See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/282955967

Path Planning and Trajectory Planning Algorithms: A General Overview

Article · March 2015

DOI: 10.1007/978-3-319-14705-5_1

CITATIONS READS

124 28,458

4 authors, including:

Paolo Boscariol Renato Vidoni

University of Padova Free University of Bozen-Bolzano
67 PUBLICATIONS   690 CITATIONS    113 PUBLICATIONS   1,189 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

E-EDU 4.0 - Engineering Education 4.0 View project

Optimization of Motion Planning and Control for Automatic Machines, Robots and Multibody Systems View project

All content following this page was uploaded by Paolo Boscariol on 04 September 2017.

The user has requested enhancement of the downloaded file.

 Path Planning and Trajectory
Planning
Algorithms: a General Overview
Alessandro Gasparetto1, Paolo Boscariol1, Albano Lanzutti2,
Renato Vidoni3
1DIEGM – Dip. di Ingegneria Elettrica, Gestionale e Meccanica –
University of Udine
Via delle Scienze, 206 - 33100 Udine (UD) - Italy
2MBP - Via Toscanini, 48/B - 46043 Castiglione delle Stiviere (MN) - Italy
3Faculty of Science and Technology - Free University of Bozen-Bolzano

Piazza Università, 1 - 39100 Bolzano - Italy

Abstract
Path planning and trajectory planning are crucial issues in the field of
Robotics and, more generally, in the field of Automation. Indeed, the trend
for robots and automatic machines is to operate at increasingly high speed,
in order to achieve shorter production times. The high operating speed may
hinder the accuracy and repeatability of the robot motion, since extreme
performances are required from the actuators and the control system.
Therefore, particular care should be put in generating a trajectory that
could be executed at high speed, but at the same time harmless for the
robot, in terms of avoiding excessive accelerations of the actuators and
vibrations of the mechanical structure. Such a trajectory is defined as
smooth.
For such reasons, path planning and trajectory planning algorithms assume
an increasing significance in robotics. Path planning algorithms generate a
geometric path, from an initial to a final point, passing through pre-defined
via-points, either in the joint space or in the operating space of the robot,
while trajectory planning algorithms take a given geometric path and
endow it with the time information. Trajectory planning algorithms are
crucial in Robotics, because defining the times of passage at the via-points
influences not only the kinematic properties of the motion, but also the
dynamic ones. Namely, the inertial forces (and torques), to which the robot
is subjected, depend on the accelerations along the trajectory, while the
vibrations of its mechanical structure are basically determined by the
values of the jerk (i.e. the derivative of the acceleration).
Path planning algorithms are usually divided according to the
methodologies used to generate the geometric path, namely:
- roadmap techniques
- cell decomposition algorithms
- artificial potential methods.
The algorithms for trajectory planning are usually named by the function
that is optimized, namely:
- minimum time
- minimum energy
- minimum jerk.
Examples of hybrid algorithms, which optimize more than a single
function, are also found in the scientific literature.
In this Chapter, the general problem of path planning and trajectory
planning will be addressed, and an extended overview of the algorithms
belonging to the categories mentioned above will be carried out, with
references to the numerous contributions to this field.

Introduction

Human activity in many sectors is nowadays supported or substituted by

robots, which range from standard robots for industrial applications to
autonomous robots for complex tasks, such as space exploration. Indeed,
the great versatility and flexibility of robots allows them to be employed in
different sectors, to perform even very diverse tasks. Referring to the
industrial environment, a robot can be defined (Sciavicco et al., 2009) as a
mechanical structure made of several rigid bodies (links) connected one to
another by means of joints. Within the robot, it is possible to identify a
structure that implements mobility, a wrist which provides dexterity, and
an end-effector which performs the task given to the robot.
Regardless of the specific mechanical structure, in all types of applications
a generic task is achieved by a robot by imposing a specific motion to the
end-effector. This motion may be free or bound: the former case applies if
the end-effector does not have a physical interaction with the environment,
while the latter case applies if the end effector interacts with the
environment by exchanging forces and/or torques.
The input of the control system of the robot is generally given by the law of
motion, which is generated by a dedicated module for motion planning.
Such motion planning module can operate off-line, by using a knowledge of
the robot and the environment which is given a priori, or can operate on-
line: in this case, suitable sensors must be employed to monitor the robot
motion and enable the control system to adjust the movements in real time.
Ultimately, controlling the robot means determining the forces and torques
that the actuators must develop at the joints, so as to ensure that the
reference trajectories are properly followed. However, this problem turns
out to be very complex, because a robot is an articulated structure, so the
motion of a single link arm affects the other links. Mathematically, this is
expressed by the fact that the dynamic equations of a robot (with the
exception of Cartesian structures), contain some terms due to the coupling
effects between different links.
In most cases, robot controllers are based on closed loops, driven by the
error between the reference and the actual position, which allows to achieve
the accuracy required to the robot in executing the planned trajectory. In
the case that, during a manipulation task, there is contact between the end-
effector and the environment, the control problem is further complicated
because not only the motion, but also the forces exchanged in the
interaction should be monitored and controlled.
In this Chapter, we will focus on the path planning and trajectory planning
problems, which constitute the two main parts of the general motion
planning problem. The interest for such topics is dramatically increasing,
because operations at high speed are required to robots in the modern
automatic systems; hence, smooth motions should be planned (where
smooth means that such motions must avoid excessive values of
accelerations of the actuators, as well as vibrations of the mechanical
structure).
Many algorithms have been proposed, both for path planning and for
trajectory planning, in the scientific literature of the robotic domain. The
aim of this Chapter is to provide a general overview of such algorithms,
which have been subdivided into suitable categories

Path planning

Path planning is a merely geometric matter, because it is defined as the

generation of a geometric path, with no mention of any specified time law.
On the other hand, trajectory planning consists in assigning a time law to
the geometric path. In most cases, path planning precedes trajectory
planning; however, these two phases are not necessarily distinct; for
instance, if point-to-point trajectories are considered (i.e. only the initial
and final positions are specified), the two problems may be solved at the
same time.
In this section the analysis of available works in literature deals with the
case of systems without non-holonomic constraints.
Different types of paths are possible, depending on the specific case. For
instance, for industrial manipulators, the standard path is usually defined
by the geometry of the task, which is defined in a static way. In more
advanced applications, or for robots operating in dynamic environments,
some extra features, such as the need for automatic obstacle avoidance,
may be added.
In applications of advanced robotics, the problem of path planning is
definitely very challenging, especially for robots characterized by a large
degree of autonomy or for robots that must operate in hostile environments
(space, underwater, nuclear, military, etc.).
The definition of the path planning problem is very straightforward: “find a
collision-free motion between an initial (start) and a final configuration
(goal) within a specified environment”. The simplest situation is when the
path is to be planned in a static and known environment; however, more
generally, the path planning problem can be formulated for any robotic
system subject to kinematic constraints, in a dynamic and unknown
environment.
Much work can be found the robotic literature, dealing with path planning.
The first definitions and algorithms date back to the 1970’s. In (Latombe,
1991) a complete overview of the path planning techniques can be found.
An overview of many techniques cited in this work can be found also in the
classic book (Choset et.al., 2005) or in the recent book (Jing, 2008). Other
useful reviews of path planning techniques are (Kazemi et.al, 2010;
Kunchev et.al.,2006).
Some basic definitions are needed to introduce the path planning problem,
namely: the configuration space (C-space), the space of free configurations
(C-free) and the obstacles’ representation in the C-space (C-obs).
The configuration space is the space of all possible robot configurations,
where a configuration q is the specification of position and orientation of
the robot A with respect to a fixed reference frame FW. Referring to Fig. 1,
the C-space of the robot A is R3, since the configuration of A is specified by
the origin of FA with respect to FW, and by its orientation.
For an articulated robot (Fig. 2), the C-space is given by its joint space (in
this case, R2). The C-obs is given by the image of the obstacles in the C-
space, and the C-free is defined as {C-space - C-obs}.
 B2
B1

Fig. 1. Mobile robot in a 2-dimensional space with obstacles

Path planning algorithms are usually divided in three categories, according

to the methodologies used to generate the geometric path, namely:
- roadmap techniques
- cell decomposition algorithms
- artificial potential methods.
 360
qstart
270 Cobs

