Res TFG02
Res TFG02
net/publication/337623130
Article in Research Journal of Applied Sciences, Engineering and Technology · November 2019
CITATIONS READS
0 561
1 author:
Zlata Jelacic
University of Sarajevo
35 PUBLICATIONS 67 CITATIONS
SEE PROFILE
All content following this page was uploaded by Zlata Jelacic on 29 November 2019.
Assistant Professor, Faculty of Mechanical Engineering, University of Sarajevo, Bosnia and Herzegovina
Abstract - Before service robots will be part of our daily Similar to the computer industry, which quickly gained
lives, large-scale developments on both hardware and momentum after a standardized, open PC architecture was
software are required. Nevertheless, service robots are introduced, the development of service robots can greatly
currently only available to large research institutions that benefit from standardized interfaces that allow modular
have the resources to develop one. The service robots that substitution of components. Modular components with
are available are usually far too expensive. Furthermore, standardized interfaces i) allow the use of hardware
the available service robots are not modular, and their components for a specific task, ii) allow a robot to be
designs are closed-source, hampering further development. assembled from components of multiple developers and iii)
Therefore, a modular service robot is required with eases maintenance of the robots.
standardized interfaces. Furthermore, it must be open
hardware so that it can act as a reference implementation Modularity has been an important issue in a number of
for future developments. This paper presents best these designs. In [2] it is recognized that a modular approach
practices in service robots and introduces a control design makes it possible to add specialized hardware and end-
for a domestic service robot. effectors. The Care-O-bot 3 also has a modular setup [1] and
has a modified, existing manipulator connected to an existing
Keywords: service robot design, sensorimotor control, hybrid end-effector. Rollin' Justin ([5], [6]) also has a separately
control, anthropomorphic robot, cooperative manipulators. designed base, torso and arms. Similarly, ARMAR-III [7] and
HoLLiE [10] have been designed as a mobile platform with a
I. INTRODUCTION modular upper body. However, only the PR2 contains open
interfaces to use different grippers, arms or sensors and the
Robots are expected to play an increasing role in society
designs in [8] and [9] are in general less modular.
over the next number of years. They can, e.g., reduce the
pressure on the home care system by assisting people with As is argued in [11], open-source software
mobility impairments so that they can live independently at implementations of a standard means that the standard is more
home for a longer time. Furthermore, robots will also likely to be of high quality, since these implementations act as
increasingly work alongside people in industry. This illustrates reference implementations, and that the standard is more likely
the need for the development of versatile, reliable, robust and to be adopted. Unfortunately, most of these robots are unique
affordable robots. research platforms that are not available to other research
institutions ([1], [6]). The platforms that are available such as
As is mentioned in [1], it is expensive and time consuming
the PR2 [2] and the Care-O-bot 3 [4] are too expensive for
to develop these robots. Nevertheless, many service robots
many research institutions. In the field of humanoid, walking
have been developed over the past ten years. The development
robots, the importance of open-hardware has long been
of the PR2 (based on the design in [2] has concentrated on the
recognized, with the iCub [12], DARwIn-OP [13] and
design and implementation of a fully integrated development
NimbRo-OP [14] as well-known examples. These robots,
platform that is designed to be safe and capable in human
however, are not suitable as service robots.
environments. The long-term goal of the Care-O-bot project
([3],[4]) is to develop a mobile robot able to assist people in The contribution of this paper is a contribution to the
their homes and combines the technological aspects with a design and development of a domestic service robot with:
user-friendly design. The two-arm system in reference [5] is
developed as a research platform to contribute to the Best practices in the service robot design, both for the
manipulation skills of humanoid robots. Together with the upper body, base platforms and the manipulator
platform introduced in [6], it also forms the mobile service Mechanical design requirements
robot Rollin' Justin. More humanoid research platforms are Description of the manipulator kinematics and
ARMAR-III [7], Dynamaid [8], TWENDY-ONE [9] and contact dynamics
HoLLiE [10]. Description of the environmental model
Proposal of a controller for a cooperative manipulator
In the next section, the design principles of the base these fully holonomic platforms did not perform sufficiently
platforms are discussed, namely the design of base platform in robust in the presence of doorway thresholds, curbs and
Section IIa, its upper body and manipulators in Section IIb and extension cords. Nevertheless, fully holonomic platforms are
IIc, respectively. Section III describes the mechanical design also successfully used in [9] and [10].
