | + lf FyFL 1 + − Mcontr
z =0
FxFR,max FzFL
αij constraint αmin ≤ αij ≤ αmax Guo et al. (2017); Joa, Park, et al. (2018); Khosravani et al. (2018); Li et al. (2007); Reinold,
Traechtler (2013)
⃒ ⃒
σxij constraint σxij ≤ ⃒σx,max ⃒ Joa et al. (2015); Joa, Park, et al. (2018); Ren et al. (2018)
∑
ωij direction for negative ωij ≥ 0 Khosravani et al. (2018)
MB

Note: Pij = (ahyp (|σxij | − 0.5σxij,peak ))/σ xij,peak

computed off-line, which can be generated by the respective actuators in authors, it does not consider the system dynamics and actuator
the current operating condition of the vehicle. A pseudoinverse formu­ interactions.
lation calculates the initial value of the control actions. Whenever uacti is
greater than one, the algorithm saturates the actuator to its Mtot z,max,acti
value, and splits the remaining required yaw moment among the other 4.3. Control structures adopted in the integrated chassis control literature
actuators. The layout, which is fault-tolerant, resembles the PS coordi­
nation, which allows to add systems without compromising the whole Table 5 summarises the most common control structures adopted
architecture. The algorithm does not require any iteration and is within the ICC implementations considered in this survey. For actuation
computationally more efficient than other solutions, but, as stated by the systems intervening only in emergency conditions, simple feedback (FB)
control structures are appropriate, as comfort and smoothness of

191
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Fig. 12. Simplified diagram of a fuzzy logic CA algorithm.

Fig. 13. Examples of off-line look-up tables expressing the yaw moment potential of different chassis actuation methods, as a function of the relevant states, from
Schiebahn et al. (2010).

intervention are not the priority. However, this is not the case for con­ in question.” As a consequence, feedforward (FFW) control can be used as
trollers characterised by continuous or very frequent activation. Ac­ an unintrusive technique to achieve the desired vehicle response
cording to Bosch (König et al., 2014), FB control can be too intrusive for through continuously active systems. FFW control is also normally less
experienced drivers; in this respect, Schmidt and König (2020) confirm affected by sensor noise, which is a comfort-related issue of FB control.
that for human-driven vehicles “innovative vehicle dynamics solutions are The literature includes several examples of FFW implementations, either
not intended to patronize drivers, rather to ensure that they get the most in the form of equations, e.g., Peters and Stadelmayer (2019), Warth
enjoyment possible from driving. To this end, the systems provide drivers with et al. (2020), and Zhao et al. (2015), or based on maps, e.g., Ahangar­
the exact amount of assistance that is perceived to be helpful in the situation nejad (2018), Ahangarnejad et al. (2019), Chatzikomis et al. (2018), and

192
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Table 5
Summary of the controllers adopted in a sample of the considered ICC literature.
Feedforward (FFW) controllers
Ahangarnejad (2018); Ahangarnejad et al. (2019); Alleyne (1997); Andreasson, Bunte (2006); Chatzikomis et al. (2018); Hirano et al. (1993); Peters, Stadelmayer (2019); Zhao et al.
(2015)
Feedback (FB) controllers
Aut. / RB Burgio, Zegelaar (2006); Falcone et al. (2008); Singh et al. (2013); Soltani et al. (2018); Tchamna et al. (2014); Vivas-Lopez et al. (2015); Xia et al. (2019); Xie et al.
(2018); Yoon et al. (2008)
BStep Ting, Lin (2004)
H∞ Cheng et al. (2020); Doumiati et al. (2011, 2013); Fergani et al. (2017); Hang et al. (2019); Hirano et al. (1993); Kissai et al. (2018); Li et al. (2008); Nagai et al. (1997,
1998); Poussot-Vassal et al. (2011); Shen and Yu (2006b, 2006a, 2007); Sun et al. (2019)
LPV Ding, Taheri (2010); Gáspár et al. (2009, 2016); Németh et al. (2017)
LQ Ahangarnejad (2018); Ahangarnejad et al. (2019); Brennan, Alleyne (2001); Chatzikomis et al. (2018); Chen et al. (2006); Fu et al. (2017); Hang, Chen (2019); Li, Arat
(2016); Matsumoto, Tomizuka (1992); Salehpour et al. (2015); Salman et al. (1992); Shuai et al. (2013); Song et al. (2015); Tchamna et al. (2014); Xie et al. (2018);
Xiujian et al. (2009); Yim (2012); Zhao et al. (2017)
MPC Chang, Gordon (2008); Cheng et al. (2019); Falcone et al. (2007, 2008); Funke et al. (2016); Guo et al. (2017); Hajiloo et al. (2020); Kirli et al. (2019); Kissai et al.
(2018); Mirzaei, Mirzaeinejad (2017); Mirzaeinejad et al. (2018); Ren et al. (2018); Tang, Khajepour (2020); Wu et al. (2020); Xiang et al. (2020); Zheng, Shyrokau
(2019); Zhu et al. (2019)
PID Alberding et al. (2014); Cao, Zheng (2019); Chatzikomis et al. (2018); Ding, Taheri (2010); Hou et al. (2008); Hwang et al. (2007); Li et al. (2008); Plochl, Lugner (1996);
Rahimi, Naraghi (2018); Reinold, Traechtler (2013); Salehpour et al. (2015); Scalzi, Marino (2008); Shyrokau, Wang (2012); Trachtler (2004); Wang et al. (2009); Zhao
et al. (2017); Zhu et al. (2014)
SMC Abe, Mokhiamar (2007); Bang et al. (2001); Cho et al. (2008, 2011, 2012); Chokor et al. (2019); Fan, Zhao (2019); He et al. (2006); Her, Koh, et al. (2015); Her, Suh,
et al. (2015); Her et al. (2016); Joa et al. (2015); Joa, Park, et al. (2018); Khosravani et al. (2018); Kim et al. (2013); Li et al. (2007,2008); Liang et al. (2020); Lou et al.
(2010); Mousavinejad et al. (2017); Nah, Yim (2019); Plochl, Lugner (1996); Rahimi, Naraghi (2018); Ren et al. (2018); Saikia, Pathak (2019); Shyrokau et al. (2013,
2015); Soltani et al. (2018); Song et al. (2015); Wang et al. (2009); Xia et al. (2020); Yim (2018); Yim, Jo (2019); Yoon et al. (2008); Yu, Moskwa (1994); Zhang, Li
(2019); Zhao et al. (2019)
Intelligent controllers
ANN Wang et al. (2018)
FL Boada et al. (2006); Cao, Zheng (2019); Elhefnawy et al. (2017); He et al. (2006); Hou et al. (2008); Li et al. (2015); March et al. (2007); Mirzaei, Mirzaeinejad (2017);
Mirzaeinejad et al. (2018); Rahimi, Naraghi (2018); Shao et al. (2007); Vivas-Lopez et al. (2015); Wang et al. (2009); Xie et al. (2018)

Ricco et al. (2020) (note that the last reference does not include ICC). (2021), Khosravani et al. (2018), Kirli et al. (2019), Wu et al. (2020),
In any case, when the vehicle is subject to external disturbances, i.e., and Xiang et al. (2020). In automated vehicle applications, Funke
crosswind, or is operating in non-nominal conditions, i.e., changing tyre- et al. (2016) and Leung et al. (2020) define a variable time step along
road friction, or is involved in extreme transients, i.e., an obstacle the prediction horizon to enhance the vehicle stabilisation and
avoidance, FB control is necessary. The literature shows a variety of FB collision avoidance performance, while reducing the computational
implementations, including: cost of implicit MPC; or ii) the explicit approach, in which the
optimisation problem is solved offline for a pre-defined set of states
• Automotive (Aut) and rule-based (RB) controllers, which include and parameters, and the online implementation of the controller
typical ABS controllers, based on complex sets of rules accounting for reduces to a function evaluation. The drawback is represented by the
wheel slip and acceleration (Singh et al., 2013), as well as suspension increased online memory requirements for storing the explicit solu­
controllers using skyhook (Lou et al., 2010; Soltani et al., 2018), tion. Relevant examples of explicit MPC are the hybrid imple­
groundhook (Valášek et al., 2004), or their combination (Vivas-Lo­ mentations in Bernardini et al. (2009), and Di Cairano et al. (2013),
pez et al., 2015). including a piecewise linear approximation of the lateral tyre force
• Nonlinear backstepping control (BStep). characteristics, as well as those in Metzler et al. (2019, 2020), which,
• H∞ controllers, providing formal guarantee of system robustness, although not including ICC, use explicit MPC for vehicle stability
although the performance in nominal conditions tends to be control, and analyse the effect of prediction model fidelity on the
conservative. controller performance. Recent MPC studies also combine i) and ii),
• Linear parameter varying (LPV) controllers, capable of adapting to e.g., see Zheng and Shyrokau (2019), which adopts the explicit so­
varying operating conditions. lutions as initial guesses for the online optimisation.
• Linear quadratic (LQ) controllers, including implementations with • Proportional integral derivative (PID) controllers.
preview for automated driving, e.g., see Chatzikomis et al. (2018), • Sliding mode controllers (SMCs), which improve tracking perfor­
and Song et al. (2015). mance in presence of modelling errors and disturbances. A common
• Model predictive controllers, which are suitable for the imple­ flaw of SMC is chattering, which can be prevented through approx­
mentation of the high level control layer (e.g., generating the refer­ imate first order formulations (Selby, 2003; Yoon et al., 2008), or
ence yaw moment of a direct yaw moment controller), the CA layer, advanced anti-chattering formulations, such as second order SMC
or control structures integrating multiple layers. MPC can be (Liang et al., 2020), integral SMC (Saikia and Pathak, 2019), integral
implemented either through: i) the implicit approach, which solves terminal SMC (Mousavinejad et al., 2017), and non-singular fast
the optimisation problem online, i.e., on the control hardware terminal SMC (Xia et al., 2020; Zhang and Li, 2019; Zhao et al.,
installed on the vehicle. The real-time execution tends to be 2019).
computationally demanding, and poses limitations to the complexity • Intelligent controllers, e,g., fuzzy logic (FL) and artificial neural
of the prediction model that is included in the MPC formulation, even network (ANN) based controllers, which are considered separately
if the introduction of efficient algorithms for the real-time solution of from conventional FB controllers in Table 5.
optimal control problems and the progressive increase of the
computational power of automotive controllers are reducing the Ding and Taheri (2010), and Warth et al. (2020) highlight that tyre
challenge, e.g., see Guo et al. (2017) and Zhu et al. (2019). Implicit wear or replacement, or other vehicle parameter variations that influ­
MPC is very frequent in the recent literature on multi-objective ICC ence cornering stiffness, can lead to inaccurate controllers; in this
and path tracking control. Relevant examples of implicit MPC respect, Ding and Taheri (2010), Fu et al. (2017), Hang and Chen
implementations for vehicle dynamics control are in Chowdhri et al. (2019), and Xiujian et al. (2009) include adaptive or robust control

193
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Fig. 14. Percentage of considered ICC studies showing simulation-only or experimental vehicle validation results, for the ICC architectures in Section 2.