180
Cfree

 90

qgoal
0 
45 90 135 180

Fig. 2. C-space, C-free and C-obs for an articulated robot with two joints

Roadmap techniques

The roadmap techniques are based upon the reduction of the N-

dimensional configuration space to a set of one-dimensional paths to
search, possibly on a graph.
In other words, this approach maps the free space connectivity into a
system of one-dimensional curves (the roadmap) in the C-free space or in
its closure. The roadmap R thus obtained contains a set of paths: hence, the
path planning consists in linking the initial and final configurations to R. In
this way a feasible path between the two configurations is found.
It is very natural to associate a graph to the roadmap and to define some
optimality index (e.g. the Euclidean length): the graph can then be searched
in order to get the optimal solution to the path planning problem (in most
cases, this is represented by the shortest path).
Fig. 3.Visibility graph

Fig. 3 represents the so-called visibility graph, i.e. the graph whose nodes
are the vertices of all the obstacles in the configuration space. Searching the
graph would lead to get the shortest Euclidean path in the C-space. The
nodes of the graphs indicate point locations, while edges represent visible
connections between the nodes. Grey areas indicate obstacles to be avoided.
The concept of visibility graph, which represents a milestone in the
literature related to path planning, was introduced by Lozano-Pérez
(Lozano Pérez & Wesley, 1979; Lozano-Pérez, 1983).
Another kind of roadmap algorithms are those based on Voronoi diagrams,
which are defined as a way to divide the space into regions having the
following characteristic: given a set of points {p1, … pn}, each point
belonging to the i-th region is closer to pi than to any other pj ≠ pi. This
approach is dual to that based on the visibility graph, because the Voronoi
diagrams enable one to obtain a path lying at the maximum distance from
the obstacles, whereas the visibility graph generates a path that passes as
close as possible to the obstacle vertices.
Fig. 4 shows some path generated by using Voronoi diagrams. The three
squares in the diagram represents obstacles, while the blue lines are the set
of points equidistant from at least two obstacles. Therefore the paths
defined with this technique are designed to be as far away as possible from
nearby obstacles. Examples of path planning algorithms may be found in
(Canny & Donald, 1988), in (Takahashi & Schilling, 1989) and in (Garrido
et. al., 2011).
Fig. 4. Paths resulting from Voronoi diagrams

Cell decomposition methods

According to the cell decomposition methods, the free space of the robot is
subdivided into several regions, called cells, in such a way that a path
between any two configurations lying in the same cell is straightforward to
generate. It is then natural to define a so-called connectivity graph, which
represents the adjacency relations between cells. Namely, the nodes of the
graph represent the cells extracted from the free space, and there is an arch
between two nodes are connected if and only if the corresponding cells are
adjacent. The path planning problem is, again, turned into a graph
searching problem, and can therefore be solved using graph-searching
techniques.
Fig. 5. Exact cell decomposition: (a) subdivision of space into numbered
polygons, (b) connectivity graph, (c) regions to be crossed, (d) path

Fig. 5 illustrates the procedure described above, which is named exact cell
decomposition technique, because the union of the cell represents exactly
the free space. In some cases, an exact computation of the free space is not
possible or convenient. Approximate cell decomposition methods must
therefore be employed. Fig. 6 shows how these techniques work:
- the whole C-space (assumed 2-dimensional) is divided into four cells;
- the algorithm checks if each cell is completely empty, completely full or
mixed (such words obviously refer to the occupancy by the obstacles);
- each mixed cell is in turn divided into four subcells, and the algorithm is
recursively applied to check the status of every subcell and recursively
divide each mixed subcell into four sub-subcells.
The graph that may be naturally associated to the approximate cell
decomposition is a tree, named quadtree for 2-dimensional spaces (Fig. 7),
octree for 3-dimensional spaces (Fig. 8), 16-tree for 4-dimensional spaces,
and so forth.

Fig. 6. Approximate cell decomposition

Mixed cell
Full cell
Empty cell
Fig. 7. Quadtree

Mixed cell
Full cell
Empty cell

Fig. 8. Octree

Artificial potential methods

The artificial potential methodologies are a different approach to the path

planning problem. The basic idea is to consider the robot in the
configuration space as a moving point subject to a potential field generated
by the goal configuration and the obstacles in the C-space: namely, the
target configuration produces an attractive potential, while the obstacles
generate a repulsive potential. The sum of these two contributions is the
total potential, which can be seen as an artificial force applied to the robot,
aimed at approaching the goal and avoiding the obstacles. Thus, given any
configuration during the robot motion, the next configuration can be
determined by the direction of the artificial force to which the robot is
subjected. This normally represents the most promising direction of motion
in terms of free path. An example of the application of the artificial
potential method is shown in Fig. 9.
The artificial potential method was originally conceived by Khatib (Khatib,
1985) and further developed by Volpe (Volpe, 1990; Volpe & Khosla, 1990).
Such a technique can find applications in many fields, because it can be
successfully implemented online, thus moving the obstacle avoidance
problem from the higher (and slower) level of path planners to the lower
(and faster) level of online motion controllers. This implies that the good
features of the artificial potential methods, especially the reactivity to
environment changes, duly detected by the robot sensors, enable the robot
controller to manage unexpected workspace changes in a fast way.
However, the artificial potential methods are intrinsically affected by a
major problem, namely the presence of local minima, where the robot may
find itself trapped. In order to overcome this problem, several solutions
have been proposed: for instance, using potential functions which do not
have local minima (Koditschek, 1992; Kim & Khosla, 1992; Connolly &
Burns, 1990; Connolly & Grupen, 1992). Such functions are called
navigation functions.
In (Guldner & Utkin, 1995) and in (Ge & Cui, 2000) alternative applications
of the artificial potential method are presented.
Another approach to solve the path planning problem is found in
(Barraquand & Latombe, 1991), where a special kind of planners, named
RPP (Random Path Planners), is proposed: local minima are avoided by
combining the concepts of artificial potential field with random search
techniques. Albeit with some limitations, RPP proved to be able to solve
path planning problems for robots with a high number of degrees of
freedom, with reasonable computation times.
Other examples of RPP can be found in (Caselli & Reggiani, 2000; Caselli et
al., 2001; Caselli et al., 2002).

Fig. 9. The artificial potential method

Alternative approaches to path planning

A possible alternative approach, which had remarkable results in very
complex path planning problems, is given by the Probabilistic Roadmap
Planners (PRM). It is a technique which employs probabilistic algorithms,
such as random sampling, to build the roadmap. The most important
advantage of PRM is that their complexity do not strictly depend on the
complexity of the environment and on the dimension of the configurations
space. The basic idea is to consider a graph where the nodes are given by a
set of random configurations in the C-free. A local planner can then try to
connect these configurations by means of a path: if a path is found, a new
node is added to the graph. In this way the graph reflects the connectivity of
the C-free. In order to find a path between two configurations, these
configurations are added to the graph, then a graph search is performed in
order to find a feasible path. Given the probabilistic nature of the
algorithm, post-processing is often necessary to improve the quality of the
path. PRM algorithms have been successfully applied to robotic
manipulators with up to 16 degrees of freedom. Examples of PRM can be
found in (Amato & Wu, 1996; Hsu et al., 2002; Nissoux et al., 1999; Clark &
Rock, 1999).
There are some examples (Donald & Xavier, 1990; Fraichard & Laugier,
1993) of path planners that take into account kinematic and dynamic
constraints of the robot, in addition to the pure geometric problem of
obstacles avoidance. This problem is referred to as kinodynamic motion
planning. Kinodynamic and nonholonomic motion planning can be
handled by the Rapidly-exploring Random Tree (RRT) method (LaValle,
2006). This method allows to search non-convex high-dimensional spaces
by randomly building a space-filling tree.
Another important version of the general problem is given by path planning
in presence of mobile obstacles. As it can be easily understood, this kind of
problem results very complex with respect to the basic version. This
approach is used, for instance, in (Fiorini & Schiller, 1996) and in
(Fraichard, 1999).
A general overview of the path planning problem can be found in (Kumar et
al. 1999) and in (Gupta & Del Pobil, 1998), where the most important
results achieved in the field of path planning, including PRM and RPP
techniques, are reported. In (Gupta & Del Pobil, 1998) it is claimed that all
the methodologies that have proven to be practically usable for path
planning are based on a discretization of the configuration space. There are
two crucial requirements in order to ensure an efficient implementation of
path planning methodologies, namely: the efficiency of collision detection
algorithms and the efficiency of graph searching techniques.