requirements. Section IV focuses on the manipulator
kinematics and contact dynamics and Section V on the As is mentioned in [6], platforms with more than three
description of the environmental model. Finally, Section VI wheels are statically over determined and hence require some
presents a proposal for a controller for cooperative form of suspension in order to maintain good ground contact.
manipulators. Wheel suspension is not specifically addressed in [8], [9] and
[10]. Rollin' Justin [6] has an independent wheel suspension
II. BEST PRACTICES IN SERVICE ROBOTS system. An example of adding compliance using a rotational
degree of freedom between the front and rear axis can be
Most of the robots introduced in Section I consist of a found in, e.g., [15].
moving base, a torso with one or more degrees of freedom and
one or two manipulators. Although many of the tasks that are b) Upper bodies
currently demonstrated using service robots can be performed
with only one manipulator, an anthropomorphic robot with The successor of the design in [2] and the robot in [8] are
two manipulators and a moving torso has the advantage that it equipped with telescopic spines. This results in a large vertical
can also perform bimanual manipulation. Therefore, focus will motion while keeping the center of gravity (CoG) in the
lie on these anthropomorphic robots. In this section, various middle of the robot. On the other hand, the robots in [5], [7],
possibilities for base platforms, torsos and manipulators are [9] and [10], have rotational DoFs. The upper bodies in [5],
discussed. [9] and [10] each have one actuator per joint, whereas one of
the DoFs in [5] is passive, i.e., the tilt of the chest is coupled
a) Base platforms to the base via tendons. The main advantage of having a pitch
joint is that the robot can move its shoulders further forward,
Base platforms for service robots can be divided in three as is indicated in [5]. The absence of a pitch joint is a
categories: non-holonomic, semi-holonomic and fully drawback because the robot needs to be positioned very
holonomic platforms. A holonomic robot can drive in any accurately with respect to an object to allow a feasible
direction without having to turn beforehand. Hence, both the grasping motion.
number of controllable DoFs and the total number of DoFs
equal 3 (two translations and one rotation). A non-holonomic For large tilt motions, however, care has to be taken to
robot has fewer controllable DoFs than the total number of prevent the robot from tipping over. In [6], this is solved by
DoFs. For example, a car is non-holonomic: it cannot drive using a variable footprint, increasing the required number of
sideways. A semi-holonomic platform is able to drive actuators and the complexity and therewith the costs of the
sideways. However, it first has to turn its wheels. system. An alternative approach is presented in [10], where a
stability measure is optimized in motion planning. However,
Non-holonomic robots have two differentially driven the safety of the robot preferably does not depend on software.
wheels and one or two passive caster wheels, e.g., the Pioneer
P3-DX, or four or more differentially driven wheels, e.g., the c) Manipulators
Pioneer P3-AT. As is recognized, however, in [6], [7] and [8],
this is not very beneficial for manipulation and maneuvering One of the key abilities of a service robot is the ability to
in tight spaces. The semi-omni directionality of the robots in transport objects. Hereto, the robots in Section I are all
[2], [6] and [8], is obtained using steering wheels, i.e., four equipped with manipulators. These manipulators typically
individually driven steerable drives. Each drive has either one have up to seven degrees of freedom and most manipulators
or two wheels, hence the total number of motors required for a are either mechanically compliant or have torque sensors so
platform is either eight or twelve. This large number of that the controllers enable compliant control. The robots in [1]
actuators is the main drawback of these platforms. and [5] have industrial robot arms with seven DoFs and
harmonic drives. The arms in [5] are equipped with torque
Therefore, a base using either Mecanum wheels or Omni sensors to allow compliant control. Both arms, however, are
wheels is an interesting alternative. These wheels have small too large and too heavy for a dual arm domestic robot. The
rolls on their circumference, allowing the wheels to move robot in [10] has arms with six DoFs using three Schunk
freely in axial direction (in case of Omni-wheels) or under an Power balls, lacking the possibility to do compliant control.