formulations with respect to tyre uncertainties. the desired – typically straight – trajectory.
Since the first generation of ICC systems, the ABS has been coordi­
5. Evaluation of integrated chassis control systems nated with controlled suspension systems to concurrently improve
braking performance and comfort. In the PeC architectures by Alleyne
Fig. 14 shows that, among the considered ICC studies, the percentage (1997), and Ting and Lin (2004), suspension actuation allows the ABS to
of cases providing some form of experimental vehicle validation is very reduce the braking distance, e.g., in the latter study by 5 m in straight
limited with respect to the simulation based verifications, across all line braking from 27 m/s (Fig. 15(a)), by providing appropriate vertical
considered ICC architectures. Such limited experimental data set is tyre load distribution profiles among the vehicle corners. With the same
likely to be caused by the lack of availability and complexity of the purpose, Hamersma and Els (2014) analyses the effect of semi-active
required demonstrator vehicles, as well as by the fact that in some cases suspension characteristics on ABS braking on rough terrains. Reul and
the real-time implementation of the proposed ICC algorithms can be Winner (2009) shows that the integration of active or semi-active sus­
rather difficult. Based on results from the literature, this section provides pension control to reduce wheel load oscillations can decrease wheel slip
guidelines to the performance analysis of ICC systems, according to the oscillations during ABS activations.
controlled vehicle behaviours, i.e., longitudinal dynamics, lateral dy­ In ICC evaluations, vehicle stability during split-μ braking is of
namics, and attitude control, defined in Section 3, with a further sub­ particular interest, see Trachtler (2004), and Wang et al. (2009); the
section on energy efficiency aspects, which are becoming of increasing target is to decrease the stopping distance and enhance lateral stability
interest. As most of the available ICC examples deal with the lateral with respect to the ABS operating on its own, e.g., by using AFS (Mir­
dynamics, this aspect is the most extensively covered one. Table 6 re­ zaeinejad et al., 2016; Trachtler, 2004) or 4WS (Reinold and Traechtler,
ports typical manoeuvres adopted for the assessment of the ICC systems 2013; Plochl and Lugner, 1996). During a split-μ braking test with zero
in the considered literature, with indication of the respective chassis steering input, Plochl and Lugner (1996) evaluates the resulting per­
control aspects of interest. formance through the lateral deviation of the centre of gravity trajectory
from the straight path, and the magnitude of the resulting ψ , ψ̇ , and β.
Fig. 15(b) compares the trajectories of: i) the passive vehicle, which
5.1. Longitudinal dynamics becomes unstable and spins; ii) the vehicle with ABS only, which re­
mains stable but is subject to significant lateral deviation from the
A significant number of ICC implementations from the literature straight line; and iii) the vehicle with ABS and 4WS, which marginally
implies forms of wheel slip control in traction and braking. The per­ reduces the braking distance and is subject only to a minor lateral de­
formance of wheel slip control systems can be assessed through the key viation from the desired trajectory. The experiments carried out by
performance indicators (KPIs) outlined in the recent survey on ABS Bosch with test vehicles (Reinold and Traechtler, 2013; Trachtler, 2004)
control in Pretagostini et al. (2020), which recommends to consider: i) confirm that the application of AFS and 4WS reduces stopping distance
the braking distance; ii) the mean deceleration; iii) the ABS efficiency, i. and driver steering interventions to follow the reference trajectory. The
e., the ratio of the mean longitudinal acceleration to its theoretical KPIs for split-μ braking are also appropriate for evaluating the split-μ
maximum value according to the available tyre-road friction coefficient; acceleration performance with relevant ICC configurations.
iv) the integral of the time-weighted average of the longitudinal jerk (in In the ICC implementations where the powertrains are used as
absolute value), to evaluate the level of smoothness of the wheel slip chassis actuators, the KPIs must be evaluated for different initial states of
control action during ABS cycling; v) the normalised error between the charge of the battery and powertrain temperatures, which can corre­
minimum wheel speed reached during the first ABS cycle and the cor­ spond to different torque limits. Recent ICC studies, e.g., Batra et al.
responding wheel speed in pure rolling conditions; vi) the time integral (2018), include integration of anti-jerk control and wheel slip control
of the absolute value of the wheel torque rate, which provides an indi­ through electric powertrains and friction brakes, and use typical driv­
cation of the level of induced actuator wear; and vii) the time integral of ability performance indicators, e.g., the integral of the absolute value of
the pitch angle. Moreover, for wheel slip control scenarios including jerk, for the assessment of the drivability aspects (Scamarcio et al., 2020;
swift tyre-road friction coefficient variations, additional recommended Pretagostini et al., 2020).
indicators are the recovery time, which quantifies how long it takes for
the controller to recover from the friction coefficient variation and go
back its steady-state operation; and the maximum absolute value of yaw 5.2. Lateral dynamics
rate induced by the friction variation, which should not exceed 1-1.5
deg/s during straight line braking. For the evaluation of split-μ According to the experience of the authors in vehicle testing, the
braking scenarios, i.e., with different tyre-road friction conditions on the performance of ICC systems targeting handling and lateral stability
left and right sides of the vehicle, a further relevant indicator is the enhancement should be evaluated according to the typical KPIs of the
magnitude of the corrective steering wheel angle to keep the vehicle on cornering response, which can be divided into four categories, namely:

194
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Table 6
Examples of manoeuvres used in the considered literature for the assessment of ICC systems.
Manoeuvre and reference ICC Reference examples Long. Lat. Att.
standard dynamics dynamics control

ABS braking on split-μ surface ( Baslamisli et al. (2011); Kawakami et al. (1992); Mastinu et al. (1994); Mirzaeinejad et al. (2016); X X
ISO 14512) Plochl, Lugner (1996); Vivas-Lopez et al. (2015)
Accelerating in turn Ahangarnejad (2018); Cho et al. (2011); Feng et al. (2020); Hirano et al. (1993); Joa et al. (2016); X X X
Xia et al. (2019)
Braking in turn (ISO 7975) Chen et al. (2012); Feng et al. (2020); Ricciardi et al. (2019); Xia et al. (2019) X X X
Circuit Feng et al. (2020); Her et al. (2016); Her, Koh et al. (2015); Joa, Park, et al. (2018); Kissai et al. X X X
(2018)
Double lane change (ISO Ding, Taheri (2010); Fergani et al. (2017); Joa et al. (2016); Li, Arat (2016); Rahimi, Naraghi X X
3888-1) (2018); Švec et al. (2019); Wu et al. (2020)
Fishhook Ahangarnejad (2018); Baslamisli et al. (2011); Tchamna et al. (2014); Vivas-Lopez et al. (2015) X
Obstacle avoidance (ISO Chang, Gordon (2008); Chatzikomis et al. (2018); Falcone et al. (2008); Hajiloo et al. (2020; X X
3888-2) Tchamna et al. (2014); Wahid et al. (2017)
J-turn manoeuvre Zhao et al. (2017) X
Sine-with-dwell (ISO 19365) Li, Arat (2016); Schiebahn et al. (2010); Yim (2015); Yoon et al. (2008) X
Sine sweep (ISO 7401) Ricciardi et al. (2019); Warth et al. (2020) X X
Single lane change Chang, Gordon (2008); Chen et al. (2019); Cho et al. (2011); Chowdhri et al. (2021); Hang et al. X
(2019); Sun et al. (2019); Zhao et al. (2017)
Sinusoidal steer (ISO 7401) Joa, Park, et al. (2018) X
Steady-state increasing steering Ahangarnejad (2018); Ahangarnejad et al. (2019); Chen et al. (2019); Hang et al. (2019); Li, Arat X
(ISO 19365) (2016); Warth et al. (2020); Tang, Khajepour (2020)
Step steer (ISO 7401) Ahangarnejad (2018); Ahangarnejad et al. (2019); Hou et al. (2008); Shen, Yu (2007); Sun et al. X
(2019); Xia et al. (2019); Yoon et al. (2008)
Straight line ABS braking (ISO Alleyne (1997); Reul, Winner (2009); Ricciardi et al. (2019); Vivas-Lopez et al. (2015); Xia et al. X X
21994) (2019)
Straight line swift acceleration Ahangarnejad (2018); Mitamura et al. (1988); Xia et al. (2019) X X
Driving on rough terrain (ISO Hamersma, Els (2014); Savitski et al. (2015); Vivas-Lopez et al. (2015); Xia et al. (2019) X X
2631-1)

a) agility, covering the cornering performance characteristics in the considered actuator, the diagram includes typical resulting vehicle
frequency domain; b) linearity, analysing the extent of the linearity of response regions, namely the loci of the relevant handling points ac­
vehicle response; c) sportiness, evaluating the reactiveness of the vehicle cording to the available literature and the experience of the authors. The
cornering response; and d) stability, analysing the sideslip and yaw rate main highlights are:
magnitude in proximity of the resonance frequency and the cornering
limit. The performance of ICC strategies should be analysed via: i) quasi- • Direct yaw moment control is effective in shaping the understeer
steady-state manoeuvres covering the entire lateral acceleration range, characteristic across the entire lateral acceleration range, and can be
e.g., along ramp steers at constant speed, or skid pads; and ii) ma­ considered the most effective form of chassis control for lateral dy­
noeuvres exciting significant transients and/or nonlinearities, e.g., sine namics enhancement. However, in practical implementations, the
sweep, step steer, sine-with-dwell, fishhook test, obstacle avoidance, variation of the steady-steady cornering response can be achieved
power-on and power-off while cornering, see also the examples in only through continuous actuation, i.e., through TV control. In fact,
Table 6. the interventions of the stability controller based on the friction
Fig. 16(a) shows a selection of KPIs used by Porsche (Warth et al., brakes, which is the DYC solution usually implemented on produc­
2020) during a slow ramp steer in high tyre-road friction conditions. The tion passenger cars, tend to cause vehicle speed reductions and
figure reports the understeer characteristic (dynamic steering angle as a additional energy dissipation, and therefore should be limited to
function of lateral acceleration, upper plot) for the evaluation of limit handling conditions.
steady-state agility, and the sideslip angle characteristic (dynamic • 4WS can shape the understeer characteristic throughout the lateral
sideslip angle as a function of lateral acceleration, lower plot) for the acceleration range, even if its effectiveness is significant only for low-
evaluation of steady-state stability. The KPIs are represented by: a) the to-medium lateral accelerations, rather than at the cornering limit.
gradients of the linear parts of the graphs, gδlin = ∂δ/∂ay and gβlin = ∂β / Moreover, the combination of DYC and 4WS enables simultaneous
∂ay (for simplicity of notations, the same symbols are used here for the and independent shaping of the understeer and sideslip angle char­
angles and their dynamic values, i.e., excluding the kinematic contri­ acteristics (see also Fig. 16), within the physical limits related to the
butions), e.g., considered at 0.4 g of lateral acceleration; b) the lateral load transfers and tyre-road friction conditions.
maximum lateral acceleration, ay,lin , at which the response of the vehicle • ARC or ASS can vary the front-to-total anti-roll moment distribution,
is close to linearity, which should be relatively large, even if some see Lee (2002), Li et al. (2008), Rahimi and Naraghi (2018), and
nonlinearity at high ay is desirable to ‘warn’ the human driver about the Velardocchia and Vigliani (2013), which has an effect on the lateral
approaching cornering limit; c) the gradients, gδ85% and gβ85% , at spe­ load transfer distribution, and on the lateral force and cornering
cific percentages (e.g., 85%) of the maximum lateral acceleration; d) the stiffness of each axle (Ricco et al., 2020). In fact, an increase of the
maximum absolute values of lateral acceleration and sideslip angle, front anti-roll moment, which is equivalent to a decrease of the rear
ay,max and βmax , which should be respectively large and small; and e) the anti-roll moment, tends to increase understeer; vice versa, a decrease
snap oversteer coefficient, gβSO = gβ85% /gβlin , which must be small, i.e., of the front anti-roll moment and/or an increase of the rear anti-roll
close to 1, to facilitate vehicle control in proximity of the limit of moment tend to reduce understeer. However, this effect is evident
handling. only for medium-to-high lateral load transfers, i.e., typically for
Fig. 17 reports the qualitative effect of different chassis actuation |ay | > 0.5g according to Schiebahn et al. (2010), which also shows
systems on the understeer characteristic. The dark dashed line repre­ that the achievable variation of the total yaw moment through ARC
sents the understeer characteristic of the passive vehicle, while for each and ASS is significantly smaller than through TV (Fig. 13).

195
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Fig. 15. Examples of enhancement of the braking performance in split-μ conditions through ICC, shown by: (a) Alleyne (1997), and (b) Plochl and Lugner (1996).

Fig. 16. Example of KPIs that can be obtained from (a) a ramp steer manoeuvre, and (b) a sine sweep steering manoeuvre, from Warth et al. (2020).

Fig. 17. Qualitative diagram of the influence of individual actuation methods on the handling performance, based on the authors’ knowledge and Ahangarnejad
(2018), Ricco et al. (2020), and Warth et al. (2018).