Trajectory planning
Solving the trajectory planning problem means generating the reference
inputs for the control system of the robot, so as to ensure that the desired
motion is performed. Usually, the algorithm employed for trajectory
planning takes as inputs the path generated by the path planner, as well as
the kinematic and dynamic constraints of the robot. The output of the
trajectory planning module is given by the trajectory of the joints, or of the
end-effector, in form of a sequence of values of position, velocity and
acceleration.
The geometric path is normally defined in the operating space of the robot,
because the task to be performed, as well as the obstacles to avoid, are
described in the operating space more naturally than in the joint space.
Thus, planning the trajectory in the operative space means generating a
sequence of values that specify the position and orientation that the end-
effector of the robot must assume at every time interval. Planning the
trajectory in the operating space is usually done when the motion follows a
path with specific geometric characteristics defined in the operating space;
in this case, the path can be specified in an exact form (i.e. taking the
original path), or in an approximate form, by allocating some path points
and connecting them by means of polynomial sequences. However, in most
cases the trajectory is planned in the joint space of the robot because, since
the control action on the manipulator is made on the joints, planning in the
operating space requires a kinematic inversion to transform the end-
effector position and orientation values into the joint values.
In order to plan a trajectory in the joint space, first a sequence of via-points
should be extracted from the desired end-effector path, then a kinematic
inversion is to be performed to get the corresponding values of the robot
joints. The trajectory is then generated in the joint space by means of
interpolation functions, taking into account the kinematic and dynamic
limits imposed to the robot joints (in terms of position, velocity,
acceleration and jerk). Normally, this way of planning the trajectory can
also avoid the problems involved in moving near singular configurations,
and can efficiently deal with the possible presence of redundant degrees of
mobility. The main drawback of planning a trajectory in the joint space is
given by the fact that the execution of a motion planned in the joint space is
not so straightforward to predict in the operative space, due to the
nonlinearities introduced by the direct kinematics. However, no matter if
the trajectory is planned in the operating space or in the joint space, it is
crucial that the laws of motion resulting from the planning do not generate
forces and torques at the joints that are not compatible with the given
constraints: in this way the possibility of exciting mechanical resonance
modes can be greatly reduced. For this reason, the planning algorithms
must output smooth trajectories, i.e. trajectories represented by a curve
whose derivatives are continuous up to a certain order. In particular, it is
highly desirable to ensure the continuity of the accelerations of the joints,
in order to get trajectories with a limited jerk, because limiting the jerk is
crucial in order to reduce the vibrations induced to the robot (which may
lead to considerable wear of the mechanical structure), as well as to avoid
the excitation of the resonance frequencies of the robot. The vibrations
caused by non-smooth trajectories may seriously damage the actuators and
degrade the tracking performance of the trajectory. Furthermore, low-jerk
trajectories can be executed faster and with a higher accuracy as
demonstrated in (Barre et al., 2005). In addition, there are some
applications where abrupt motions can jeopardize the quality of the work or
constitute a risk to the human operators working near the robot.
In order to classify the different trajectory planning methodologies into
categories, it is useful to consider that a trajectory is usually planned after
some optimality criterion has been set. The most significant optimality
criteria that can be found in the literature are:

• minimum execution time;

• minimum energy (or actuator effort);
• minimum jerk.

In addition to the above, hybrid optimality criteria have been proposed,

such as, for instance, time-energy optimal trajectory planning. With respect
to the minimum energy criterion, a short clarification is necessary. In most
of the cases related with trajectory planning, the term “energy” does not
correspond to a physical quantity measured in Joules, but it is defined as
the integral of squared torques: in other words, it measures the effort of the
robot actuators. However, in the robotic literature it is possible to find also
trajectory planning algorithms where the optimality index is “energy” in its
strict meaning. Actually, this is not really a problem, because in the electric
motors used on the robots, the torque can be assumed proportional to the
current, so there is a correlation between the actuators’ effort and the
energy required to the system.

