Académique Documents
Professionnel Documents
Culture Documents
Winter 2016
1
Contents
Abstract .............................................................................................................................................. 6
Objective ............................................................................................................................................ 7
Introduction ........................................................................................................................................ 8
1.
1.1.
1.2.
1.3.
1.3.1.
1.3.2.
1.4.
1.4.1.
1.4.2.
1.4.3.
1.4.4.
1.4.5.
1.5.
1.5.1.
.1.5.2
1.6.
1.6.1.
1.6.2.
1.6.3.
1.7.
1.7.1.
1.7.2.
1.7.3.
1.7.4.
1.8.
1.8.1.
Chapter 2 .......................................................................................................................................... 28
Literature Review ............................................................................................................................. 28
2.
Introduction ............................................................................................................................... 28
2.1.
2.2.
2.3.
2.3.1.
2.3.2.
2.3.3.
Problem Definition.................................................................................................................... 37
3.1.
3.2.
Methodology ......................................................................................................................... 38
3.3.
References ........................................................................................................................................ 40
Chapter 1: Introduction
Abstract
Redundant robot has become widely used in industrial applications to perform repetitive tasks
such as welding and cutting materials. Redundancy in robots can be achieved by adding more
DOF than they needs, which will allow them to perform their tasks with more dexterity such as
dealing with uncertainties in their workspace. However, the maintaining of end-effector trajectory
in the presence of kinematic uncertainties such as free motion or constrained motion in redundant
robots becomes an important issue. Designing an adaptive control to use for trajectory tracking is
considered as an appropriate solution for this problem. Specific rules will be applied to generate
intelligent algorithm profile for all the possible solutions to perform the repetitive task with
acceptable quality and minimum errors. Experiments will be made on a redundant robot (ETS
laboratory) to validate the proposed models and to illustrate the robustness of control laws
developed in the context of real-time control.
Objective
The main objective: designing adaptive control for redundant robots to perform repetitive
task with knowledge of planning trajectory in the presence of uncertainty in the workspace
environment.
The specific objective: designing specific rules to identify the appropriate solution and
perform it with minimum errors and high possible velocity in the presence of uncertainties
in their workspace
Introduction
Developments in the electromechanical industrial sector has reached a great progress by making
robots to become more cost effective at their jobs, and as human labor continues to become more
expensive, more and more industrial jobs have become candidates for robotic automation. This is
the most important trend pushing growth of the industrial robot market. Additionally to the
economic side, the fact that robots became more intelligent and able to perform tasks that are
either dangerous or impossible for human workers to accomplish. The applications, in which
industrial robots perform, are gradually increasing in the year 2000; approximately 78% of the
robots installed in the US were welding or material-handling robots. Figure 1.1 shows the Robot
prices compared with human labor costs in the 1990s [12]. This report will introduce the
characteristics and operations of the redundant robotics in industrial to enhance the productivity
and how repetitive tasks are performed with redundant robotics under uncertainty in the workspace
environment.
Figure1.1 Robot prices compared with human labor costs in the 1990s
1. Redundant robot
In general, a robot can be defined as a goal oriented machine that can sense plan and act, and can
be classified to four categories; manipulator, mobile robot, machine vision and artificial
intelligence. This research problem will concentrate on manipulator that has more degree of
freedom (DOF) than it need, which is known as redundant robot. This chapter will introduce
briefly the kinematic and dynamic of redundant robot. In addition, the concept of adaptive control
rules will be explained.
cylindrical, planar, screw, and spherical. However, most manipulators have revolute joints or
prismatic joints. In uncommon case, that mechanism is built with a joint having n degree of
freedom, it can be modeled as n joints of one degree of freedom connected with n-1 links of zero
length. The links are numbered starting from the fixed link called base link, which is called link 0.
The first moving body is link 1, and so on; out to the end free link which is called link n. Typical
redundant manipulators have more than 6 joints. A single link of a typical robot has many
attributes which is a mechanical designer had to consider during its design. For instance, the type
of material used the strength and stiffness of the link, the location and type of the joint bearing, the
external shape, the weight and inertia.