angle (in case of Mecanum wheels). In [2], it was argued that The arms of the robot in [8] are actuated by Dynamixel servo
actuators. Although these have a compliant mode in which wheel. Compared to omni wheels, Mecanum wheels have a
they are back drivable, true impedance control is not possible. number of advantages:
The robots in [2] and [9] also have seven DoFs but are
mechanically compliant. With the same total width and wheel size, a
Mecanum wheel platform has a wider track and a
AMIGO's manipulators [16] have seven DoFs, of which longer wheelbase (0.55 m) compared to an omni
six DoFs are driven by three differential joints. The use of wheel platform (0.46 m), resulting in a more stable
differential joints results in a compact and anthropomorphic platform (see Figure 1).
design [17]. All joints are equipped with absolute position Together with a more favorable position of the
sensors as well as torque sensors to allow impedance control. motors this also results in more space for peripheral
Nevertheless, this design is also not open hardware. Although equipment.
recent developments show that variable-stiffness actuation is a Finally, driving over doorway thresholds is smoother
promising research direction, it is concluded that a rigid arm because the wheels roll along their circumference
with torque sensors leads to a simpler design of both hardware instead of the circumference of the small rollers.
and controllers. To keep the compact design and
anthropomorphic appearance, a design with differential joints
similar to, e.g., [16] and [17], is desired. Compared to the
current manipulators, a number of possible improvements is
identified: i) reduce the backlash in the differential joints, ii)
improve the absolute position sensors and iii) redesign the I/O
to be compatible with the rest of the robot.
The surface is assumed to be sufficiently smooth and contact friction is solely motivated by theoretical reasons and
convex. Common examples of such surfaces are planar, the effect of contact friction is left as a perturbation to the
nominal closed-loop system.
𝜑(𝑝𝑒)=𝑛𝑇(𝑝𝑒−𝑝0) (11)
VI. CONTROLLER FOR COOPERATIVE
Where 𝑝𝑒=[𝑝𝑒,𝑥𝑝𝑒,𝑦𝑝𝑒,𝑧]𝑇,𝑛∈ℝ3,𝑝0∈ℝ3 MANIPULATION
Are the Cartesian coordinates, normal to the plane and the Over the past decades, several control algorithms have
offset of the plane from the origin and spherical, been proposed for cooperative manipulators. The algorithms
that can control both forces and motion can be divided into
𝜑(𝑝𝑒)=‖𝑝𝑒−𝑝0‖−𝑟0
hybrid position/force control schemes and
Where 𝑝0∈ℝ3,𝑟0∈ℝ3>0 impedance/admittance control schemes. In the hybrid control
schemes, the coordination space is decoupled into motion and
Represent the position of its center and radius. For a force-controlled directions, using a predefined and fixed
frictionless and compliant surface, the normal component of selection matrix. Unexpected contact in motion-controlled
the contact force exerted by the end-effector on the surface is directions can lead to damage of the object and manipulators,
modeled as, since the force in these directions is not controlled. In the
impedance control schemes the dynamic relation between the
𝑓𝑒,𝑛={−𝑘𝑒ℎ 𝑖𝑓 ℎ≤0 0 𝑖𝑓 ℎ>0 ( 12) forces and motion of the system is taken into account. Using
impedance control, the task can be executed without leading to
Where ke characterizes the stiffness of the surface. This contact instability in the absence of precise knowledge of the
model is known in the literature as the Kelvin model. The contact directions. However, to achieve a satisfactory level of
complete contact force vector exerted by the end-effector on position/force tracking performance, the precise knowledge of
the surface is given by, contact directions is still required.
𝑓𝑒=𝑛(𝑝𝑒)𝑓𝑒,𝑛 𝑤ℎ𝑒𝑟𝑒 𝑛(𝑝𝑒)=(𝜕𝜑(𝑝𝑒)𝜕𝑝𝑒)𝑇 (13) A cascade controller for cooperative manipulation is
introduced. The block diagram of this controller is shown in
Where 𝑛(𝑝𝑒) is the normal direction of the surface. The
Figure 2. A motion controller is at the lowest level of this
effect of contact friction can also be included in the model by
scheme. The inner motion control loop is added to improve the
considering the tangential components of the contact force
tracking performance. The reference for this motion control
𝑓𝑒=𝑛(𝑝𝑒)𝑓𝑒,𝑛+𝑡(𝑝𝑒)𝑓𝑒,𝑡 (14) law is obtained from an impedance relationship driven by the
internal force error. The desired trajectory for each
Where 𝑡(𝑝𝑒) represent the tangential directions of the manipulator is obtained from the kinematic constraints
surface and 𝑓𝑒,𝑡 the friction force which depends on the between the object and the respective manipulator. Finally, the
magnitude of the normal force (i.e. |𝑓𝑒,𝑛|) and velocity of the desired object motion is obtained from the external force-
end-effector. For the case h ≤ 0, the robot is in contact with the based impedance controller.