• AAS modifies the front-to-total vertical axle force distribution understeer characteristic expressed as a function of the steering wheel
(Ahangarnejad, 2018; Ahangarnejad et al., 2019), and thus can be angle, rather than the steering angle at the wheel, i.e., AFS modifies the
used to shape the understeer characteristic. However, the AAS effect understeer characteristic perceived by the human driver without
varies with vehicle speed, as the aerodynamic forces and moments changing the fundamental cornering response of the passive vehicle, and
depend on the square of speed, which implies that this actuation therefore is beneficial only in human driven vehicles. The study in
method: i) is effective only at very high speeds; ii) cannot generate a Shimada and Shibahata (1994) assesses the direct yaw moment control,
consistent vehicle response at different speeds; and iii) is suitable front-to-total anti-roll moment distribution control and 4WS capabilities
only for very high performance cars, e.g., designed for being driven of compensating the variation of cornering response induced by the
not only on public roads with relatively low speed limits, but on race longitudinal vehicle acceleration, as a function of |β|. The conclusions
tracks as well. confirm the analysis of Fig. 17, i.e., direct yaw moment control can
compensate the longitudinal acceleration effect for the whole consid­
AFS is not included in the diagram, as its effect is limited to the ered range of |β|, roll moment distribution control is effective only at

196
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

large |β|, while 4WS is effective only at small-to-medium |β|. 2019), and especially the overshoot of the yaw rate and sideslip angle
Another potentially relevant chassis actuation technology (not response, which can be critical during extreme transients (Ahangarne­
included in Fig. 17) that could be integrated in future ICC systems for the jad, 2018; Li and Arat, 2016; Yim, 2015).
enhancement of the understeer characteristic is represented by variable Figure 18(a) is a qualitative example of spider chart, according to the
geometry suspension systems, i.e., based on the active control of camber, current industrial vehicle assessment practices, based on 4WS results by
toe and/or caster angles, which influence rolling resistance and tyre slip Porsche (Warth et al., 2018) and Fiat Chrysler Automobiles (Pascali
power losses in straight line, as well as directional stability and steering et al., 2003), which are compared with those of the passive car, used as
wheel feeling in cornering conditions (Kavitha et al., 2019; Tranquillo baseline (its indicators represent the 100% level in the chart), and the
et al., 2020). Examples of active camber system (ACS) devices have been same vehicle with an ICC system including 4WS, AAS and TV (the ICC
recently involved in patent applications, see Seo and Park (2020), and values are based on the experience of the authors of this survey). The
Bakker (2020). The dual axis steering (DAS) system developed for F1 greater is the amplitude, the better is the performance related to the
applications by Mercedes, which allows the driver to manage the toe respective KPI.
angle by moving the steering wheel along its rotation axis, permits to The handling and lateral stability aspects of ICC for automated ve­
manually switch the front suspension between two setups (Tranquillo hicles are evaluated along manoeuvres with reference paths requiring a
et al., 2020): i) a zero-toe position that reduces tyre heating in straight combination of good agility and stability characteristics, like the
line; and ii) a toe-out position for desirable cornering response. Kavitha obstacle avoidance, see Table 6. As in these applications the main ICC
et al. (2018, 2019), and Vo et al. (2018) are examples of studies dealing function is to track the reference path, the most meaningful indicators
with active suspension kinematics with double wishbone suspensions, are the lateral displacement and heading angle path tracking errors,
for enhanced vehicle dynamics performance. Gáspár and Németh (2016, ΔyCG and Δψ CG , see Guo et al. (2018), Hang and Chen (2019), and Peng
2017), and Shyrokau et al. (2015) integrate ACS with TV and AFS for et al. (2020), and Figure 18(b). Chatzikomis et al. (2018) also considers
vehicle stability and trajectory tracking. The variation of camber angle the maximum initial speed to complete the manoeuvre, vin,max , and the
modifies tyre forces as well as the scrub radius. According to Gáspár and vehicle speed at the exit of the obstacle avoidance course, vfin , that must
Németh (2016, 2017), variable geometry suspension systems can be be close to vin,max .
beneficial also in the context of driver assistance systems. As a summary of the previous discussions, Table 7 reports a selection
Fig. 16(b) shows a generic frequency response characteristic of a of the KPIs of the cornering response, together with an overview of the
vehicle subject to a sine sweep steering test, with increasing frequency effectiveness level of the most common chassis actuators.
and constant amplitude of the steering wheel input, executed at
approximately constant speed. Typical indicators are: a) the yaw dy­
⃒ ⃒
namic amplification, Aψ̇ = max⃒Gψ̇ ⃒/G0 , i.e., the ratio of the maximum 5.3. Attitude control
magnitude of the yaw rate response to its steady-state value, which
should be as small as possible; b) the vehicle state eigenfrequencies, fGn As the ICC implementations from the literature focus on the lateral
= f(max|Gi˙|), with n = ψ̇ , β, namely the frequencies corresponding to vehicle dynamics, the most relevant performance indicators of attitude
the peak amplitudes of the yaw rate and sideslip angle response; c) fGay = control deal with the roll response of the vehicle body, e.g., in terms of
roll gradient, which is the steady-state roll angle magnitude variation
f(0.9 max|Gay |), i.e., the frequency at which the magnitude of the lateral
per unit of lateral acceleration variation (gϕlin = ∂ϕ/∂ay ), and the phase
acceleration frequency response, |Gay |, reaches 90% of its maximum
value of roll angle with respect to a lateral acceleration input, which is
value; d) G1ψ̇ = |ψ̇ /δ|1 Hz and G1β = |β/δ|1 Hz , i.e., the magnitudes of the
an indicator of the sportiness level of the response, and is evaluated
yaw rate and sideslip angle response to the steering input at a specific
through the corresponding time delays, e.g., t0.5ϕay and t1ϕay , computed
frequency value, e.g., 1 Hz; e) G1β = |ay /ψ̇ |1 Hz and G1ay = |ay /δ|1 Hz , i.
at 0.5 Hz and 1 Hz. The rollover prevention performance can be evalu­
e., the magnitudes of the yaw rate to lateral acceleration and steering
ated through the rollover index and the critical rollover velocity, where
angle to lateral acceleration characteristics at the same frequency; and f)
the latter is the maximum initial speed for a given test (e.g., the fishhook
t1ψ̇ , t1β , t1ay ψ̇ , and t1ay , i.e., the time delays of the frequency response
manoeuvre) at which the car is not affected by unstable roll dynamics.
characteristics mentioned in d) and e), computed from the phase angles The attitude control performance should also be assessed in terms of
at the same frequency (1 Hz). The delay values define the level of pitch angle and vehicle heave induced by the tangential vehicle accel­
reactiveness of the car, and need to be carefully specified. For example, a eration (Ricciardi et al., 2019). With respect to ride comfort, which is not
reduction of t1ay ψ̇ is usually desirable, however, this is usually accom­ the focus of the available ICC examples, but is dealt with by a significant
panied by a reduction of t1ψ̇ , which corresponds to fast front-end steering body of literature on suspension control, typical indicators are the root
response, subjectively perceived by typical drivers as aggressive mean square or vibration dose values of the frequency weighted sprung
behaviour. mass accelerations, see the standard ISO 2631-1. In general, ride comfort
Many ICC studies (see Table 6) use extreme transient tests, such as can be directly evaluated through the observation of the frequency
the sine-with-dwell (the reference test for the homologation of stability response characteristics of roll, heave and pitch accelerations, for
control systems, see UN/ECE 140), fishhook, step steer, and obstacle assigned road inputs.
avoidance tests, to assess the performance of the proposed ICC systems. In terms of attitude control in the ICC examples from the considered
The results are mostly evaluated in terms of qualitative observation of literature, Li et al. (2008) controls the roll and yaw motions through
the reference yaw rate tracking performance, as well as the sideslip 4WS, DYC and ARC; the results show the better roll dynamics of the
angle limitation capability (Chatzikomis et al., 2018; Hajiloo et al., integrated system with respect to the standalone actuations during a
2020; Xie et al., 2018). For an objective evaluation of the yaw rate double lane change. The ICC of Gáspár and Németh (2016) applies the
tracking performance, the root mean square value of the yaw rate error emerging technology of ACS to limit the roll motion of the vehicle body
can be used. Moreover, the cornering response of a vehicle subject to a by controlling the camber angles of the individual front wheels and the
steering input tends to be similar to that of an underdamped second roll centre heights. Yoon et al. (2008) increases the critical rollover
order system, where the damping ratio decreases with vehicle speed, see speed in comparison with the passive vehicle during a simulated severe
the discussion of the transient response of the single track vehicle model turn, through the combination of VSC and continuous damping control
in Milliken and Milliken (1995). Therefore, the transient response of a (CDC) in a SuC architecture. Tchamna et al. (2014) designs a PeC ar­
vehicle subject to fast steering inputs can be evaluated through the chitecture for the coordination of VSC and ASS, where the latter reduces
typical indicators for assessing the step response of second order sys­ the roll angle, roll rate, pitch angle, heave displacement, rollover index
tems, such as the rise time, the settling time (Hou et al., 2008; Xia et al., as well as lateral load transfers during fishhook and obstacle avoidance

197
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Fig. 18. Examples of: (a) qualitative spider chart showing the comparison between a passive car, its 4WS version, and its version with ICC, based on Pascali et al.
(2003), and Warth et al. (2018), and the target performance of ICC; and (b) path tracking KPIs, adapted from Chatzikomis et al. (2018), where the dashed trajectory is
the reference path, and the red line is the actual vehicle path.

Table 7
Qualitative influence of different chassis actuation systems on a selection of lateral dynamics and attitude control performance indicators.

tests. The reduction of the vehicle body motion brought by the ASS two MuC implementations, a first one consisting of two control layers
improves the yaw rate tracking and sideslip angle limitation perfor­ and a second one with three layers, shows that the three-layer case tends
mance of the VSC. Similar analyses in terms of roll dynamics and lateral to reduce the peak and average values of the sprung mass accelerations
stability improvements through semi-active suspensions and DYC are during simulated J-turn and single lance change manoeuvres, in addi­
presented in Soltani et al. (2018). Zhao et al. (2017) develops an ICC tion to improving the lateral dynamics. A third PeC implementation is
system for a vehicle with AFS, DYC and ASS; the comparison between also presented, which, however, “results in an unsatisfactory performance

198
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

in terms of the vehicle dynamics.” the development direction indicated in Shibahata (2005). Very recent
developments in ICC, e.g., see Tang and Khajepour (2020), also show
5.4. Energy efficiency attention towards distributed optimal control through innovative PeC
arrangements, which could achieve similar performance to upstream
An area of increasing interest is the development of chassis control architectures.
systems targeting, among other objectives, the reduction of energy The growing tendency towards the development and implementa­
consumption, e.g., to make the powertrain/s operate in regions with tion of ICC systems is expected to continue, the main reasons being:
high efficiency, which includes extensions towards the topic of con­
nected and automated vehicles (Montanaro et al., 2018). The literature • The increasing number of actuators implemented in modern pas­
on energy-efficient chassis control mostly deals with TV control imple­ senger cars, normally introduced on the most expensive and
mentations on electric vehicles with multiple powertrains, including performance-oriented models, e.g., see the Porsche Taycan (Tracy,
consideration of powertrain power losses and/or longitudinal and 2019). In the last 15 years, car makers and their suppliers, e.g.,
lateral tyre slip power losses, see the analyses in De Filippis et al. (2018), General Motors (Ghoneim, 2006; Lee and Litkouhi, 2013), Bosch
and De Novellis et al. (2013, 2014), targeting power loss or battery (Lohner et al., 2007), and Delphi (Bedner et al., 2004), have been
power minimisation in straight line and cornering conditions. In the patenting several supervisory state machines to coordinate power­
previous analyses, the power consumption and/or power losses are trains, braking and steering systems, and active aerodynamics.
mostly evaluated in quasi-steady-state cornering, as functions of lateral • The progressive electrification of passenger car powertrains, which
acceleration, sometimes also including non-zero longitudinal accelera­ can be characterised by rather complex arrangements, including
tion conditions, while the MPC implementation in Parra et al. (2020) multiple electric machines. For instance, the latest generation Honda
shows benefits during transient manoeuvres as well. NSX (Toyoshima, 2016) is equipped with a hybrid powertrain on the
In real electric vehicles, the power consumption of the powertrains is rear axle and an electric powertrain for each front wheel, which al­
estimated from the measurement of the appropriate voltage/s and cur­ lows continuous front-to-total wheel torque distribution and direct
rent/s. Consideration of power consumption and energy efficiency is still yaw moment control. In general, many recent production electric
rather limited in the ICC literature, with a few exceptions. Table 4 shows vehicles (Shao et al., 2020) have AWD layouts with one powertrain
examples of energy-efficiency-related formulations that are minimised per axle with high torque capability, in some cases with different
in the CA cost functions, many of them also proposed as KPIs. Joa, Park motors on each axle, see the Audi e-tron and Tesla Model S, which –
et al. (2018) suggest consideration of the energy efficiency aspects together with the other available chassis actuators, e.g., for braking
through the longitudinal and lateral tyre slip power losses. Edrén et al. and suspension control – offers significant ICC potential.
(2019) and Jalaliyazdi (2016) evaluate the electric powertrain power • The interest towards the integration of vehicle dynamics control and
loss (see Table 4) together with the mechanical power of the powertrains energy management, e.g., through the development of advanced
to obtain the energy consumption (also used in the allocation problem in model based controllers, combining aspects such as powertrain
Edrén et al. (2019)) or energy efficiency (Jalaliyazdi, 2016) through the power losses, actuation power losses, and tyre slip power losses, with
respective models. The studies in Shyrokau et al. (2013, 2015) discuss a the more conventional yaw rate tracking, sideslip angle limitation,
parameter related to the lateral tyre slip power losses, see the ‘tyre en­ wheel slip control, roll-over prevention, and ride comfort improve­
ergy efficiency during Fy generation’ entry in Table 4. Most of the ment, even if the most significant preliminary implementations
considered ICC literature does not include detailed analyses of energy (Ataei et al., 2020; Parra et al., 2020) refer to direct yaw moment
recuperation in braking through the electric powertrains, which is a control only, rather than ICC.
major point to be further investigated, as mentioned in Wen et al. • The significant efforts towards driving automation (see the auto­
(2018). An example of regeneration-related KPI is the power recupera­ mation levels defined in the standard SAE J3016), which requires the
tion index of the electric motors, adopted by Shyrokau et al. (2013, implementation of path tracking controllers with high level of
2015) and defined in Table 4. adaptability to the varying operating conditions and environmental
In terms of energy efficiency results, the allocation of the TV and the scenarios. Figure 19(b) shows the distribution of the ICC architec­
RWS control actions in Edrén et al. (2019) accomplishes an energy tures adopted in automated vehicles, and confirms that in such ap­
consumption improvement of 6–8% in typical handling manoeuvres plications ICC requires the adoption of upstream coordination. High
with closed-loop path tracking. In Jalaliyazdi (2016) the optimal torque levels of driving automation could bring new vehicle dynamics
distribution unit achieves an overall powertrain efficiency increase by control paradigms, e.g., with the integration of path tracking and
~2.3%. In electric vehicles, the energy consumed by the electric pow­ direct yaw moment control, and the relaxation of the sideslip con­
ertrains is in general much higher than the total energy demand of the straints typical of vehicle stability control, with the purpose of pri­
remaining subsystems. For this reason, the energy consumption of the oritising reference trajectory tracking in emergency conditions
chassis actuators is usually neglected. A rare exception is represented by (Funke et al., 2015). Moreover, one of the long-term aims of driving
the study in Shyrokau et al. (2015), which calculates the energy con­ automation is to allow the vehicle users to carry out other activities
sumption of each chassis actuator by considering its inertia, the resis­ during the trips (Gwak et al., 2019), which is not possible in current
tance induced by the road-related loads, and the internal actuator power cars, because of the insufficient ride comfort level. This long-term
losses. Along a sine-with-dwell test, the results show that the electric objective is likely to bring an increasing effort towards the
power consumption is ~0.13 kJ for the active camber actuators, improvement of the comfort and motion sickness (Elbanhawi et al.,
~0.15 kJ for the wheel steering actuators, ~0.61 kJ for of the active 2015) aspects, through the next generation of suspension actuation
suspension actuators, and ~0.52 kJ for the friction brake actuators, systems (e.g., see Sun et al., 2019) as well as path planning and
while the electric motors consume a total of ~13 kJ. tracking controllers (Gallep and Muller, 2018).
• Personalisation as an emerging trend for next generation vehicles, i.
6. Trends and future developments e., future ICC systems should be able to modify their behaviour in
accordance with the requirements of the specific vehicle occupants,
Fig. 19(a) provides an overview of the distribution throughout the which will require the development of appropriate user-oriented
years of the ICC implementations proposed in the references of this supervisory strategies (BBC Top Gear Magazine, 2019; König et al.,
literature survey among the ICC architectures defined in Section 2. The 2014).
trend shows an increasing number of upstream architectures, and in • The progressive increase of the level of information sharing within
particular of SuC and MuC implementations, which is consistent with the same vehicle from systems from multiple domains, for example