Minimum execution time algorithms

The optimality criterion based on minimum execution time was the first to
be considered in trajectory planning, because short execution times are
strictly related to high productivity in automatized production plants in
industrial environments. Thus, no wonder that many papers can be found,
in the robotic literature, proposing trajectory planning algorithms aimed at
minimizing the performance index given by the execution time.
The algorithms described in (Bobrow et al., 1985) and in (Shin & McKay,
1985) are defined in the position-velocity phase plane. The basic idea of
these algorithms is to write the dynamic equation of manipulator in a
parametric form using the curvilinear abscissa s of the path as the
independent parameter. The curvilinear abscissa s (path parameter) and its
derivative s’ (pseudo-velocity) constitute the state of the system, while
the second derivative of s (i.e. the pseudo-acceleration s’’) is chosen as the
control variable. In this way, it is possible to transform the constraints
given by the nonlinear robot dynamics, as well as the constraints on the
actuators, into constraints on the control variable depending from the state
of the system. For every point on the path, the maximum admissible value
for the pseudo-velocity of the end-effector is determined from the
constraints; it is then possible to build in the position-velocity phase plane
(i.e. in the (s, s’) plane), a velocity limit curve (VLC). The optimal trajectory
is then computed by finding the admissible control that yields, for each
point of the path, the maximum velocity that does not exceed the limit
curve. The solution turns out to be in the form of a curve (named switching
curve) in the phase plane.
An alternative approach to minimum time trajectory planning consists in
using dynamic programming techniques, such as those described in (Shin &
McKay, 1986) and in (Balkan, 1998). The basic idea is to take the state
space and discretize it by building a grid of points (called state points). On
the basis of the limits set on velocity, acceleration and jerk, it is possible to
associate to each point the set of the subsequent admissible state points,
and to define the cost of each possible solution by considering the time
needed for the motion. This cost is defined by assuming a constant value of
acceleration for each step. Finally, an algorithm based on dynamic
programming generates the minimum time trajectory. Compared with the
phase plane methods, the dynamic programming methods do not require
the parameterization of the path and enables to choose an arbitrary
performance index. Therefore, such algorithms may be used as a general
technique for trajectory optimization. On the other hand, the phase plane
approach turns out to be very efficient in terms of computational load;
moreover, it may also be used for on-line trajectory planning, as in (Croft et
al., 1995) and in (Pardo-Castellote & Cannon, 1996).
A model-based approach is used to maximize the speed of industrial robots
by obtaining the minimum-time trajectories that satisfy various constraints
commonly given in the application of industrial robots in (Kim et. al. 2010).
Conventional trajectory patterns, such as trapezoidal velocity profiles and
cubic polynomial functions.
The algorithms described above produce trajectories with discontinuous
accelerations and joint torques, because the dynamic models used consider
the robot members as perfectly rigid and do not take into account the
actuator dynamics. Neglecting the link flexibility and the actuator dynamics
normally leads to some undesired effects. First, in reality the robot
actuators cannot generate discontinuous torques: this causes the joint
motion to be delayed with respect to the reference trajectory. This accuracy
in trajectory following is thus greatly reduced, and the tracking controller
has to be often activated during the execution of the trajectory. Moreover,
each switching of the actuators may cause the so-called chatter
phenomenon, i.e. high frequency oscillations inducing vibrations of the
mechanical structure of the robot. This obviously results in wearing of the
mechanical components and in a decrease of the accuracy in trajectory
following. Again, the tracking controller is activated more frequently and
the actuators are further stressed. Another undesired effect resulting from
an inaccurate model is that, since the time-optimal control requires
saturation of at least one robot actuator at any time instant, it is impossible
for the controller to correct the tracking errors arising from disturbances or
modelling errors.
In (Constantinescu, 1998) and in (Constantinescu & Croft, 2000) a possible
solution to these kind of problems is proposed: in these works, the phase
plane method is used, together with a limitation set on the torque
variations (actuator jerks). The proposed algorithm takes the pseudo-jerk,
defined as the third derivative of the curvilinear abscissa, as the control
variable: a dynamic equation of the third order is thus obtained. The
experimental results presented in (Constantinescu, 1998) show that, if
some upper bound is set on the pseudo-jerk, time-optimal trajectories can
be practically obtained by simply employing a conventional PID controller.
This proves the correlation between accuracy in trajectory following and
low values of jerk.
A different way to limit the torque variations is to consider in the objective
function not only the execution time, but also an energy contribution: for
instance, in (Shiller, 1996) the integral of squared torques along the whole
trajectory is taken into account. The experimental results presented in
(Shiller, 1996) show that the increase of the overall motion time is
compensated by a greater accuracy in trajectory following, even if
conventional PD controllers are used. This results in a reduction of actuator
stresses, with obvious advantages in the total lifetime of the electro-
mechanical components of the robot.
It is possible to approach the problem of minimum-time trajectory
planning by defining a priori the primitives of the motion, i.e. the curves
that define the trajectory in the joint space. Such curves must be smooth
functions, so that the control signals and, consequently, the torque signals
at the actuators, result also smooth functions. The most common situation
is that in which the path is specified using a limited number of via-points:
the solution is then given by spline interpolation. In the literature, several
methodologies are proposed to compute time-optimal trajectories for robot
manipulators based on optimization of splines, whose order may be three
(cubic splines) or higher. The main differences among these techniques are:
• the type of constraints considered (either kinematic or dynamic);
• the algorithm used to compute the optimal trajectory;
• the possibility to extend the optimization problem, by taking into
account other optimization criteria, in addition to the minimum time.
The distinction based on the type of constraints can be considered the most
important. It can be extended to any type of trajectory planning algorithm,
so that the two categories of kinematic trajectory planning and dynamic
trajectory planning can be defined. The kinematic trajectory planning
algorithms take as their input upper (sometimes also lower) bounds on
velocity, acceleration and jerk. In most cases such bounds are considered
constant. The dynamic trajectory planning algorithms consider the
dynamic model of the robot and define an optimization problem taking into
account dynamic constraints, such as bounds on the actuator torques, or on
the actuator jerks, defined as the variation of the torques. In some cases,
kinematic constraint (typically the velocity) are also considered. Both
approaches have pros and cons: the kinematic trajectory planning has its
main advantage in the simplicity and in the lower computational load; on
the other hand, the dynamic trajectory planning features a better capacity
to use the robot actuators. In other words, the kinematic methods are based
on a simplified computational model that yields a non-optimal use of the
robot actuators, although in most cases reasonably good trajectories are
planned. Dynamic methods are based on a more accurate model and
therefore produce better solutions, but at the cost of a heavier
computational load, since they have to deal with non-trivial issues, such as
identification of the dynamic parameters of the robot, or the efficiency in
implementing efficient algorithms to solve the robot dynamic equations.
An interesting example of an algorithm based on the inverse dynamic of a
parallel robot is given by (Carbone et al., 2008). In this work, a multi-
objective optimisation problem is formulated and a dedicated genetic
algorithm is employed to find an optimal trajectory based upon spline
functions.
Splines function are therefore used as trajectory primitives in order to
ensure the continuity of the acceleration. Another example can be found in
(Lin et al., 1983), where a nonlinear optimization problem is set, namely
the computation of the value of the time intervals between the via-points,
so as to minimize the total execution time of the trajectory subject to
kinematic constraints. The technique is based upon an unconstrained
optimization algorithm named FPS (Flexible Polyhedron Search), in
combination with an algorithm called FSC (Feasible Solution Converter),
which converts the solutions that are not physically feasible (i.e. that are
not compatible with the kinematic constraints) into feasible ones, by
implementing a suitable time scaling of the trajectory generated by the FPS
algorithm. In (Wang & Horng, 1990), the same optimization algorithm
presented in (Lin et al., 1983) is used, but instead of cubic splines, cubic B-
splines are taken as primitives of motion.
The algorithms described above produce a local optimal solution, while
other minimum-time trajectory planning methods output a global optimal
solution. Piazzi and Visioli use interval analysis to calculate a minimum-
time trajectory subject to kinematic constraints at the joints. Such
kinematic constraints are on the maximum value of velocity, acceleration
and jerk. In (Piazzi & Visioli, 1998) they extend the results already
presented in (Piazzi & Visioli, 1997a) and (Piazzi & Visioli, 1997b). The
simulations presented in (Piazzi & Visioli, 1998) showed an improvement
of 18% of the total execution time with respect to the results yielded by a
local optimization algorithm.
In (Guarino Lo Bianco & Piazzi, 2001a) and in (Guarino Lo Bianco & Piazzi,
2001b) a global optimization method is presented, which combines a
stochastic technique, such as a genetic algorithm, with a deterministic
procedure based on interval analysis. The proposed technique can be
applied to solve general global optimization problems where semi-infinite
constraints are defined. In (Guarino Lo Bianco & Piazzi, 2001a) this
algorithm is applied to the problem of minimum-time trajectory planning
with specific kinematic and dynamic constraints: namely, the trajectories,
represented by cubic splines, are subject to restrictions on the maximum
actuator torques, as well as on the linear and angular velocities of the end-
effector in the operating space. It is remarkable that, differently from usual,
in (Guarino Lo Bianco & Piazzi, 2001a) the velocity constraint is not
imposed in the joint space, but in the operating space of the end-effector.
A composition of polynomial functions of different orders are used in
(Boscariol et. al. 2012b; Boscariol et. al. 2012b) to obtain jerk continuity
along a trajectory planned from a set of pre-defined via-points, obtaining a
global minimum time solution.
Another example of minimum-time trajectory planning for robotic
manipulators can be found in (Cao & Dodds, 1994). In this case the
objective function is made of two terms: the first term takes the squared
values of the optimization variables (i.e. of the time intervals between the
via-points), while the second term is the sum of the squared accelerations
computed at the via- points. The introduction of this second term has the
effect of increasing the trajectory smoothness with respect to a pure
minimum-time approach. The optimization is performed by using the DFP
(Davidon-Fletcher-Powell) algorithm, which does not consider the
kinematic bounds, therefore performing an unconstrained minimization.
The solution obtained by means of the DFP algorithm is then subjected to a
procedure of time-scaling, until the more restrictive kinematic bound has
been saturated. The resulting trajectory, although respecting the limits on
velocity, acceleration and jerk, is sub-optimal with respect to time.
In (Dong et al., 2007) a technique for determining time-optimal path-
constrained trajectories subject to velocity, acceleration and jerk
constraints, acting on both the robot actuators and on the task to be
executed, is presented. The solution of the optimization problem is based
upon a hybrid optimization strategy, which takes into account the path
description, the kinematic model of the robot and constraints defined by
the user. The resulting trajectories are optimal with respect to time, but not
with respect to smoothness.
In the work (Liu et. al. 2013) a combination of spline functions up to the
seventh order are used together to achieve minimum time solutions with
velocities, acceleration and jerk bounds. Other examples of minimum-time
algorithms subject to kinematic constraints may be found in (Van Dijk et
al., 2007; Dongmei et al., 2006; Tangpattanakul & Artrit, 2009;
Tangpattanakul et. al., 2010; Joon-Young et al., 2010). In (Rubio et al.,
2012) the minimum-time trajectory problem is solved under kinematic and
dynamic constraints, i.e. teorque, power, jerk and energy, taking into
account both the robot dynamics and the obstacle presence.