11
12
The inverse
kinematic problem is to determine the link positions from the knowledge of the end effector. Using
redundant robot to perform repetitive task in industrial sector the planning trajectory is always
known. In addition, the task space will be within the dexterous space of the robot and that will lead
to multi-solution or infinite solution problem.
13
The fact that a manipulator has multiple solutions will cause problems because the system has to
be able to choose one. However, multiple solutions are necessary to solve the uncertainty problems
that might be accrued environment of the robot such as obstacles and any parameters changes. The
14
criteria upon which to base a decision vary, but a very reasonable choice would be the closet
solution as shown in Figure 1.6. Moreover, the number of solution depends upon the number of
degree of freedom. For example, completely general rotary-jointed manipulator with six degrees of
freedom, there are up to sixteen solutions possible
Figure 1.6 one of the two possible solutions to reach point B causes a collision
15
Newton's equation: the force, F, acting at the center of mass of a rigid body which
16
Figure 1.7 force F acting on the center of mass of the body causes the body to accelerate at
II.
Euler's equation
The moment N, which must be acting on the rigid body to cause this motion, is given by Euler's
equation and shown at fig.1.8
N = C I + x C I Euler's equation
Where C I is the inertia tensor of the rigid body written in a frame, {C}, whose origin
is located at the center of mass, angular velocity and with angular acceleration .
Figure 1.8 A moment N is acting on a body, and the body is rotating with
velocity
cv and accelerating at th.
17
1
2
K = 2 m . +
1
2
The first term is kinetic energy due to linear velocity ( ) of link's center of mass and the second
term is kinetic energy due to angular velocity () of the link.
The kinetic energy of a manipulator (quadratic form) is given by:
K (, ) =
1
2
()
Where
constant.
The lagrangian of a manipulator is
L (,) = K (,) u ()
The equation of motion for the manipulator are then given by
18
+ =
[12]
end-effector, or workpieces. One way to include more detail in a path description is to give a
sequence of desired via points (intermediate points between the initial and final positions). The
name path points includes all the via points plus the initial and final points. Along with these
spatial constraints on the motion, the user could also wish to specify temporal attributes of the
motion. Usually, it is desirable for the motion of the manipulator to be smooth function that is
continuous and has a continuous first derivative and second derivative as well. Rough, jerky
motions tend to cause increased wear on the mechanism and cause vibrations by exciting
resonances in the manipulator. Any smooth functions of time that pass through the via points
could be used to specify the exact path shape.
(0) = 0 ,
( 0) = 0,
20
( ) = 0
( ) = 0
(1)
(2)
() = 0 + 1 + 2 2 + 3 3
(3)
Where 0 , 1 , 2 , 3 polynomial coefficients the joint velocity and acceleration along this
path are clearly
() = 1 + 2 2 + 33 2
() = 2 2 + 63
(4)
3
( 0 )
2
3 =
2
( 0 )
3
(5)
By using (5), we can calculate the cubic polynomial that connects any initial joint angle position
with any desired final position.
(6)
(7)
Solving these equations for the we get we get
0 = 0
1 = 0
2 =
3 =
2
3
3
2
( 0 )
( 0 ) +
1
2
(8)
( + 0 )
Using (8), we can calculate the cubic polynomial that connects any initial and final positions with
any initial and final velocities. In Figure 10, shows a reasonable choice of joint velocities at the via
points, as indicated with small line segments representing tangents to the curve at each via point.
22
23
Where SKA is formed by scaling the unit vector SKA by the amount or rotation SA,
When via points are passed, the linear splines with parabolic blends are used and angular velocity
of the end-effector are changed smoothly. One minor problem arises from the fact that the angle
axis representation of orientation is not unique that is,
Where n is any positive or negative integer. Figure 1.11 shows four different possible of SKB and
their relations to SKA [12] .
Figure 1.11 shows four different possible of SKB and their relations to SKA
25
to robust control in dealing with uncertainties in constant or slow-varying parameters. Figure 1.12
is shown the operation signals of adaptive control.