199
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Fig. 19. ICC architecture distribution along the years for: (a) the whole considered set of ICC references; and (b) the considered ICC references dealing with
automated vehicle applications.

radars, cameras, lidars and navigation system, as well as information • Ongoing and expected future developments include: i) systematic
sharing from other vehicles, road users and infrastructure (Mon­ consideration of electric powertrain layouts with multiple motors as
tanaro et al., 2018), together with the trend towards predictive chassis actuators; ii) implementation of ICC systems integrating the
control (Guo et al., 2017; Hajiloo et al., 2020; Kirli et al., 2019; Lin energy management and vehicle dynamics control functions; iii) a
et al., 2019; Xiang et al., 2020). The ongoing progress in these areas new generation of vehicle dynamics control systems specifically
is likely to bring a new generation of vehicle controllers, capable of designed for automated driving applications, benefitting from sensor
pre-emptive interventions, e.g., as a function of the expected path or fusion, and including pre-emptive and predictive components in
road profile ahead, see Gao and Gordon (2019), Wang et al. (2020), their control structures; iv) increased attention to vehicle comfort
and Wu et al. (2020). and motion sickness, in the context of ICC for automated vehicles;
• The continued effort of the automotive industry to limit complexity, and v) increased personalisation capability of the ICC systems, to
computational requirements and costs, while encouraging modu­ meet varying user requirements with the same vehicle.
larity and flexibility of the electric and electronic architectures
(Askaripoor et al., 2020; Eder et al., 2020; Horst et al., 2014; Som­
mer et al., 2013). Declaration of Competing Interest

7. Conclusions The authors declare that they have no known competing financial
interests or personal relationshipsthat could have appeared to influence
Integrated chassis control solutions are becoming widely adopted on the work reported in this paper.
production cars because of the increasing number of available chassis
actuators, including complex electrified powertrain architectures, References
brake-by-wire, semi-active and active suspension systems, as well as
Abe, M. (1992). Roll moment distribution control in active suspension for improvement
active front steering and 4-wheel-steering. This paper reviewed the ac­ of limit performance of vehicle handling. In Proceedings of the International
ademic and industrial literature on ICC, and included: i) the classifica­ Symposium on Advanced Vehicle Control.
tion and critical analysis of ICC architectures according to the position of Abe, M., & Mokhiamar, O. (2007). An integration of vehicle motion controls for full
drive-by-wire vehicle. Proceedings of the Institution of Mechanical Engineers, Part K:
the coordination layer; ii) the presentation of the reference variables Journal of Multi-Body Dynamics, 221(1), 116–127.
adopted by the different controllers within ICC systems; iii) the classi­ Ahangarnejad, A. H. (2018). Integrated control of active vehicle chassis control systems.
fication and detailed discussion of coordination strategies and control Politecnico di Milano.
Ahangarnejad, A. H., Melzi, S., & Ahmadian, M. (2019). Integrated vehicle dynamics
structures; and iv) an overview of typical manoeuvres and key perfor­
system through coordinating active aerodynamics control, active rear steering,
mance indicators for ICC assessment. torque vectoring and hydraulically interconnected suspension. International Journal
The main conclusions are: of Automotive Technology, 20(5), 903–915.
Alberding, M. B., Tjønnås, J., & Johansen, T. A. (2014). Integration of vehicle yaw
stabilisation and rollover prevention through nonlinear hierarchical control
• The recent papers, some of them dealing with automated vehicle allocation. Vehicle System Dynamics, 52(12), 1607–1621.
applications, show a clear shift towards multi-objective ICC imple­ Alleyne, A. (1997). Improved vehicle performance using combined suspension and
mented through upstream architectures. In particular, multi-layer braking forces. Vehicle System Dynamics, 27(4), 235–265.
Ando, N., & Fujimoto, H. (2010). Yaw-rate control for electric vehicle with active front/
coordination is earning increasing attention due to its modularity rear steering and driving/braking force distribution of rear wheels. In Proceedings of
and adaptability, on top of the intuitive separation and management the 11th IEEE International Workshop on Advanced Motion Control (AMC).
of tasks. Andreasson, J., & Bunte, T. (2006). Global chassis control based on inverse vehicle
dynamics models. Vehicle System Dynamics, 44(sup1), 321–328.
• On-line model-based optimisation techniques, either in the form of Askaripoor, H., Farzaneh, M. H., & Knoll, A. (2020). Considering safety requirements in
control allocation algorithms or model predictive controllers, are design phase of future E/E architectures. In Proceedings of the 25th IEEE International
becoming very frequent in the recent literature, given the compu­ Conference on Emerging Technologies and Factory Automation (ETFA).
Ataei, M., Khajepour, A., & Jeon, S. (2020). Model predictive control for integrated
tational power increase of modern automotive controllers, and are lateral stability, traction/braking control, and rollover prevention of electric
progressively substituting rule-based and off-line optimisation-based vehicles. Vehicle System Dynamics, 58(1), 49–73.
implementations. Bakker, C. (2020). Suspension with active damping to tune caster dynamics (Patent No. US
10,773,566 B2).
• In terms of ICC results and performance assessment, the literature
Bang, M. S., Lee, S. H., Han, C. S., MacIuca, D. B., & Hedrick, J. K. (2001). Performance
shows that: i) most of the ICC implementations target the enhance­ enhancement of a sliding mode wheel slip controller by the yaw moment control.
ment of the lateral vehicle dynamics through improved utilisation of Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile
the tyre-road friction potential; and ii) the available ICC results are Engineering, 215(4), 455–468.
Baslamisli, S. C., Kose, I. E., & Anlac, G. (2011). Handling stability improvement through
mostly based on simulations, i.e., there is a substantial lack of robust active front steering and active differential control. Vehicle System Dynamics,
experimental demonstrations on real vehicles. 49(5), 657–683.