Minimum energy algorithms

As already remarked, the minimum-time trajectory planning algorithms

received a lot of consideration in the robotic literature, mainly because of
the strong industrial interest to reduce the length of the production cycles.
However, the minimum-time optimization criterion is not the only one that
can be considered: other criteria are definitely more suitable for different
needs and requirements.
The trajectory planning based on energetic criteria is interesting under
many aspects. On one hand, it generates smooth trajectories which are
easier to track, and reduce the stresses induced to the actuators and to the
mechanical structure of the robot. On the other hand, this optimization
criterion enables one to better comply with energy saving requirements,
which are driven not only by mere economic considerations, but may be
imposed by specific applications in which the energy source is limited by
technical factors, such as robotic applications for outer space, for
underwater exploration or for military tasks.
A classical example of minimum-energy trajectory planning algorithm is
contained in (Martin & Bobrow, 1999), where a trajectory is optimized with
respect to energy taking into account constraints on the motion of the end-
effector, as well as the physical limits of the joints. The proposed objective
function is the integral of squared torques. The trajectories are expressed
by cubic B-splines and, by exploiting some property of the convex hull, it is
possible to transform the joint limits into some limits set on the
optimization parameter, which are the control points of the B-splines. The
resulting motion thus minimizes the effort of the actuators.
In (Balkan, 1998) and in (Shiller, 1996) some techniques for optimal
trajectories planning, with respect to energy and time, are described: the
function to optimize is made of two terms, the first related to the execution
time, the second related to the energy consumption. Such algorithms are
intended to reduce the stresses of the actuators and to facilitate the
trajectory tracking. In (Shiller, 1996), the integral of the squared torques
along the trajectory is considered in the objective function, while in
(Balkan, 1998) the function of total energy is considered.
Other examples of optimized trajectories, with respect to energy as well as
to time, are presented in (Saramago & Steffen, 1998; Saramago & Steffen,
2000; Saravan et al., 2009; Verscheure et al., 2008; Xu et al., 2009). In
(Saramago & Steffen, 1998) the Authors consider a trajectory
parameterized by cubic splines, subject to kinematic constraints set on the
maximum value of velocity, acceleration and jerk, and to dynamic
constraint given by the maximum torque applicable to the joints. In
(Saramago & Steffen, 2000) the same Authors consider a trajectory
parameterized by cubic B-splines, where the physical limits of the joints are
added to the torque and kinematic constraints. The objective function
includes also an additional term (penalty function), in order to avoid
mobile obstacles expressed as spherical or hyperspherical safety zones. In
(Saravan et al., 2009), two strategies for offline 3-dimensional optimal
trajectory planning of industrial robots, in presence of fixed obstacles, are
presented. In (Verscheure et al., 2008), a nonlinear change of variables is
employed to convert the time-energy optimal trajectory planning problem
into a convex control problem based on only one state variable. In (Xu et
al., 2009), a methodology based on the minimization of an objective
function which considers both the total execution time and the total energy
spent along the whole trajectory is presented; the via-points of the
trajectory are interpolated by means of cubic splines. Kinematic and
dynamic constraints, in terms of upper bounds on velocity, acceleration,
jerk and input forces and torques are also considered. It is worth noting
that in algorithms such as the one presented in (Shiller, 1996) the energy
term is added in order to produce trajectories which result slower but
smoother with respect to those generated by minimum-time trajectory
planning algorithms; on the other hand, in approaches such as the one
presented in (Saramago & Steffen, 2000) the objective function is primarily
designed to minimize the energy and to plan trajectories with no regard to
the execution time.
Recently, due to the development and installation of energy recovery and
redistribution devices in robotic systems, the minimum-energy topic has
gained new interest among the research community, e.g. (Pellicciari et al.,
2013; Hansen et al., 2012).

Minimum jerk algorithms

The importance of generating trajectories that do not impose

discontinuities of the actuator torques at the robot joints has already been
remarked; for instance, in (Constantinescu, 1998) and in (Constantinescu &
Croft, 2000) this result is obtained by imposing upper bounds to the rate of
change of the actuator torques. However, this kind of approach requires the
computation of the third order dynamics of the robot.
An alternative method to obtain smooth profiles of the actuator torques is
based on the idea of limiting the jerk, defined as the time derivative of the
acceleration. Indeed, the torque variations depend upon the dominant term
of the matrix of inertia multiplied by the vector of the joint jerk. Thus, some
trajectory planning methods take the jerk as the variable to be minimized,
in order to obtain smooth trajectories. The minimization of the jerk yields
positive results, such as: reduction of the error during the trajectory
tracking phase, reduction of the excitation of resonance frequencies,
reduction of the stresses induced to the mechanical structure of the robot
and to the actuators.
This results in a natural and coordinated motion: indeed, some studies
suggest that the movements of the human arm satisfy an optimization
criterion based upon the minimization of the jerk, or of the torque
variations (Simon, 1993). The minimum-jerk trajectory planning for
robotic manipulators are an example of optimization based on physical
criteria which mimic the human ability to produce natural movements
(Bobrow et al., 2001).
In (Kyriakopoulos & Saridis, 1998) the analytical solution of a trajectory
planning problem for a point-to-point path, based on a minimum-jerk
optimization criterion, is presented. The optimization, performed by
applying Pontryagin’s principle, involves two objective functions, namely:
the maximum absolute value of jerk (minimax approach) and the time
integral of the squared jerk.
In some cases, the total execution time of the trajectory is not imposed, so it
can be chosen so as to comply with the kinematic limits on velocity and
acceleration. However, most of the minimum-jerk algorithms that can be
found in the robotic literature consider an execution time imposed a priori.
In (Simon, 1993), the integral of the squared jerk is minimized along the
executed trajectory. In order to have a trajectory with a smooth start and
stop, the values of velocity, acceleration and jerk are set to zero at the first
and at the last via-points. The proposed algorithm is based upon a
stochastic optimization technique performed by means of neural networks.
The algorithm does not ensure the exact interpolation of intermediate
nodes, but allows a tolerance, which can be set by tuning appropriate
weights. This does not constitute a problem in cases where the exact
interpolation is not needed, but just the passage in the neighbourhood of
the via-points is required. The main limitation of this technique is that the
resulting trajectories are not analytical functions, but are numerically
defined.
Another approach is contained in (Simon & Isik, 1993), where the
interpolation of the via-points is performed by means a trigonometric
spline, thus ensuring the continuity of the jerk. The algorithm assumes that
the time interval between the via-points is known and constant, and takes
as input the values for the velocity, the acceleration and the jerk, at the first
and at the last via-points (such values are typically all set to zero). There are
some advantages in using trigonometric splines to interpolate the trajectory
via-points, for instance the property of locality: namely, if a via-point is
changed, it is not necessary to recalculate the whole trajectory, but only the
two splines that are connected to the via-point need to be recomputed. This
property allows fast computation, thus making it possible to implement
obstacle avoidance procedures in real time. The most significant aspect, in
terms of trajectory optimization, is that parameterizing the trajectory
allows some degrees of freedom, namely those given by the values of the
first three derivatives (velocity, acceleration and jerk) at the intermediate
via-points. Such values can be adjusted in order to minimize an objective
function, such as the time integral of the squared jerk. The optimization
presented in (Simon & Isik, 1993) is not bounded, since no kinematic limits
are imposed, and yields a closed form solution, thus not requiring iterative
minimization procedures.
In (Piazzi & Visioli, 1997c) and in (Piazzi & Visioli, 2000) an algorithm
based on interval analysis is presented. This technique seeks the minimum
of the maximum absolute value of the jerk along a trajectory whose
execution time is imposed a priori. It is therefore a so-called minimax
approach bounded on the trajectory execution time. The trajectories
primitives are cubic splines and the intervals between the via-points are
computed, so as to obtain the lowest maximum absolute jerk value. In
(Piazzi & Visioli, 2000) the Authors present a comparison with the method
based on trigonometric splines (Simon & Isik, 1993), reporting the highest
values of the jerk, of the torques and of the torque variations. The
simulation, which calculates the robot dynamics using the MatLab™
Robotics Toolbox, highlights the efficiency of the minimax algorithm with
respect to other approaches.

Hybrid optimization approaches

Optimal trajectory planning with respect to time, energy and jerk has been
discussed in the foregoing. Hybrid optimization approaches have also been
proposed in the robotic literature. For instance, in order to get the
advantages of the jerk reduction while executing fast trajectories, hybrid
time-jerk optimal techniques are proposed, for instance (Huang et al.,
2006; Petrinec & Kovacic, 2007; Gasparetto & Zanotto, 2007; Gasparetto &
Zanotto, 2008; Gasparetto et al., 2010; Boscariol et. al. 2011, Gasparetto et.
al. 2012). These algorithms differ from the primitives used to interpolate
the path, or from the optimization procedures implemented.
In (Gasparetto & Zanotto, 2007; Gasparetto & Zanotto, 2008; Gasparetto et
al., 2010: Boscariol et. al. 2011; Gasparetto et. al. 2012) a minimum time-
jerk trajectory planning technique is described, based upon two algorithms
aimed at the minimization of an objective function, which is designed so as
to ensure fastness in execution and smoothness of the trajectory at the
same time. Such an objective function is composed of a term which is
proportional to the total execution time and of a term which is proportional
to the integral of the squared jerk along the path. The proposed algorithm
enables one to define constraints on the robot motion before the execution
of the trajectory. The constraints are expressed in form of upper bounds on
the velocity, acceleration and jerk values of all robot joints. In this way, any
physical limitation of the real robot can be taken into account when
planning the trajectory. Unlike most jerk-minimization methods, this
technique does not ask for an a priori setting of the total execution time.
In (Lombai & Szederkenyi, 2008) and in (Lombai & Szederkenyi, 2009),
the methodology is extended by taking into account also the power
consumption of the actuators and physical limits of the joints. In this way,
the technique becomes a time-jerk-energy planning algorithm.