environment (i.e. contact motion phase) and the surface is
assumed frictionless. The assumption on the absence of
VI. DISCUSSION AND CONCLUSION robots. Possible motor tasks that could be used in such
paradigms include but are not limited to: handshake, partner
For cooperative manipulation tasks, it is important to dance, sawing, carrying objects, leading an individual with
simultaneously control the motion of the system comprised of visual impairment, and as a more direct clinical task, therapist-
the manipulators and the object, the interaction forces between patient interactions during rehabilitation.
the object and the manipulators (internal forces) and the
contact forces between the object and the environment While several of these tasks have served as motivation for
(external forces). a number of human-human interface studies, there is a basic
lack of understanding of the magnitude of forces used in these
The proposed hybrid control, consisting of the hybrid tasks. Additionally, how interaction forces contribute to task
position/force control and impedance control, takes the performance, or provide a channel for communicating
previously described major contact problem issues into information about motor performance, intent, and skill
account. It is a control law suitable for implementation in remains unknown. The identification of general principles of
unstructured environments. sensorimotor interaction between human partners may also be
applicable to an alternative perspective of human-robot-
A cascade control algorithm, based on the cooperative
interaction in rehabilitation. While the application of human-
manipulation, is designed for the service robot control. With
robot interaction in rehabilitation has traditionally focused on
the proposed control algorithm for cooperative manipulation,
the robot-patient interaction, future work may also wish to
the motion, internal and external forces of the object can be
consider human-robot interaction in rehabilitation from the
controlled. Using impedance relationships, a commanded
perspective of a robot interacting with a therapist.
object trajectory is converted into reference trajectories for the
motion controllers of the manipulators such that the desired While the nature and challenges presented by the
internal and contact forces can be achieved. interaction between robot and therapist are likely to be
different from those of robot-patient interactions, principles
The next step towards the implementation in the
derived from the study of human-human interaction may serve
rehabilitation would be to experimentally map the applied
to inform such interactions as well. Overall, the identification
forces during the rehabilitation training and implement it in
of guiding principles that drive human-human sensorimotor
the proposed control algorithm.
interactions and service robot design have the potential to
VII. FUTURE RECOMMETDATIONS further the design, control and use of rehabilitation robots that
can physically interact with humans in intuitive and
The desire to develop robots that can physically interact biologically inspired ways, thereby enhancing rehabilitation
with humans in intuitive and biologically inspired ways has outcomes.
revealed a vast field of human-human physical interaction that
has only begun to be studied. It may be advantageous to REFERENCES
design robots that can be optimized to the functioning of the
[1] Weisshardt F, Reiser U, Parlitz C, Verl A (2010)
human nervous system. However, much work is to be done in
Making high-tech service robot platforms available.
human-human sensorimotor interactions before any general
In: Joint International Symposium on Robotics and
principles of cooperative sensorimotor control can be firmly
German Conference on Robotics, vol 2, pp 1115-
established. As a first step, some areas of potentially fruitful
1120.
investigation to reveal principles of sensorimotor cooperation
[2] Wyrobek K, Berger E, Van Der Loos H, Salisbury J
in human-human interface that are directly relevant to the
(2008) Towards a personal robotics development
design of physical interfaces and control schemes for
platform: Rationale and design of an intrinsically safe
rehabilitation robots have been identified.
personal robot. In: Proceedings of the IEEE
Specifically, new experimental paradigms should be International Conference on Robotics and
developed that can address open questions of how motor Automation, pp 2165-2170.