200
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Batra, M., Maitland, A., McPhee, J., & Azad, N. L. (2018). Non-linear model predictive De Novellis, L., Sorniotti, A., & Gruber, P. (2014). Wheel torque distribution criteria for
anti-jerk cruise control for electric vehicles with slip-based constraints. In Proceedings electric vehicles with torque-vectoring differentials. IEEE Transactions on Vehicular
of the American Control Conference. Technology, 63(4), 1593–1602.
BBC Top Gear Magazine. (2019). Ten things you need to know about the Ferrari SF90 De Novellis, L., Sorniotti, A., & Gruber, P. (2015). Driving modes for designing the
Stradale. https://www.topgear.com/car-news/supercars/ten-th cornering response of fully electric vehicles with multiple motors. Mechanical
ings-you-need-know-about-ferrari-sf90-stradale#7. Systems and Signal Processing, 64–65, 1–15.
Beal, C. E., & Gerdes, J. C. (2012). Model predictive control for vehicle stabilization at De Pinto, S., Chatzikomis, C., Sorniotti, A., & Mantriota, G. (2017). Comparison of
the limits of handling. IEEE Transactions on Control Systems Technology, 21(4), traction controllers for electric vehicles with on-board drivetrains. IEEE Transactions
1258–1269. on Vehicular Technology, 66(8), 6715–6727.
Bedner, E. J., Boguslaw Hac, A., Chen, H. H., Chandy, A., & Check, M. J. (2004). Vehicle Deng, W. (2012). Function decomposition and control architecture for complex vehicle control
chassis control with coordinated brake and steering control on split coefficient surface. system. Patent No. US 8,260,498 B2.
Patent No. US 6,681,167 B2. Di Cairano, S., Tseng, H. E., Bernardini, D., & Bemporad, A. (2013). Vehicle yaw stability
Bedner, E. J., & Chen, H. H. (2004). A supervisory control to manage brakes and four- control by coordinated active front steering and differential braking in the tire
wheel-steer systems. SAE Technical Paper, 2004-01–1059. sideslip angles domain. IEEE Transactions on Control Systems Technology, 21(4),
Bernardini, D., Di Cairano, S., Bemporad, A., & Tseng, H. E. (2009). Drive-by-wire 1236–1248.
vehicle stabilization and yaw regulation: A hybrid model predictive control design. Ding, N., & Taheri, S. (2010). An adaptive integrated algorithm for active front steering
In Proceedings of the IEEE Conference on Decision and Control. and direct yaw moment control based on direct Lyapunov method. Vehicle System
Boada, M. J. L., Boada, B. L., Muñoz, A., & Díaz, V (2006). Integrated control of front- Dynamics, 48(10), 1193–1213.
wheel steering and front braking forces on the basis of fuzzy logic. Proceedings of the Doumiati, M., Sename, O., Dugard, L., Martinez-Molina, J. J., Gaspar, P., & Szabo, Z.
Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 220(3), (2013). Integrated vehicle dynamics control via coordination of active front steering
253–267. and rear braking. European Journal of Control, 19(2), 121–143.
Borrelli, F., Falcone, P., Keviczky, T., Asgari, J., & Hrovat, D. (2005). MPC-based Doumiati, M., Sename, O., Martinez, J., Dugard, L., Gaspar, P., & Szabo, Z. (2011).
approach to active steering for autonomous vehicle systems. International Journal of Vehicle yaw control via coordinated use of steering/braking systems. IFAC
Vehicle Autonomous Systems, 3(2), 265–291. Proceedings Volumes, 44(1), 644–649.
Brennan, S., & Alleyne, A. (2001). Integrated vehicle control via coordinated steering and Drakunov, S., Özguner, U., Dix, P., & Ashrafi, B. (1995). ABS control using optimum
wheel torque inputs. In Proceedings of the 2001 American Control Conference. search via sliding modes. IEEE Transactions on Control Systems Technology, 3(1),
Burgio, G., & Zegelaar, P. (2006). Integrated vehicle control using steering and brakes. 79–85.
International Journal of Control, 79(5), 534–541. Duffie, N. A., Chitturi, R., & Mou, J. I. (1988). Fault-tolerant heterarchical control of
Cao, Z., & Zheng, S. (2019). MR-SAS and electric power steering variable universe fuzzy heterogeneous manufacturing system entities. Journal of Manufacturing Systems, 7(4),
PID integrated control. Neural Computing and Applications, 31(4), 1249–1258. 315–328.
Chandy, A. (2003). System and method incorporating feedforward for motor vehicle chassis Durham, W., Bordignon, K. A., & Beck, R. (2017). Aircraft control allocation. Wiley.
control. Patent No. US 6,567,731 B2. Eder, J., Voss, S., Bayha, A., Ipatiov, A., & Khalil, M. (2020). Hardware architecture
Chang, S., & Gordon, T. J. (2008). A flexible hierarchical model-based control exploration: automatic exploration of distributed automotive hardware
methodology for vehicle active safety systems. Vehicle System Dynamics, 46(sup1), architectures. Software and Systems Modeling, 19(4), 911–934.
63–75. Edrén, J., Jonasson, M., Jerrelind, J., Stensson, A., & Drugge, L. (2019). Energy efficient
Chatrath, K. (2019). Vehicle Dynamics Control Using Control Allocation. Delft University of cornering using over-actuation. Mechatronics, 59, 69–81.
Technology. Elbanhawi, M., Simic, M., & Jazar, R. (2015). In the passenger seat : Investigating ride
Chatrath, K., Zheng, Y., & Shyrokau, B. (2020). Vehicle dynamics control using model comfort measures in autonomous cars. IEEE Intelligent Transportation Systems
predictive control allocation combined with an adaptive parameter estimator. SAE Magazine, 7(3), 4–17.
International Journal of Connected and Automated Vehicles, 3(12-03-02–0009), Elhefnawy, A., Sharaf, A., Ragheb, H., & Hegazy, S. (2017). Design of an integrated yaw-
103–117. roll moment and active front steering controller using fuzzy logic control. SAE
Chatzikomis, C., Sorniotti, A., Gruber, P., Bastin, M., Shah, R. M., & Orlov, Y. (2017). International Journal of Vehicle Dynamics, Stability, and NVH, 1(2017-01-1569),
Torque-vectoring control for an autonomous and driverless electric racing vehicle 270–282.
with multiple motors. SAE International Journal of Vehicle Dynamics, Stability, and Falcone, P., Borrelli, F., Asgari, J., Tseng, H. E., & Hrovat, D. (2007). Predictive active
NVH, 1(2017-01-1597), 338–351. steering control for autonomous vehicle systems. IEEE Transactions on Control
Chatzikomis, C., Sorniotti, A., Gruber, P., Zanchetta, M., Willans, D., & Balcombe, B. Systems Technology, 15(3), 566–580.
(2018). Comparison of path tracking and torque-vectoring controllers for Falcone, P., Eric Tseng, H., Borrelli, F., Asgari, J., & Hrovat, D (2008). MPC-based yaw
autonomous electric vehicles. IEEE Transactions on Intelligent Vehicles, 3(4), 559–570. and lateral stabilisation via active front steering and braking. Vehicle System
Chen, T., Chen, L., Xu, X., Cai, Y., & Sun, X. (2019). Simultaneous path following and Dynamics, 46(sup1), 611–628.
lateral stability control of 4WD-4WS autonomous electric vehicles with actuator Falcone, P., Tseng, H. E., Asgari, J., Borrelli, F., & Hrovat, D. (2007). Integrated braking
saturation. Advances in Engineering Software, 128, 46–54. and steering model predictive control approach in autonomous vehicles. IFAC
Chen, W., Xiao, H., Chu, C., & Zu, J. W. (2012). Hierarchical control of automotive Proceedings Volumes, 40(10), 273–278.
electric power steering system and anti-lock brake system: Theory and experiment. Fan, X., & Zhao, Z. (2019). Vehicle dynamics modeling and electronic stability program/
International Journal of Vehicle Design, 59(1), 23–43. active front steering sliding mode integrated control. Asian Journal of Control, 21(5),
Chen, W., Xiao, H., Liu, L., & Zu, J. W. (2006). Integrated control of automotive electrical 2364–2377.
power steering and active suspension systems based on random sub-optimal control. Feng, J., Chen, S., & Qi, Z. (2020). Coordinated chassis control of 4WD vehicles utilizing
International Journal of Vehicle Design, 42(3–4), 370–391. differential braking, traction distribution and active front steering. IEEE Access, 8,
Cheng, S., Li, L., Liu, C., & Wu, X. (2020). Robust LMI-based H-infinite controller 81055–81068.
integrating AFS and DYC of autonomous vehicles with parametric uncertainties. IEEE Fergani, S., Menhour, L., Sename, O., Dugard, L., & D’Andréa-Novel, B. (2017).
Transactions on Systems, Man, and Cybernetics: Systems,, 1–10. In press. Integrated vehicle control through the coordination of longitudinal/lateral and
Cheng, S., Li, L., Mei, M. M., Nie, Y. L., & Zhao, L. (2019). Multiple-objective adaptive vertical dynamics controllers: Flatness and LPV/H∞-based design. International
cruise control system integrated with DYC. IEEE Transactions on Vehicular Technology, Journal of Robust and Nonlinear Control, 27(18), 4992–5007.
68(5), 4550–4559. Fruechte, R. D., Karmel, A. M., Rillings, J. H., Schilke, N. A., Boustany, N. M., &
Cheon, J. S. (2010). Brake by wire system configuration and functions using front EWB Repa, B. S. (1989). Integrated vehicle control. In Proceedings of the IEEE 39th
(Electric Wedge Brake) and rear EMB (Electro-mechanical brake) actuators. SAE Vehicular Technology Conference.
Technical Paper, 2010-01–1708. Fu, Z., Li, B., & Ning, X. (2017). Model free adaptive optimal tracking controller design
Cho, W., Choi, J., Kim, C., Choi, S., & Yi, K. (2012). Unified chassis control for the for AFS/DYC based integrated chassis control system. In Proceedings of the 2017 9th
improvement of agility, maneuverability, and lateral stability. IEEE Transactions on International Conference on Modelling, Identification and Control (ICMIC).
Vehicular Technology, 61(3), 1008–1020. Funke, J., Brown, M., Erlien, S. M., & Gerdes, J. C. (2015). Prioritizing collision
Cho, W., Heo, H., & Yi, K. (2011). An investigation into unified chassis control for agility, avoidance and vehicle stabilization for autonomous vehicles. In Proceedings of the
maneuverability and lateral stability. In Proceedings of the 14th international IEEE 2015 IEEE Intelligent Vehicles Symposium (IV).
Conference on Intelligent Transportation Systems (ITSC). Funke, J., Brown, M., Erlien, S. M., & Gerdes, J. C. (2016). Collision avoidance and
Cho, W., Yoon, J., Kim, J., Hur, J., & Yi, K. (2008). An investigation into unified chassis stabilization for autonomous vehicles in emergency scenarios. IEEE Transactions on
control scheme for optimised vehicle stability and manoeuvrability. Vehicle System Control Systems Technology, 25(4), 1204–1216.
Dynamics, 46(sup1), 87–105. Gao, Yangyan, & Gordon, T. J. (2019). Optimal control of vehicle dynamics for the
Chokor, A., Talj, R., Doumiati, M., & Charara, A. (2019). A global chassis control system prevention of road departure on curved roads. IEEE Transactions on Vehicular
involving active suspensions. IFAC-PapersOnLine, 52(5), 444–451. Technology, 68(10), 9370–9384.
Chowdhri, N., Ferranti, L., Santafé, F., & Shyrokau, B. (2021). Integrated nonlinear Gao, Yiqi, Lin, T., Borrelli, F., Tseng, E., & Hrovat, D. (2010). Predictive control of
model predictive control for automated driving. Control Engineering Practice, 106, autonomous ground vehicles with obstacle avoidance on slippery road. In
Article 104654. Proceedings of the ASME 2010 Dynamic Systems and Control Conference.
De Filippis, G., Lenzo, B., Sorniotti, A., & Gruber, P. (2018). Energy-efficient torque- Gáspár, P., & Németh, B. (2016). Integrated control design for driver assistance systems
vectoring control of electric vehicles with multiple drivetrains. IEEE Transactions on based on LPV methods. International Journal of Control, 89(12), 2420–2433.
Vehicular Technology, 67(6), 4702–4715. Gáspár, P., Szabó, Z., & Bokor, J. (2009). Active suspension in integrated vehicle control.
De Novellis, L., Sorniotti, A., & Gruber, P. (2013). Optimal wheel torque distribution for a IntechOpen.
four-wheel-drive fully electric vehicle. SAE International Journal of Passenger Cars -
Mechanical Systems, 6(2013-01-0673), 128–136.