Several objectives are taken into account in the work (Khouki et. al, 2009):
in particular minimum elctrical and kinematic energy, mimum time and
maximum maniuplability are obtained with the solution of a single
optimization problem.

Minimum effort trajectories planned trough model-based approaches are

presented in (Boscariol et. al, 2013a; Boscariol et. al, 2013b; Boscariol et. al,
2013b). The first one includes bounds on jerk, while second one has
bounded joint speed. The work (Boscariol et. al, 2013c) introduces the
novel topic of robustness in trajectory planning algorithms. Such approach
allows to increase the tolerance of the resuting trajectory to the inevitable
mismatches between the dynamic model used for the planning and the
actual robot dynamics.

The problem of finding minimum time-effort trajectories for motor-driven

parallel platform manipulators, subject to the constraints imposed by the
kinematics and dynamics of the manipulator structure is the topic of the
paper (Chen et. al., 2011). Computational efficiency is obtained trough a
hybrid scheme comprising the particle swarm optimization method and the
local conjugate gradient method. Also in (Chen et. al., 2012) a constrained
multi-objective genetic algorithm (MOGA) based technique is proposed to
address this problem for a general motor-driven parallel kinematic
manipulator. The planning process is composed of searching for a motion
ensuring the accomplishment of the assigned task, minimizing the traverse
time, and expended energy subject to various constraints imposed by the
associated kinematics and dynamics of the manipulator.
All the trajectory planning methods introduced above are applicable to
rigid link robot, with either serial or parallel kinematic configurations.
However, it is worthwhile to mention that also cable-driven robots
application are gaining a growing interest in robotics. Among the
advantages brought by this class of manipulators, low overall mass and
high stiffness make them very advantageous in many applications. On the
other hand, the fact that they often require using actuation redundancy and
that they operation must avoid cable interference (Williams and Gallina,
2002), has led to the development of trajectory planning algorithms
specifically designed for them. The work (Trevisani, 2010) presents a
method to compute trajectories for underconstrained parallel robot that
ensures positive and bounded cable tension, while in (Trevisani, 2013) a
similar procedure is also experimentally validated. A detailed study of the
dynamics of cable-driven parallel robot is reported in (Ismail et. al., 2013),
as a tool for developing accurate path planning algorithms. The time-
optimality of trajectories designed for cable-driven robot is the topic
covered in the works (Barnett & Gosselin, 2013) and (Bamdad, 2013).

Conclusions

In this paper, the fundamental problems of path planning and trajectory

planning in Robotics have been addressed. An overview of the most
significant methods, that can be found in the robotic literature to generate
collision-free paths, has been presented. Then, the problem of finding an
optimal trajectory given a planned path has been discussed and the most
significant approaches have been described.