redundancy, varying skill level, specific role assignment and [3] Graf B, Hans M, Schraft R (2004) Care-O-bot II -
reliance on haptic feedback play a role in the haptic cues and development of a next generation robotic home
physical interactions between individuals. In particular, more assistant. Advanced Robotics 16(2):193-205.
attention in the specific areas of sensorimotor assistance and [4] Graf B, Parlitz C, Haeagele M (2009) Robotic home
sensorimotor education could provide some guidance in the assistant Care-O-bot R3 product vision and
design of haptic interfaces and controllers for rehabilitation
innovation platform. In: International Conference on and software architecture of the bimanual mobile
Human-Computer Interaction , pp 312-320. manipulation robot hollie and its actuated upper
[5] Ott C, Eiberger O, Friedl W, Baeuml B, Hillenbrand body. In: IEEE/ASME International Conference on
U, Borst C, Albu-Schaeuer A, Brunner B, Advanced Intelligent Mechatronics, pp 286-292.
Hirschmueller H, Kielhoefer S, Konietschke R, [11] Reynolds C, Wyatt J (2011) Open source, open
Suppa M, Wimboeck T, Zacharias F, Hirzinger G standards, and health care information systems 13(1):
(2006) A humanoid two-arm system for dexterous http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3221
manipulation. In: Proceedings of the IEEE-RAS 346/
International Conference on Humanoid Robots, pp [12] Metta G, Sandini G, Vernon D, Natale L, Nori F
276-283. (2008) The icub humanoid robot: An open platform
[6] Fuchs M, Borst C, Giordano P, Baumann A, Kraemer for research in embodied cognition. In: Workshop on
E, Langwald J, Gruber R, Seitz N, Plank G, Kunze K, Performance Metrics for Intelligent Systems, pp 50-
Burger R, Schmidt F, Wimboeck T, Hirzinger G 56.
(2009) Rollin' Justin - design considerations and [13] Ha I, Tamura Y, Asama H, Han J, Hong D (2011)
realization of a mobile platform for a humanoid Development of open humanoid platform darwin-op.
upper body. In: Proceedings of the IEEE In: Proceedings of the 2011 SICE Annual
International Conference on Robotics and Conference, pp 2178-2181.
Automation, pp 4131-4137. [14] Schwarz M, Pastrana J, Allgeuer P, Schreiber M,
[7] Asfour T, Regenstein K, Azad P, Schr•oder J, Schueller S, Missura M, Behnke S (2014) Humanoid
Bierbaum A, Vahrenkamp N, Dillmann R (2006) teensize open platform NimbRo-OP. In: Lecture
ARMAR-III: An integrated humanoid platform for Notes in Computer Science, vol 8371, Springer
sensory-motor control. In: Proceedings of the IEEE- Berlin Heidelberg, pp 568-575.
RAS International Conference on Humanoid Robots, [15] Bischo_ R, Huggenberger U, Prassler E (2011) Kuka
pp 169-175. youbot - a mobile manipulator for research and
[8] Stueckler J, Behnke S (2009) Integrating indoor education. In: IEEE International Conference on
mobility, object manipulation, and intuitive Robotics and Automation, pp 1-4.
interaction for domestic service tasks. In: [16] Rijs R, Beekmans R, Izmit S, Bemelmans D (2010)
Proceedings of the IEEE-RAS International Philips experimental robot arm, user instruction
Conference on Humanoid Robots, pp 506-513. manual v1.1. Tech. Rep. APT536-09-9962, Philips
[9] Iwata H, Sugano S (2009) Design of human Applied Technologies.
symbiotic robot TWENDY-ONE. In: IEEE [17] Le Tien L, Scha_er A, Hirzinger G (2007) Mimo
International Conference on Robotics and state feedback controller for a exible joint robot with
Automation, pp 580-586. strong joint coupling. In: Proceedings of the 2007
[10] Hermann A, Sun J, Xue Z, Ruehl S, Oberlaender J, IEEE International Conference on Robotics and
Roennau A, Zoellner J, Dillmann R (2013) Hardware Automation, pp 3824-3830.
Zlata Jelačić, “Design and Control of a Domestic Service Robot” Published in International Research Journal of
Innovations in Engineering and Technology (IRJIET), Volume 3, Issue 11, pp 66-72, November 2019.
*******