201
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Gallep, J., & Muller, S. (2018). Vehicle Guidance Control for Automated Driving ISO 7401. (2011). Road vehicles - Lateral transient response test methods - Open-loop test
considering Motion Sickness. AVEC’18- Proceedings of the 2018 on Audio/Visual methods.
Emotion Challenge and Workshop. ISO 7975. (2006). Passenger cars — Braking in a turn — Open-loop test method.
Ghoneim, Y. A. (2006). Method and apparatus for estimating steering behavior for integrated Ivanov, V., & Savitski, D. (2015). Systematization of integrated motion control of ground
chassis control. Patent No. US 7,099,759 B2. vehicles. IEEE Access, 3, 2080–2099.
Gillespie, T. D. (1992). Fundamentals of vehicle dynamics. Society of Automotive Ivanov, V., Shyrokau, B., Augsburg, K., & Vantsevich, V. (2010). System fusion in off-
Engineers, Inc. road vehicle dynamics control. In Proceedings of the Joint 9th Asia-Pacific ISTVS
Gordon, T. J. (1996). An integrated strategy for the control of a full vehicle active Conference.
suspension system. Vehicle System Dynamics, 25(sup1), 229–242. Jalaliyazdi, M. (2016). Integrated vehicle stability control and power distribution using model
Gordon, T. J., Howell, M., & Brandao, F. (2003). Integrated control methodologies for predictive control. University of Waterloo.
road vehicles. Vehicle System Dynamics, 40(1–3), 157–190. Joa, E., Park, K., Koh, Y., Yi, K., & Kim, K. (2018). A tyre slip-based integrated chassis
Grip, H. F., Imsland, L., Johansen, T. A., Kalkkuhl, J. C., & Suissa, A. (2009). Estimation control of front/rear traction distribution and four-wheel independent brake from
of road inclination and bank angle in automotive vehicles. In Proceedings of the moderate driving to limit handling. Vehicle System Dynamics, 56(4), 579–603.
American Control Conference (pp. 426–432). Joa, E., Yi, K., & Kim, K. (2015). An integrated control of front/rear traction distribution
Guo, H., Liu, F., Xu, F., Chen, H., Cao, D., & Ji, Y. (2017). Nonlinear model predictive and differential braking for limit handling. In Proceedings of the 24th International
lateral stability control of active chassis for intelligent vehicles and its FPGA Symposium on Dynamics of Vehicles on Roads and Tracks (IAVSD).
implementation. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 49(1), Joa, E., Yi, K., & Kim, K. (2016). Development of integrated chassis control for limit
2–13. handling. SAE Technical Paper, 2016-01–1638.
Guo, J., Luo, Y., Li, K., & Dai, Y. (2018). Coordinated path-following and direct yaw- Johansen, T. A., & Fossen, T. I. (2013). Control allocation — A survey. Automatica, 49(5),
moment control of autonomous electric vehicles with sideslip angle estimation. 1087–1103.
Mechanical Systems and Signal Processing, 105, 183–199. Jonner, W. D., Winner, H., Dreilich, L., & Schunck, E. (1996). Electrohydraulic brake
Gwak, J., Jung, J., Oh, R., & Park, M. (2019). A review of intelligent self-driving vehicle system-the first approach to brake-by-wire technology. SAE Transactions,
software research. KSII Transactions on Internet and Information Systems, 13(11), 1368–1375, 960991.
5299–5320. Kaiser, G. (2015). Torque Vectoring: Linear Parameter-Varying Control for an Electric
Hac, A. B., Chen, H. H., Bedner, E. J., & Loudon, S. P. (2002). Integrated control of active Vehicle. Hamburg-Harburg University of Technology.
tire steer and brakes. Patent No. US 6,453,226 B1. Kapania, N. R., & Gerdes, J. C. (2015). Design of a feedback-feedforward steering
Hajiloo, R., Abroshan, M., Khajepour, A., Kasaiezadeh, A., & Chen, S. K. (2020). controller for accurate path tracking and stability at the limits of handling. Vehicle
Integrated steering and differential braking for emergency collision avoidance in System Dynamics, 53(12), 1687–1704.
autonomous vehicles. IEEE Transactions on Intelligent Transportation Systems. In press. Katsuyama, E. (2013). Decoupled 3D moment control for vehicle motion using in-wheel
Hamersma, H. A., & Els, P. S. (2014). Improving the braking performance of a vehicle motors. SAE International Journal of Passenger Cars-Mechanical Systems, 6(2013-01-
with ABS and a semi-active suspension system on a rough road. Journal of 0679), 137–146.
Terramechanics, 56, 91–101. Kavitha, C., Shankar, S. A., Ashok, B., Ashok, S. D., Ahmed, H., & Kaisan, M. U. (2018).
Hang, P., & Chen, X. (2019). Integrated chassis control algorithm design for path tracking Adaptive suspension strategy for a double wishbone suspension through camber and
based on four-wheel steering and direct yaw-moment control. Proceedings of the toe optimization. Engineering Science and Technology, an International Journal, 21(1),
Institution of Mechanical Engineers, Part I: Journal of Systems and Control Engineering, 149–158.
233(6), 625–641. Kavitha, C., Shankar, S. A., Karthika, K., Ashok, B., & Ashok, S. D. (2019). Active camber
Hang, P., Chen, X., & Luo, F. (2019). LPV / H ∞ controller design for path tracking of and toe control strategy for the double wishbone suspension system. Journal of King
autonomous ground vehicles through four-wheel steering and direct yaw moment Saud University - Engineering Sciences, 31(4), 375–384.
control. International Journal of Automotive Technology, 20(4), 679–691. Kawakami, H., Sato, H., Tabata, M., Inoue, H., & Itimaru, H. (1992). Development of
He, J., Crolla, D. A., Levesley, M. C., & Manning, W. J. (2006). Coordination of active integrated system between active control suspension, active 4WS, TRC and ABS. SAE
steering, driveline, and braking for integrated vehicle dynamics control. Proceedings Transactions, 101(6), 326–333.
of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 220 Khosravani, S., Jalali, M., Khajepour, A., Kasaiezadeh, A., Chen, S. K., & Litkouhi, B.
(10), 1401–1420. (2018). Application of lexicographic optimization method to integrated vehicle
He, L., Shen, T., Yu, L., Feng, N., & Song, J. (2013). A model-predictive-control-based control systems. IEEE Transactions on Industrial Electronics, 65(12), 9677–9686.
torque demand control approach for parallel hybrid powertrains. IEEE Transactions Kim, E., Kim, J., & Sunwoo, M. (2014). Model predictive control strategy for smooth path
on Vehicular Technology, 62(3), 1041–1052. tracking of autonomous vehicles with steering actuator dynamics. International
He, X., Yang, K., & Liu, Y. (2018). A novel direct yaw Moment control system for Journal of Automotive Technology, 15(7), 1155–1164.
autonomous vehicle. SAE Technical Paper, 2018-01–1594. Kim, H., Park, J., Jeon, K., & Choi, S. (2013). Integrated control strategy for torque
Heißing, B., & Ersoy, M. (2011). Chassis handbook: fundamentals, driving dynamics, vectoring and electronic stability control for in wheel motor EV. In Proceedings of the
components, mechatronics, perspectives. Vieweg+Teubner. 2013 World Electric Vehicle Symposium and Exhibition (EVS27).
Her, H., Joa, E., Yi, K., & Kim, K. (2016). Integrated chassis control for optimized tyre Kirli, A., Chen, Y., Okwudire, C. E., & Ulsoy, A. G. (2019). Torque-Vectoring-Based
force coordination to enhance the limit handling performance. Proceedings of the Backup Steering Strategy for Steer-by-Wire Autonomous Vehicles With Vehicle
Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 230(8), Stability Control. IEEE Transactions on Vehicular Technology, 68(8), 7319–7328.
1011–1026. Kissai, M., Monsuez, B., & Tapus, A. (2017). Review of integrated vehicle dynamics
Her, H., Koh, Y., Joa, E., Yi, K., & Kim, K. (2015). An integrated control of differential control architectures. In Proceedings of the 2017 European Conference on Mobile Robots
braking, front/rear traction, and active roll moment for limit handling performance. (ECMR).
IEEE Transactions on Vehicular Technology, 65(6), 4288–4300. Kissai, M., Mouton, X., Monsuez, B., Martinez, Di., & Tapus, A. (2018). Optimizing
Her, H., Suh, J., & Yi, K. (2015). Integrated control of the differential braking, the vehicle motion control for generating multiple sensations. In Proceedings of the 2018
suspension damping force and the active roll moment for improvement in the agility IEEE Intelligent Vehicles Symposium.
and the stability. Proceedings of the Institution of Mechanical Engineers, Part D: Journal König, L., Schindele, F., & Zimmermann, A. (2018). ITC–model-based feed forward
of Automobile Engineering, 229(9), 1145–1157. traction control. In Proceedings of the 9th International Munich Chassis Symposium
Hirano, Y., Harada, H., Ono, E., & Takanami, K. (1993). Development of an integrated 2018.
system of 4WS and 4WD by H∞ control. SAE Technical Paper, Article 930267. König, L., Schindele, F., & Zimmermann, A. (2019). ITC–Integrated traction control for
Hiraoka, T., Nishihara, O., & Kumamoto, H. (2009). Automatic path-tracking controller sports car applications. In Proceedings of the 9th Internationales Stuttgarter Symposium.
of a four-wheel steering vehicle. Vehicle System Dynamics, 47(10), 1205–1227. König, L., Walter, T., Gutmayer, B., & Merlein, D. (2014). Integrated vehicle dynamics
Horst, O., Heinrich, P., & Langer, F. (2014). ICT-Architecture for Multi-Modal Electric control – an optimized approach for linking multiple chassis actuators. In Proceedings
Vehicles. CoFAT. of the 14th Internationales Stuttgarter Symposium.
Hou, Y., Zhang, J., Zhang, Y., & Chen, L. (2008). Integrated chassis control using ANFIS. Kou, Y., Kim, W., Yoon, S., Lee, J., & Kim, D. (2004). Integration chassis control (ICC)
In Proceedings of the 2008 IEEE International Conference on Automation and Logistics. systems of Mando. SAE Technical Paper, 2004-01–2044.
Hrovat, D. D., Tran, M. N., & Yester, J. L. (1998). Traction control for moving a vehicle from Kritayakirana, K., & Gerdes, J. C. (2012). Using the centre of percussion to design a
deep snow. Patent No. US 5,735,362. steering controller for an autonomous race car. Vehicle System Dynamics, 50(sup1),
Hwang, J. Y., Lee, H., Kim, J. H., Kim, J. H., & Kwak, B. H. (2007). Coordinated control of 33–51.
the brake control system and the driveline control system. In Proceedings of the 2007 Lee, A. Y. (2002). Coordinated control of steering and anti-roll bars to alter vehicle
International Conference on Control, Automation and Systems. rollover tendencies. ASME. Journal of Dynamic Systems, Measurement, and Control,
ISO 14512. (1999). Passenger cars — Straight-ahead braking on surfaces with split coefficient 124(1), 127–132.
of friction — Open-loop test procedure. Lee, J. W., & Litkouhi, B. (2013). Method to adaptively control vehicle operation using an
ISO 19365. (2016). Passenger cars — Validation of vehicle dynamic simulation — Sine with autonomic vehicle control system. Patent No. US 8,428,843 B2.
dwell stability control testing. Lee, J. W., Salinger, J. A., & Chen, X. (2012). Method to assess risk associated with operating
ISO 21994. (2007). Passenger cars — Stopping distance at straight-line braking with ABS — an autonomic vehicle control system. Patent No. US 8,244,408 B2.
Open-loop test method. Lenzo, B., Zanchetta, M., Sorniotti, A., Gruber, P., & Nijs, W. De (2020). Yaw rate and
ISO 2631-1. (1997). Mechanical Vibration and Shock - Evaluation of Human Exposure to sideslip angle control through single input single output direct yaw moment control.
Whole-Body Vibration - Part 1: General Requirements. IEEE Transactions on Control Systems Technology, 29(1), 124–139.
ISO 3888-1. (2018). Passenger cars–test track for a severe lane-change manoeuvre–Part 1: Leung, K., Schmerling, E., Zhang, M., Chen, M., Talbot, J., Gerdes, J. C., & Pavone, M.
double lane-change. (2020). On infusing reachability-based safety assurance within planning frameworks
ISO 3888-2. (2011). Passenger cars — Test track for a severe lane-change manoeuvre — Part for human – robot vehicle interactions. The International Journal of Robotics Research,
2: Obstacle avoidance. 39(10–11), 1326–1345.