References

Amato N.M., Wu Y. (1996). A randomized roadmap method for path and

manipulation planning, Proceedings of the 1996 IEEE
International Conference on Robotics and Automation, pp. 113-
120.
Balkan T. (1998). A dynamic programming approach to optimal control of
robotic manipulators, Mechanics Research Communications, vol.
25, n. 2, pp. 225-230.
Bamdad, M. (2013). Time-Energy Optimal Trajectory Planning of Cable-
Suspended Manipulators, in Cable-Driven Parallel Robots.
Springer Berlin Heidelberg, pp. 41-51
Barnett, E., & Gosselin, C. (2013, August). Time-Optimal Trajectory
Planning of Cable-Driven Parallel Mechanisms for Fully-Specified
Paths With G1 Discontinuities, ASME 2013 International Design
Engineering Technical Conferences and Computers and
Information in Engineering Conference. American Society of
Mechanical Engineers.
Barraquand J., Latombe J.C. (1991). Robot motion planning: a distributed
representation approach, The International Journal of Robotics
Research, vol. 10, n. 6, pp. 628-649.
Barre P.J., Bearee R., Borne P., Dumetz E. (2005) Influence of a jerk
controlled movement law on the vibratory behaviour of high-
 dynamics systems, Journal of Intelligent and Robotic Systems,
vol. 42, n. 3, pp. 275–293.
Bobrow J.E., Dubowsky S., Gibson J.S. (1985). Time-optimal control of
robotic manipulators along specified paths, The International
Journal of Robotics Research, vol. 4, n. 3, pp. 554-561.
Bobrow J.E., Martin B.J., Sohl G., Wang E.C., F.C. Park, Kim J. (2001).
Optimal robot motion for physical criteria, Journal of Robotic
Systems, vol. 18, n. 12, pp. 785-795.
Boscariol P., Gasparetto A., Lanzutti A., Vidoni R., Zanotto V. (2011).
Experimental validation of minimum time-jerk algorithms for
industrial robots. Journal of Intelligent & Robotic Systems, vol.
64, n. 2, pp. 197-219.
Boscariol P., Gasparetto A., Vidoni R. (2012). Jerk-Continous Trajectories
For Cyclic Tasks. Proceedings of the ASME 2012 International
Design Engineering Technical Conferences (IDETC), pp. 1-10.
Boscariol P., Gasparetto A., Vidoni R. (2012). Planning continuous-jerk
trajectories for industrial manipulators. Proceedings of the ESDA
2012 11^th Biennial Conference on Engineering System Design
and Analysis, pp. 1-10.
Boscariol P., Gasparetto A. (2013). Model-based trajectory planning for
flexible link mechanisms with bounded jerk, Robotics and
Computer Integrated Manufacturing, vol. 29, n. 4, pp. 90-99.
Boscariol P., Gasparetto A., Vidoni R., Romano A. (2013). A model-based
trajectory planning approach for flexible-link mechanisms.
Proceedings of ICM 2013 - IEEE International Conference on
Mechatronics, pp. 1-6.
Boscariol P., Gasparetto A., Vidoni R. (2013). Robust trajectory planning
for flexible robots. Proceedings of the 2013 ECCOMAS Multibody
Dynamics Conference, pp. 293-294.
Canali, F., Guarino Lo Bianco, C., & Locatelli, M. (2013). Minimum-jerk
online planning by a mathematical programming approach.
Engineering Optimization, (ahead-of-print), pp. 1-21.
Canny J., Donald B. (1988). Simplified Voronoi diagrams. Discrete &
Computational Geometry, vol. 3, n. 1, pp. 219-236.
Cao B., Dodds G.I (1994). Time-optimal and smooth constrained path
planning for robot manipulators, Proceedings of the 1994 IEEE
International Conference on Robotics and Automation, pp. 1853-
1858.
Carbone G., Ceccarelli M., Oliveira P. J., Saramago S. F., Carvalho J. C. M.
(2008). An optimum path planning for Cassino Parallel
Manipulator by using inverse dynamics. Robotica, vol. 26, n. 2, pp.
229-239.
Caselli S., Reggiani M. (2000). ERPP: an experience-based randomized
path planner, Proceedings of the ICRA ’00- IEEE International
Conference on Robotics and Automation, pp. 1002-1008.
Caselli S., Reggiani M., Rocchi R. (2001). Heuristic methods for
randomized path planning in potential fields, Proceedings of the
2001 IEEE International Symposium on Computational
Intelligence in Robotics and Automation, pp. 426-431.
Caselli S., Reggiani M., Sbravati R. (2002). Parallel path planning with
multiple evasion strategies, Proceedings of the ICRA ’02 - IEEE
International Conference on Robotics and Automation, pp. 260-
266.Chen, C. T., & Liao, T. T. (2011). A hybrid strategy for the
time-and energy-efficient trajectory planning of parallel platform
manipulators. Robotics and Computer-Integrated Manufacturing,
vol. 27, n.1, pp. 72-81.
Chen, C. T., & Pham, H. V. (2012). Trajectory planning in parallel kinematic
manipulators using a constrained multi-objective evolutionary
algorithm. Nonlinear Dynamics, vol. 67, n.2, pp. 1669-1681.
Choset H.M., Lynch K.M., Hutchinson S., A. Kantor G.A., Burgard W.,
Kavraki L.E., Thrun S. (2005). Principles of robot motion: theory,
algorithms, and implementation. MIT press.
Clark C.M, Rock S. (2001). Randomized motion planning for groups of
nonholonomic robots, Proceedings of the 6th International
Symposium on Artificial Intelligence, Robotics and Automation in
Space, pp. 1-8.
Connoly C.I, Burns J.B. (1985). Path planning using Laplace’s equation,
Proceedings of the 1985 IEEE International Conference on
Robotics and Automation, pp. 2102-2106.
Connoly C.I. Grupen R.A. (1992). On the application of harmonic functions
to robotics, Proceedings of the 1992 IEEE International
Symposium on Intelligent Control, pp. 498-502.
Constantinescu D. (1998). Smooth time optimal trajectory planning for
industrial manipulators, PhD Thesis, The University of British
Columbia, 1998.
Constantinescu D., Croft E.A. (2000). Smooth and time-optimal trajectory
planning for industrial manipulators along specified paths,
Journal of Robotic Systems, vol. 17, n. 5, pp. 233-249.
Croft E.A., Benhabib B., Fenton R.G. (1995). Near time-optimal robot
motion planning for on-line applications, Journal of Robotic
Systems, vol. 12, n. 8, pp. 553-567.
Donald B.R., Xavier P.G. (1990). Provably good approximation algorithms
for optimal kinodynamic planning for Cartesian robots and open
chain manipulators, Proceedings of the sixth Annual Symposium
on Computational Geometry, pp. 290-300.
Dong J., Ferreira P.M., Stori J.A. (2007). Feed-rate optimization with jerk
constraints for generating minimum-time trajectories,
International Journal of Machine Tools & Manufacture, vol. 47, n.
12-13, pp. 1941-1955.
Dongmei X., Daokui Q., Fang X. (2006). Path constrained time-optimal
robot control, Proceedings of the International Conference on
Robotics and Biomimetics, pp. 1095-1100.
Fiorini P., Shiller Z. (1996). Time optimal trajectory planning in dynamic
environments, Proceedings of the 1996 IEEE International
Conference on Robotics and Automation, pp. 1553-1558.
Fraichard T., Laugier C. (1993). Dynamic trajectory planning, path-velocity
decomposition and adjacent paths, Proceedings of the 1993
International Joint Conference on Artificial Intelligence, pp. 1592-
1597.
Fraichard T (1999). Trajectory planning in a dynamic workspace: a state-
time space approach, Advanced Robotics, vol. 13, n. 1, pp. 74-94.
Garrido, S., Moreno, L., & Lima, P. U. (2011). Robot formation motion
planning using Fast Marching. Robotics and Autonomous
Systems, vol. 59, n. 9, pp. 675-683.
Gasparetto A., Zanotto V. (2007). A new method for smooth trajectory
planning of robot manipulators, Mechanism and Machine Theory,
vol. 42, n. 4, pp. 455-471.
Gasparetto A., Zanotto V. (2008). A technique for time-jerk optimal
planning of robot trajectories, Robotics and Computer-Integrated
Manufacturing, vol. 24, n. 3, pp. 415-426.
Gasparetto A., Lanzutti A., Vidoni R., Zanotto V. (2012). Experimental
validation and comparative analysis of optimal time-jerk
algorithms for trajectory planning. Robotics and Computer-
Integrated Manufacturing, vol. 28, n. 2, 164-181.
Ge S.S., Cui Y.J. (2000). New potential functions for mobile robot path
planning, IEEE Transactions on Robotics and Automation, vol. 16,
n. 5, pp. 615-620.
Guarino Lo Bianco C., Piazzi A. (2001a). A semi-infinite optimization
approach to optimal spline trajectory planning of mechanical
manipulators, Semi-infinite programming: recent advances,
Goberna M.A and Lopez M.A. (Eds.), Springer Verlag, pp. 271-297.
Guarino Lo Bianco C., Piazzi A. (2001b). A hybrid algorithm for infinitely
constrained optimization, International Journal of Systems
Science, vol. 32, n. 1, pp. 91-102.
Guldner J., Utkin V.I. (1995). Sliding mode control for gradient tracking
and robot navigation using artificial potential fields, IEEE
Transactions on Robotics and Automation, vol. 11, n.2, pp. 247-
254.
Gupta K., Del Pobil A.P. (1998). Practical motion planning in robotics:
current approaches and future directions, John Wiley & Sons,
1998.
Hsu D., Kindel R., Latombe J.C., Rock S. (2002). Randomized kinodynamic
motion planning with moving obstacles, The International
Journal of Robotics Research, vol. 21, n. 3, pp. 233-255.
Hansen C., Oltjen J., Meike D., Ortmaier T. (2012). Enhanced approach for
energy-efficient trajectory generation of industrial robots,
Proceedings of the IEEE International Conference on Automation
Science and Engineering 2012 (CASE 2012), pp.1-7.
Huang P., Xu Y., Liang B. (2006). Global minimum-jerk trajectory planning
of space manipulator, International Journal of Control,
Automation and Systems, vol. 4, n. 4, pp. 405-413.
Ismail, M., Samir, L., & Romdhane, L. (2013). Dynamic in Path Planning of
a Cable Driven Robot, in Design and Modeling of Mechanical
Systems. Springer Berlin Heidelberg, pp. 11-18
LaValle, S. M. (2006). Planning algorithms. Cambridge university press.