202
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Li, D., Du, S., & Yu, F. (2008). Integrated vehicle chassis control based on direct yaw Ni, J. (2017). Envelope control for four-wheel independently actuated autonomous
moment, active steering and active stabiliser. Vehicle System Dynamics, 46(sup1), ground vehicle through. IEEE Transactions on Vehicular Technology, 66(11),
341–351. 9712–9726.
Li, D., Li, B., Yu, F., Du, S., & Zhang, Y. (2007). A top-down integration approach to Nishio, Y., & Shen, T. (2019). Model predictive control with traffic information-based
vehicle stability control. In Proceedings of the 2007 IEEE International Conference on driver’s torque demand prediction for diesel engines. International Journal of Engine
Vehicular Electronics and Safety, ICVES. Research, 2(2), 674–684.
Li, L., Ran, X., Wu, K., Song, J., & Han, Z. (2015). A novel fuzzy logic correctional Nocedal, J., & Wright, S. J. (2006). Numerical optimization. Springer.
algorithm for traction control systems on uneven low-friction road conditions. Ono, E., Takanami, K., Iwama, N., Hayashi, Y., Hirano, Y., & Satoh, Y. (1994). Vehicle
Vehicle System Dynamics, 53(6), 711–733. integrated control for steering and traction systems by μ-synthesis. Automatica, 30
Li, S., & Arat, M. A. (2016). A hierarchical integrated chassis control strategy based on (11), 1639–1647.
wheel force feedback. International Journal of Vehicle Performance, 2(4), 327–352. Ono, E., Hattori, Y., Muragishi, Y., & Koibuchi, K. (2006). Vehicle dynamics integrated
Liang, Z., Zhao, J., Dong, Z., Wang, Y., & Ding, Z. (2020). Torque vectoring and rear- control for four-wheel-distributed steering and four-wheel-distributed traction/
wheel-steering control for vehicle’s uncertain slips on soft and slope terrain using braking systems. Vehicle System Dynamics, 44(2), 139–151.
sliding mode algorithm. IEEE Transactions on Vehicular Technology, 69(4), Paden, B., Cáp, M., Yong, S. Z., Yershov, D., & Frazzoli, E. (2016). A survey of motion
3805–3815. planning and control techniques for self-driving urban vehicles. IEEE Transactions on
Lin, F., Zhang, Y., Zhao, Y., Yin, G., Zhang, H., & Wang, K. (2019). Trajectory Tracking of Intelligent Vehicles, 1(1), 33–55.
Autonomous Vehicle with the Fusion of DYC and longitudinal – lateral control. Parra, A., Tavernini, D., Gruber, P., Sorniotti, A., & Zubizarreta, A. (2020). On nonlinear
Chinese Journal of Mechanical Engineering, 32(1), 16. model predictive control for energy-efficient torque-vectoring. IEEE Transactions on
Lohner, H., Traechtler, A., Futterer, S., Verhagen, A., Frese, K., Gerdes, M., Vehicular Technology, 70(1), 173–188, 2021.
Sackmann, M., & Martini, D. (2007). Coordination of a vehicle dynamics control system Pascali, L., Gabrielli, P., & Caviasso, G. (2003). Improving vehicle handling and comfort
with other vehicles stability systems. Patent No. US 7,305,292 B2. performance using 4WS. SAE Transactions, 998–1006, 2003-01-0961.
Lou, S., Fu, Z., Zhang, L., & Xu, C. (2010). Integrated control of semi-active suspension Peng, H., Wang, W., An, Q., Xiang, C., & Li, L. (2020). Path tracking and direct yaw
and ABS based on sliding mode theory. In Proceedings of the 29th Chinese Control moment coordinated control based on robust MPC with the finite time horizon for
Conference. autonomous independent-drive vehicles. IEEE Transactions on Vehicular Technology,
Lu, Q., Gentile, P., Tota, A., Sorniotti, A., Gruber, P., Costamagna, F., & De Smet, J. 69(6), 6053–6066.
(2016). Enhancing vehicle cornering limit through sideslip and yaw rate control. Peters, Y., & Stadelmayer, M. (2019). Control allocation for all wheel drive sports cars
Mechanical Systems and Signal Processing, 75, 455–472. with rear wheel steering. Automotive and Engine Technology, 4(3–4), 111–123.
Lv, C., Wang, H., & Cao, D. (2018). Brake-Blending Control of EVs. Modeling, dynamics, Plochl, M., & Lugner, P. (1996). Braking behaviour of a 4-wheel-steered automobile with
and control of electrified vehicles. Elsevier Inc. an antilock braking system. Vehicle System Dynamics, 25(sup1), 547–558.
Majersik, L. E., Katrak, K. K., Palazzolo, S. D., & Seifert, M. A. (2010). Method and system Poussot-Vassal, C., Sename, O., Dugard, L., Gáspár, P., Szabó, Z., & Bokor, J. (2011).
for coordinating a vehicle stability control system with a suspension damper control sub- Attitude and handling improvements through gain-scheduled suspensions and
system. Patent No. US 7,706,941 B2. brakes control. Control Engineering Practice, 19(3), 252–263.
March, C., Shim, T., & Zhang, Y. (2007). Integrated control of suspension and front Pretagostini, F., Ferranti, L., Ivanov, V., & Shyrokau, B. (2020). Survey on wheel slip
steering to enhance vehicle handling. Proceedings of the Institution of Mechanical control design strategies, evaluation and application to antilock braking systems.
Engineers, Part D: Journal of Automobile Engineering, 221(4), 377–391. IEEE Access, 8, 10951–10970.
Mares, M. E. (2014). Active aerodynamic chassis control. Patent No. US 8,798,868 B2. Rahimi, S., & Naraghi, M. (2018). Design of an integrated control system to enhance
Mastinu, G., Babbel, E., Lugner, P., Margolis, D., Mittermayr, P., & Richter, B. (1994). vehicle roll and lateral dynamics. Transactions of the Institute of Measurement and
Integrated controls of lateral vehicle dynamics. Vehicle System Dynamics, 23(sup1), Control, 40(5), 1435–1446.
358–377. Reif, K. (2014). Automotive mechatronics. Springer Vieweg.
Matsumoto, N., & Tomizuka, M. (1992). Vehicle lateral velocity and yaw rate control Reinold, P., & Traechtler, A. (2013). Closed-loop control with optimal tire-force
with two independent control inputs. ASME. Journal of Dynamic Systems, distribution for the horizontal dynamics of an electric vehicle with single-wheel
Measurement, and Control, 114(4), 606–613. chassis actuators. In Proceedings of the 2013 American Control Conference.
Metzler, M., Tavernini, D., Gruber, P., & Sorniotti, A. (2020). On Prediction Model Ren, Y., Zheng, L., & Khajepour, A. (2018). Integrated model predictive and torque
Fidelity in Explicit Nonlinear Model Predictive Vehicle Stability Control. IEEE vectoring control for path tracking of 4-wheel-driven autonomous vehicles. IET
Transactions on Control Systems Technology, 1–17. In press. Intelligent Transport Systems, 13(1), 98–107.
Metzler, M., Tavernini, D., Sorniotti, A., & Gruber, P. (2019). Explicit non-linear model Rengaraj, C. and Crolla, D., Integrated chassis control to improve vehicle handling
predictive control for vehicle stability control. In Proceedings of the 9th International dynamics performance. SAE Technical Paper 2011, 2011-01–0958.
Munich Chassis Symposium 2018. Reul, M., & Winner, H. (2009). Enhanced braking performance by integrated ABS and
Mihailescu, A., Scharfenbaum, I., Schaaf, U., & Schimmel, C. (2019). Efficient chassis semi-active damping control. In Proceedings of the 21st International Technical
application with integral controller and virtual methodology. ATZ Worldwide, 121 Conference on the Enhanced Safety of Vehicles, (ESV).
(3), 42–47. Reuss, H. C., Herold, J. H., & Pascali, L. (2014). Integration of chassis control system
Milliken, W. F., & Milliken, D. L. (1995). Race car vehicle dynamics. Society of Automotive networking into the vehicle dynamics development process. In Proceedings of the 14
Engineers, Inc. Internationales Stuttgarter Symposium.
Mirzaei, M., & Mirzaeinejad, H. (2017). Fuzzy scheduled optimal control of integrated Ricciardi, V., Ivanov, V., Dhaens, M., Vandersmissen, B., Geraerts, M., Savitski, D., &
vehicle braking and steering systems. IEEE/ASME Transactions on Mechatronics, 22 Augsburg, K. (2019). Ride blending control for electric vehicles. World Electric
(5), 2369–2379. Vehicle Journal, 10(36), 1–13.
Mirzaeinejad, H., Mirzaei, M., & Kazemi, R. (2016). Enhancement of vehicle braking Ricco, M., Zanchetta, M., Rizzo, G. C., Tavernini, D., Sorniotti, A., Chatzikomis, C.,
performance on split- μ roads using optimal integrated control of steering and Velardocchia, M., Geraerts, M., & Dhaens, M. (2020). On the design of yaw rate
braking systems. Journal of Multi-Body Dynamics, 230(4), 401–415. control via variable front-to-total anti-roll moment distribution. IEEE Transactions on
Mirzaeinejad, H., Mirzaei, M., & Rafatnia, S. (2018). A novel technique for optimal Vehicular Technology, 69(2), 1388–1403.
integration of active steering and differential braking with estimation to improve Robert Bosch Gmbh. (2007). Bosch automotive electrics and automotive electronics: systems
vehicle directional stability. ISA Transactions, 80, 513–527. and components, networking and hybrid drive. Springer Vieweg.
Mitamura, R., Tani, M., Tanaka, T., & Yuasa, M. (1988). System integration for new SAE J3016. (2014). Taxonomy and definitions for terms related to on-road motor vehicle
mobility. SAE Transactions, 1418–1428, 881773. automated driving systems.
Montanaro, U., Dixit, S., Fallah, S., Dianati, M., Oxtoby, D., & Mouzakitis, A. (2018). Saikia, A., & Pathak, M. (2019). Vehicle stability enhancement based on unified chassis
Towards connected autonomous driving : Review of use-cases. Vehicle System control with electronic stability control and active front steering. Advanced Research
Dynamics, 57(6), 779–814. in Electrical and Electronic Engineering, 6(1), 37–42.
Morrison, G., & Cebon, D. (2017). Extremum-seeking algorithms for the emergency Salehpour, S., Pourasad, Y., & Taheri, S. H. (2015). Vehicle path tracking by integrated
braking of heavy goods vehicles. Proceedings of the Institution of Mechanical Engineers, chassis control. Journal of Central South University, 22(4), 1378–1388.
Part D: Journal of Automobile Engineering, 231(14), 1961–1977. Salman, M., Zhang, Z., & Boustany, N. (1992). Coordinated control of four wheel braking
Mousavinejad, E., Han, Q. L., Yang, F., Zhu, Y., & Vlacic, L. (2017). Integrated control of and rear steering. In Proceedings of the 1992 American Control Conference.
ground vehicles dynamics via advanced terminal sliding mode control. Vehicle Sato, S., & Inoue, H. (1993). Integrated chassis control system for improved vehicle
System Dynamics, 55(2), 268–294. dynamics. In Proceedings of the International Symposium on Advanced Vehicle Control.
Nagai, M., Hirano, Y., & Yamanaka, S. (1997). Integrated control of active rear wheel Savitski, D., Ivanov, V., Augsburg, K., Dhaens, M., Els, S., & Sandu, C. (2015). State-of-
steering and direct yaw moment control. Vehicle System Dynamics, 27(5–6), 357–370. the-art and future developments in integrated chassis control for ground vehicles. In
Nagai, M., Hirano, Y., & Yamanaka, S. (1998). Integrated robust control of active rear Proceedings of the 13th European Conference of the ISTVS.
wheel steering and direct yaw moment control. Vehicle System Dynamics, 29(sup1), Scalzi, S., & Marino, R. (2008). Integrated active front steering and semiactive rear
416–421. differential control in rear wheel drive vehicles. IFAC Proceedings Volumes, 41(2),
Nagai, M., Shino, M., & Gao, F. (2002). Study on integrated control of active front steer 10732–10737.
angle and direct yaw moment. JSAE Review, 23(3), 309–315. Scamarcio, A., Gruber, P., De Pinto, S., & Sorniotti, A. (2020). Anti-jerk controllers for
Nah, J., & Yim, S. (2019). Optimization of control allocation with ESC, AFS, ARS and automotive applications: A review. Annual Reviews in Control, 50, 174–189.
TVD in integrated chassis control. Journal of Mechanical Science and Technology, 33 Schiebahn, M., Zegelaar, P. W. A., Lakehal-Ayat, M., & Hofmann, O. (2010). The yaw
(6), 2941–2948. torque influence of active systems and smart actuators for coordinated vehicle
Németh, B., Gáspár, P., Fenyes, D., & Bokor, J. (2017). Robust control design for the dynamics controls. Vehicle System Dynamics, 48(11), 1269–1284.
integration of steering and torque vectoring using a variable-geometry suspension
system. In Proceedings of the 2017 American Control Conference (ACC).

203
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Schmidt, F., & König, L. (2020). Will driving still be fun in the future? Vehicle dynamics Trachtler, A. (2004). Integrated vehicle dynamics control using active brake, steering
systems through the ages. In Proceedings of the 10th international Munich Chassis and suspension systems. International Journal of Vehicle Design, 36(1), 1–12.
Symposium 2019. Tracy, D. (2019). An Extremely Detailed Look At The Porsche Taycan’s Engineering Designed
Selby, M. A. (2003). Intelligent vehicle motion control. University of Leeds. To Take On Tesla. Jalopnik. https://jalopnik.com/an-extremely-detailed-look-at-the-
Seo, I. S., & Park, J.-I. (2020). Active geometry control suspension (Patent No. US porsche-taycans-engin-1837802533.
10,710,634 B2). Tranquillo, G., Sorrentino, A., & Van, V. (2020). From mechanical system to tire
Shao, J., Zheng, L., Li, Y. N., Wei, J. S., & Luo, M. G. (2007). The integrated control of performance impact: D.A.S. (Dual Axis Steering) explained thanks to advanced modeling.
anti-lock braking system and active suspension in vehicle. In Proceedings of the 4th vi-grade, https://www.vi-grade.com/en/products/public_download/1688.
International Conference on Fuzzy Systems and Knowledge Discovery (FSKD 2007). Tseng, H. E., Ashrafi, B., Madau, D., Brown, T. A., & Recker, D. (1999). The development
Shao, L., Ece, A. H. U., Karci, H., Tavernini, D., Sorniotti, A., & Cheng, M (2020). Design of vehicle stability control at Ford. IEEE/ASME Transactions on Mechatronics, 4(3),
approaches and control strategies for energy-efficient electric machines for electric 223–234.
vehicles — A review. IEEE Access, 8, 116900–116913. UN/ECE 140. (2018). Uniform provisions concerning the approval of passenger cars with
Shen, P., Zhao, Z., Zhan, X., & Li, J. (2017). Particle swarm optimization of driving regard to Electronic Stability Control (ESC) Systems, Official Journal of the European
torque demand decision based on fuel economy for plug-in hybrid electric vehicle. Union [2018/1592].
Energy, 123, 89–107. Valášek, M., Vaculin, O., & Kejval, J. (2004). Global chassis control : integration synergy
Shen, X. M., & Yu, F. (2006a). Investigation on integrated vehicle chassis control based of brake and suspension control for active safety. In Proceedings of the 7th
on vertical and lateral tyre behaviour correlativity. Vehicle System Dynamics, 44 International Symposium on Advanced Vehicle Control.
(sup1), 506–519. Van Zanten, A. T. (2000). Bosch ESP systems: 5 years of experience. SAE Transactions,
Shen, X. M., & Yu, F. (2006b). Study on vehicle chassis control integration based on a 2000-01-1633, 428-436.
main-loop-inner-loop design approach. Proceedings of the Institution of Mechanical Van Zanten, A. T., Erhardt, R., Lutz, A., Neuwald, W., & Bartels, H. (1996). Simulation for
Engineers, Part D: Journal of Automobile Engineering, 220(11), 1491–1502. the development of the Bosch-VDC. SAE Transactions, 544–554, 960486.
Shen, X. M., & Yu, F. (2007). Study on vehicle chassis control integration based on Van Zanten, A. T., Erhardt, R., & Pfaff, G. (1995). VDC, the vehicle dynamics control
vehicle dynamics and separate loop design approach. International Journal of Vehicle system of Bosch. SAE Transactions, 1419–1436, 950759.
Autonomous Systems, 5. Velardocchia, M., & Vigliani, A. (2013). Control systems integration for enhanced vehicle
Shibahata, Y. (2005). Progress and future direction of chassis control technology. Annual dynamics. The Open Mechanical Engineering Journal, 7(1), 58–69.
Reviews in Control, 29(1), 151–158. Velenis, E., Tsiotras, P., & Lu, J. (2008). Trail-Braking driver input parameterization. SAE
Shibahata, Y., Abe, M., Shimada, K., & Furukawa, Y. (1994). Improvement on limit Technical Paper, 2008-01-2986.
performance of vehicle motion by chassis control. Vehicle System Dynamics, 23(sup1), Vivas-Lopez, C. A., Tudon-Martinez, J. C., Hernandez-Alcantara, D., & Morales-
449–468. Menendez, R. (2015). Global chassis control system using suspension, steering, and
Shimada, K., & Shibahata, Y. (1994). Comparison of three active chassis control methods braking subsystems. Mathematical Problems in Engineering, 2015, 1–18.
for stabilizing yaw moments. SAE Transactions, 1178–1187, 940870. Vo, D. Q., Marzbani, H., Fard, M., & Jazar, R. N. (2018). Variable caster steering in
Shuai, Z., Zhang, H., Wang, J., Li, J., & Ouyang, M. (2013). Combined AFS and DYC vehicle dynamics. International Journal of Automobile Engineering, 232(9),
control of four-wheel-independent-drive electric vehicles over CAN network with 1270–1284.
time-varying delays. IEEE Transactions on Vehicular Technology, 63(2), 591–602. Wahid, N., Zamzuri, H., Hassan, N., & Rahman, M. A. A. (2017). Potential field based
Shyrokau, B., & Wang, D. (2012). Control allocation with dynamic weight scheduling for motion planning with steering control and DYC for ADAS. Telkomnika
two-task integrated control allocation with dynamic weight scheduling for two-task (Telecommunication Computing Electronics and Control), 15(2), 853–860.
integrated vehicle control. In Proceedings of the 11th International Symposium on Wang, C., Zhao, W., Luan, Z., Gao, Q., & Deng, K. (2018). Decoupling control of vehicle
Advanced Vehicle Control. chassis system based on neural network inverse system. Mechanical Systems and
Shyrokau, B., Wang, D., & Lienkamp, M. (2013). Integrated vehicle dynamics control Signal Processing, 106, 176–197.
based on control allocation with subsytem coordination. In Proceedings of the 23rd Wang, J. X., Chen, N., Pi, D. W., & Yin, G. D. (2009). Agent-based coordination
International Symposium on Dynamics of Vehicles on Roads and Tracks (IAVSD). framework for integrated vehicle chassis control. Proceedings of the Institution of
Shyrokau, B., Wang, D., Savitski, D., Hoepping, K., & Ivanov, V. (2015). Vehicle motion Mechanical Engineers, Part D: Journal of Automobile Engineering, 223(5), 601–621.
control with subsystem prioritization. Mechatronics, 30, 297–315. Wang, S., Zhao, X., & Yu, Q. (2020). Vehicle Stability Control Strategy Based on
Singh, K. B., Arat, M. A., & Taheri, S. (2013). An intelligent tire based tire-road friction Recognition of Driver Turning Intention for Dual-Motor Drive Electric Vehicle.
estimation technique and adaptive wheel slip controller for antilock brake system. Mathematical Problems in Engineering, 2020.
Journal of Dynamic Systems, Measurement and Control, Transactions of the ASME, 135 Warth, G., Frey, M., & Gauterin, F. (2018). Usage of the cornering stiffness for an
(3). adaptive rear wheel steering feedforward control. IEEE Transactions on Vehicular
Smakman, H. T. (2000). Functional integration of slip control with active suspension for Technology, 68(1), 264–275.
improved lateral vehicle dynamics. Delft University of Technology. Warth, G., Frey, M., & Gauterin, F. (2020). Design of a central feedforward control of
Soltani, A., Bagheri, A., & Azadi, S. (2018). Integrated vehicle dynamics control using torque vectoring and rear-wheel steering to beneficially use tyre information. Vehicle
semi-active suspension and active braking systems. Proceedings of the Institution of System Dynamics, 58(12), 1789–1822.
Mechanical Engineers, Part K: Journal of Multi-Body Dynamics, 232(3), 314–329. Wen, H. H., Chen, W., & Hui, J. (2018). A single-pedal regenerative braking control
Sommer, S., Camek, A., Becker, K., Buckl, C., Zirkler, A., Fiege, L., Armbruster, M., strategy of accelerator pedal for electric vehicles based on adaptive fuzzy control
Spiegelberg, G., & Knoll, A. (2013). RACE: A centralized platform computer based algorithm. Energy Procedia, 152, 624–629.
architecture for automotive applications. In Proceedings of the 2013 IEEE International Wu, H., Si, Z., & Li, Z. (2020). Trajectory Tracking Control for Four-Wheel Independent
Electric Vehicle Conference, IEVC 2013. Drive Intelligent Vehicle Based on Model Predictive Control. IEEE Access, 8,
Song, P., Tomizuka, M., & Zong, C. (2015). A novel integrated chassis controller for full 73071–73081.
drive-by-wire vehicles. Vehicle System Dynamics, 53(2), 215–236. Wu, J., Zhou, H., Liu, Z., & Gu, M. (2020). Ride Comfort Optimization via Speed Planning
Sorniotti, A., Barber, P., & De Pinto, S. (2017). Path tracking for automated driving: A and Preview Semi-Active Suspension Control for Autonomous Vehicles on Uneven
tutorial on control system formulations and ongoing research. In D. Watzenig, & Roads. IEEE Transactions on Vehicular Technology, 69(8), 8343–8355.
M. Horn (Eds.), Automated Driving: Safer and More Efficient Future Driving (pp. Xia, G., Guo, D., Tang, X., & Chen, W. (2019). Coordinated control of a semi-active
71–140). Springer International Publishing. suspension system and an electronic stability program on a finite-state basis. In
Sun, J., Cong, J. Y., Gu, L., & Dong, M. (2019). Fault-tolerant control for vehicle with Proceedings of the INTER-NOISE and NOISE-CON Congress and Conference.
vertical and lateral dynamics. Proceedings of the Institution of Mechanical Engineers, Xia, H., Chen, J., Liu, Z., & Lan, F. (2020). Coordinated motion control for automated
Part D: Journal of Automobile Engineering, 233(12), 3165–3184. vehicles considering steering and driving force saturations. Transactions of the
Sun, S., Tang, X., Yang, J., Ning, D., Du, H., Zhang, S., & Li, W. (2019). A new generation Institute of Measurement and Control, 42(1), 157–166.
of magnetorheological vehicle suspension system with tunable stiffness and damping Xiang, C., Peng, H., Wang, W., Li, L., An, Q., & Cheng, S. (2020). Path tracking
characteristics. IEEE Transactions on Industrial Informatics, 15(8), 4696–4708. coordinated control strategy for autonomous four in-wheel-motor independent-drive
Švec, M., Hrvatinic, K., Ileš, & Matuško, J (2019). Predictive torque vectoring vehicle vehicles with consideration of lateral stability. Proceedings of the Institution of
control based on a linear time varying model. In Proceedings of the 42nd International Mechanical Engineers, Part D: Journal of Automobile Engineering, 2(4), 1023–1036.
Convention on Information and Communication Technology, Electronics and Xiao, H., Chen, W., Zhou, H., & Zu, J. W. (2011). Integrated control of active suspension
Microelectronics, MIPRO. system and electronic stability programme using hierarchical control strategy:
Taheri, S., & Law, E. H. (1990). Investigation of a combined slip control braking and Theory and experiment. Vehicle System Dynamics, 49(1–2), 381–397.
closed loop four wheel steering system for an automobile during combined hard Xie, X., Jin, L., Jiang, Y., & Guo, B. (2018). Integrated dynamics control system with ESC
braking and sever steering. In Proceedings of the 1990 American Control Conference. and RAS for a distributed electric vehicle. IEEE Access, 6, 18694–18704.
Tang, C., & Khajepour, A. (2020). Wheel modules with distributed controllers: A multi- Xiujian, Y., Zengcai, W., & Weili, P. (2009). Coordinated control of AFS and DYC for
agent approach to vehicular control. IEEE Transactions on Vehicular Technology, 69 vehicle handling and stability based on optimal guaranteed cost theory. Vehicle
(10), 10879–10888. System Dynamics, 47(1), 57–79.
Tchamna, R., Youn, E., & Youn, I. (2014). Combined control effects of brake and active Yim, S. (2012). Design of a robust controller for rollover prevention with active
suspension control on the global safety of a full-car nonlinear model. Vehicle System suspension and differential braking. Journal of Mechanical Science and Technology, 26
Dynamics, 52(sup1), 69–91. (1), 213–222.
Ting, W. E., & Lin, J. S. (2004). Nonlinear control design of anti-lock braking systems Yim, S. (2015). Integrated chassis control with adaptive algorithms. Journal of Automobile
combined with active suspensions. In Proceedings of the 2004 5th Asian Control Engineering, 1(9).
Conference. Yim, S. (2018). Selection of actuator combination in integrated chassis control using
Toyoshima, H. (2016). Hybrid powertrain for the new super sports car Honda NSX. MTZ Taguchi method. International Journal of Automotive Technology, 19(2), 263–270.
Worldwide, 77(6), 42–47.

204
V. Mazzilli et al. Annual Reviews in Control 51 (2021) 172–205

Yim, S., Choi, J., & Yi, K. (2012). Coordinated control of hybrid 4WD vehicles for Zhang, J., & Li, J. (2019). Integrated vehicle chassis control for active front steering and
enhanced maneuverability and lateral stability. IEEE Transactions on Vehicular direct yaw moment control based on hierarchical structure. Transactions of the
Technology, 61(4), 1946–1950. Institute of Measurement and Control, 41(9), 2428–2440.
Yim, S., & Jo, Y. H. (2019). Integrated chassis control with AFS, ARS and ESC under Zhang, L., Ding, H., Guo, K., Zhang, J., Pan, W., & Jiang, Z. (2018). Cooperative chassis
lateral force constraint on AFS. JMST Advances, 1(1–2), 13–21. control system of electric vehicles for agility and stability improvements. IET
Yim, S., Kim, S., & Yun, H. (2016). Coordinated control with electronic stability control Intelligent Transport Systems, 13(1), 134–140.
and active front steering using the optimum yaw moment distribution under a lateral Zhao, H., Chen, W., Zhao, J., Zhang, Y., & Chen, H. (2019). Modular integrated
force constraint on the active front steering. Proceedings of the Institution of longitudinal, lateral, and vertical vehicle stability control for distributed electric
Mechanical Engineers, Part D: Journal of Automobile Engineering, 230(5), 581–592. vehicles. IEEE Transactions on Vehicular Technology, 68(2), 1327–1338.
Yin, G., Li, J., Jin, X., Bian, C., & Chen, N. (2015). Integration of motion planning and Zhao, H., Gao, B., Ren, B., & Chen, H. (2015). Integrated control of in-wheel motor
model- predictive-control-based control system for autonomous electric vehicles. electric vehicles using a triple-step nonlinear method. Journal of the Franklin Institute,
Transport, 30(3), 353–360. 352(2), 519–540.
Yoon, J., Cho, W., Koo, B., & Yi, K. (2008). Unified chassis control for rollover prevention Zhao, J., Wong, P. K., Ma, X., & Xie, Z. (2017). Chassis integrated control for active
and lateral stability. IEEE Transactions on Vehicular Technology, 58(2), 596–609. suspension, active front steering and direct yaw moment systems using hierarchical
Yu, F., Li, D. F., & Crolla, D. A. (2008). Integrated vehicle dynamics control-state-of-the strategy. Vehicle System Dynamics, 55(1), 72–103.
art review. In Proceedings of the 2008 IEEE Vehicle Power and Propulsion Conference. Zheng, Y., & Shyrokau, B. (2019). A real-time nonlinear MPC for extreme lateral
Yu, L., & Liu, X. (2016). Review of brake-by-wire system used in modern passenger car. stabilization of passenger vehicles. In Proceedings of the IEEE Conference on
In Proceedings of the ASME 2016 International Design Engineering Technical Conference Mechatronics.
and Computers and Information in Engineering Conference. Zhou, Q., Wang, F., & Li, L. (2005). Robust Sliding mode control of 4WS Vehicles for
Yu, S. H., & Moskwa, J. J. (1994). A global approach to vehicle control: coordination of automatic path tracking. In Proceedings of the IEEE Intelligent Vehicles Symposium.
four wheel steering and wheel torques. Journal of Dynamic Systems, Measurement, and Zhu, B., Chen, Y., & Zhao, J. (2014). Integrated chassis control of active front steering
Control, 116(4), 659–667. and yaw stability control based on improved inverse nyquist array method. The
Zarkadis, K., Velenis, E., Siampis, E., & Longo, S. (2018). Predictive torque vectoring Scientific World Journal, 2014, 1–14.
control with active trail-braking. In Proceedings of the 2018 European Control Zhu, J., Wang, Z., Zhang, L., & Dorrell, D. G. (2019). Braking/steering coordination
Conference, ECC 2018. control for in-wheel motor drive electric vehicles based on nonlinear model
predictive control. Mechanism and Machine Theory, 142, 1–20.

205