Joonyoung K., Sung-Rak K., Soo-Jong K., Dong-Hyeok K. (2010). A
practical approach for minimum-time trajectory planning for
industrial robots, Industrial Robots: An International Journal,
vol. 37, n. 1, pp. 51-61.
Lin C.S., Chang P.R., Luh J.Y.S. (1983). Formulation and optimization of
cubic polynomial joint trajectories for industrial robots, IEEE
Transactions on Automatic Control, vol. 28, n. 12, pp. 1066-1073.
Kazemi, M., Gupta, K., & Mehrandezh, M. (2010). Path-planning for visual
servoing: A review and issues, Visual Servoing via Advanced
Numerical Methods (pp. 189-207). Springer London.
Kim J.O, Khosla P.K. (1992). Real-time obstacle avoidance using harmonic
potential functions, IEEE Transactions on Robotics and
Automation, vol. 8, n. 3, pp. 338-349.
Koditschek D.E. (1992). Exact robot navigation using artificial potential
functions, IEEE Transactions on Robotics and Automation, vol. 8,
n. 5, pp. 501-518.
Khatib O. (1985). Real-time obstacle avoidance for manipulators and
mobile robots. Proceedings of the 1985 IEEE International
Conference on Robotics and Automation, pp. 500-505.
Khoukhi A., Baron L., Balazinski M. (2009). Constrained multi-objective
trajectory planning of parallel kinematic machines, Robotics and
Computer-Integrated Manufacturing, vol. 25, n. 4-5, pp. 756-769.
Kim, J., Kim, S. R., Kim, S. J., & Kim, D. H. (2010). A practical approach for
minimum-time trajectory planning for industrial robots. Industrial
Robot: An International Journal, vol. 37, n. 1, pp. 51-61.
Kumar V., Zefran M, Ostrowski J.P. (2007). Motion planning and control of
robots, Handbook of Industrial Robotics (2nd edition), Shimon Y.
Nof editor, John Wiley & Sons, 2007.
Kunchev, V., Jain, L., Ivancevic, V., & Finn, A. (2006). Path planning and
obstacle avoidance for autonomous mobile robots: A review,
Knowledge-Based Intelligent Information and Engineering
Systems (pp. 537-544). Springer Berlin Heidelberg
Kyriakopoulos K.J., Saridis G.N. (1988). Minimum jerk path generation,
Proceedings of the 1988 IEEE International Conference on
Robotics and Automation, pp. 364-369.
Latombe J.C. (1991). Robot motion planning, Kluwer, 1991.
Lombai F., Szederkenyi G. (2008). Trajectory tracking control of a 6-
degree-of-freedom robot arm using nonlinear optimization,
Proceedings of the 10th IEEE International Workshop on
Advanced Motion Control, pp. 655-660.
Lombai F., Szederkenyi G. (2009). Throwing motion generation using
nonlinear optimization on a 6-degree-of-freedom robot
manipulator, Proceedings of the 2009 IEEE International
Conference on Mechatronics, pp. 1-6.
Lozano-Pérez T., Wesley M. A. (1979). An algorithm for planning collision-
free paths among polyhedral obstacles. Communications of the
ACM, vol. 22, n. 10, pp. 560-570.
Lozano-Perez T. (1983). Spatial planning: A configuration space approach.
IEEE Transactions on Computers, vol. 100, n. 2, pp. 108-120.
Liu, H., Lai, X., & Wu, W. (2013). Time-optimal and jerk-continuous
trajectory planning for robot manipulators with kinematic
constraints. Robotics and Computer-Integrated Manufacturing,
vol 29, n. 2, pp 309-317.
Martin B.J., Bobrow J.E. (1999). Minimum effort motions for open chain
manipulators with task-dependent end-effector constraints, The
International Journal of Robotics Research, vol. 18, n. 2, pp. 213-
224.
Nissoux C., Simon T., Latombe J.C. (1999). Visibility based probabilistic
roadmaps, Proceedings of the 1999 IEEE International
Conference on Intelligent Robots and Systems, pp. 1316-1321.
Pardo-Castellote G., Cannon R.H. (1996). Proximate time-optimal
algorithm for on-line path parameterization and modification,
Proceedings of the 1996 IEEE International Conference on
Robotics and Automation, pp. 1539-1546.
Petrinec K., Kovacic Z. (2007). Trajectory planning algorithm based on the
continuity of jerk, Proceedings of the 2007 Mediterranean
Conference on Control & Automation, pp. 1-5.
Pellicciari M., Berselli G., Leali F., Vergnano A. (2013) A method for
reducing the energy consumption of pick-and-place industrial
robots, Mechatronics, vol. 23, n. 3, pp. 326-334.
Piazzi A., Visioli A. (1997a). A global optimization approach to trajectory
planning for industrial robots, Proceedings of the 1997 IEEE-RSJ
International Conference on Intelligent Robots and Systems, pp.
1553-1559.
Piazzi A., Visioli A. (1997b). A cutting-plane algorithm for minimum-time
trajectory planning of industrial robots, Proceedings of the 36th
Conference on Decision and Control, pp. 1216-1218.
Piazzi A., Visioli A. (1997c). An interval algorithm for minimum-jerk
trajectory planning of robot manipulators, Proceedings of the 36th
Conference on Decision and Control, pp. 1924-1927.
Piazzi A., Visioli A. (1998). Global minimum-time trajectory planning of
mechanical manipulators using interval analysis, International
Journal of Control, vol. 71, n. 4, pp. 631-652.
Piazzi A., Visioli A. (2000). Global minimum-jerk trajectory planning of
robot manipulators, IEEE Transactions on Industrial Electronics,
vol. 47, n. 1, pp. 140-149.
Rubio F., Valero F., Sunyer J., Cuadrado J., (2012) Optimal time
trajectories for industrial robots with torque, power, jerk and
energy consumed constraints, Industrial Robot: An International
Journal, vol. 39, n. 1, pp. 92 – 100.
Saramago S.F.P., Steffen Jr. V. (1998). Optimization of the trajectory
planning of robot manipulators tacking into account the dynamics
of the system, Mechanism and Machine Theory, vol. 33, n. 7, pp.
883-894.
Saramago S.F.P., Steffen Jr. V. (2000). Optimal trajectory planning of
robot manipulators in the presence of moving obstacles,
Mechanism and Machine Theory, vol. 35, n. 8, pp. 1079-1094.
Saravan R., Ramabalan R., Balamurugan C. (2009). Evolutionary multi-
criteria trajectory modeling of industrial robots in the presence of
obstacles, Engineering Applications of Artificial Intelligence, vol.
22, n. 2, pp. 329-342.
Sciavicco L., Siciliano B., Villani L., Oriolo G. (2009). Robotics. Modelling,
planning and control, Springer Verlag, 2009.
Shiller Z. (1996). Time-energy optimal control of articulated systems with
geometric path constraints, Journal of Dynamic Systems,
Measurement, and Control, vol. 118, pp. 139-143.
Shin K.G., McKay N.D. (1985). Minimum-time control of robotic
manipulators with geometric path constraints, IEEE Transactions
on Automatic Control, vol. 30, n. 6, pp. 531-541.
Shin K.G., McKay N.D. (1986). A Dynamic programming approach to
trajectory planning of robotic manipulators, IEEE Transactions on
Automatic Control, vol. 31, n. 6, pp. 491-500.
Simon D. (1993). The application of neural networks to optimal robot
trajectory planning, Robotics and Autonomous Systems, vol. 11, n.
1, pp. 23-34.
Simon D., Isik C. (1993). A trigonometric trajectory generator for robotic
arms, International Journal of Control, vol. 57, n. 3, pp. 505-517.
Takahashi O., Schilling R. J. (1989). Motion planning in a plane using
generalized Voronoi diagrams. IEEE Transactions on Robotics
and Automation, vol. 5, n.2, pp. 143-150.
Tangpattanakul P., Artrit P. (2009), Minimum-Time Trajectory of Robot
Manipulator Using Harmony Search Algorithm, Proceedings of the
 IEEE 6th International Conference on ECTI-CON 2009, pp. 354-
357.
Tangpattanakul, P., Meesomboon, A., & Artrit, P. (2010). Optimal
Trajectory of Robot Manipulator Using Harmony Search
Algorithms. In Recent Advances In Harmony Search Algorithm,
pp. 23-36. Springer Berlin Heidelberg.
Trevisani, A. (2010). Underconstrained planar cable-direct-driven robots: A
trajectory planning method ensuring positive and bounded cable
tensions. Mechatronics, vol. 20, no. 1, pp. 113-127.
Trevisani, A. (2013). Experimental validation of a trajectory planning
approach avoiding cable slackness and excessive tension in
underconstrained translational planar cable-driven robots, in
Cable-Driven Parallel Robots. Springer Berlin Heidelberg, pp 23-
29.
Van Dijk N.J.M., Van de Wouw N., Nijmeijer H., Pancras W.C.M. (2007).
Path-constrained motion planning for robotics based on kinematic
constraints, Proceedings of the ASME 2007 International Design
Engineering Technical Conference & Computers and Information
in Engineering Conference, pp. 1-10.
Verscheure D., Demeulenaere B., Swevers J., De Schutter J., Diehl M.
(2008). Time-energy optimal path tracking for robots: a
numerically efficient optimization approach, Proceedings of the
10th International Workshop on Advanced Motion Control, pp.
727-732.
Volpe R.A. (1990). Real and artificial forces in the control of
manipulators: theory and experiments, The Robotics Institute,
Carnegie Mellon University, Pittsburgh, 1990.
Volpe R.A., Khosla P.K (1990). Manipulator control with superquadric
artificial potential functions: theory and experiments, IEEE
Transactions on Systems, Man, and Cybernetics, vol. 20, n.6, pp.
1423-1436.
Wang C.H., Horng J.G. (1990). Constrained minimum-time path planning
for robot manipulators via virtual knots of the cubic B-Spline
functions, IEEE Transactions on Automatic Control, vol. 35, n. 5,
pp. 573-577.
Williams, R. L., & Gallina, P. (2002). Planar cable-direct-driven robots:
Design for wrench exertion, Journal of intelligent and robotic
systems, vol. 35, n. 2, pp. 203-219.
Jing X.J., Edited by, (2008). Motion Planning, InTech
Xu H., Zhuang J., Wang S., Zhu Z. (2009). Global Time-Energy Optimal
Planning of Robot Trajectories, Proceedings of the International
Conference on Mechatronics and Automation, pp. 4034-4039.

View publication stats