26
the system with minimum efforts. Optimization also, can be defined as the process of finding the
conditions that give the maximum or minimum value of a function [15].
27
Chapter 2
Literature Review
2. Introduction
In the last decade, development and research in robotics has been very active. Adding more degree
of freedom to robots than they normally require causes redundancy. This redundancy enhances the
performance of robotics and adds more cleverness to perform their tasks like human, and allow
them to avoid any unforeseen condition in their work space. Generally, a redundancy indicates that
the robot has an internal movement which does not influence the trajectory of the robots end28
effectors and allows the robot to perform auxiliary tasks such as, obstacle avoidance and any other
uncertainties conditions. At the same time redundancy add more complexity to the control design
methods and optimisation techniques of the redundant robots. This chapter will introduce some of
the research papers retreated to the control and optimization of robots.
29
easily implemented in real time to the physical system. This implementation would be the future
work of this research project [2].
In 2011, Fabricio G. proposed sliding mode speed auto-regulation technique for robotic tracking
path tracking. Fuzzy controller evaluated the freely accessible 6R robot model PUMA-560, and
path tracking is divided into a geometric (desired error) and a dynamic (desired speed), where
speed profiles are assigned for nonlinear systems to track non-smooth paths. It is completely
independent of the robot parameters. This proposed technique can be extended to include
acceleration and/or torque constraints.[5]
In 2012, Madani, T. proposed an adaptive variable structure controller of redundant robots with
mobile/fixed obstacles avoidance. The controller is designed directly in Cartesian space and no
knowledge on the dynamic model is needed, except its structure, and the parameters of the
controller are adapted using adaptive laws obtained via Lyapunov stability analysis of the closed
loop. The Simulation and results obtained using a 3DOF robot manipulator evolved in the vertical
plan, and can be implemented on a robot that has more than 3DOF [18].
In 2013, Jing Zhao published a paper addressed as Generating human-like movements for robotic
arms. The control of criteria based method is combining both kinematic and dynamic criteria. A
new inverse kinematic algorithm was developed to plan the human-like reaching movements.
The Gradient Projection Method based Rapidly-exploring Random Tree (GPM-RRT) algorithm is
proposed to plan the human-like grasping movement. The criteria for human-like control in
robotics are incomplete because the obstacle avoidance problem is not considered, and GPM-RRT
can directly solve this problem [10].
In 2015 Hariharan S. in his PhD thesis designed a complete path planning of higher DOF
manipulators in human like environments. A multi-pass sequential localized search technique to
30
solve the problem of path planning of hyper-redundant manipulators for the shortest path in realtime in the presence of obstacles is used. He divided the path to nodes and used finite state systems
to represent the state at each stage and the lines connecting those represents gk (xk, uk). The
control sequence corresponds to the path originating at the initial state x0 and terminating at the
final state xN. The numerical solution is used to solve invers kinematic. The original task of
planning motions for hyper-redundant manipulator arms was implemented only in the presence of
obstacles. The future work will deal with other uncertainty problems such as load change or
surface friction [9].
In 2013, Farzin Piltan proposed a robust fuzzy sliding mode control technique for robot
manipulator systems with modeling uncertainties. In this research, a joint level controller for
continuum robots is described which utilizes a fuzzy methodology component to compensate for
dynamic uncertainties. Therefore, a fuzzy sliding mode tracking controller for robot manipulators
with uncertainty in the kinematic and dynamic models is designed and analyzed. The design
satisfied the desired performance objectives in the presence of noises, disturbances, parameter
perturbations, unmodeled dynamics, sensor failures, actuator failures, time delays. Sliding mode
method provides an effective tool to control nonlinear systems through the switching function and
dynamic formulation of nonlinear system [8].
In 2015, Qi Zhou presents a novel design method for adaptive tracking control which is proposed
for robot finger dynamics. First, the dynamics are described by considering the robot finger as a
largescale system since it has many joints and multi (DOFs). Second, by employing the direct
adaptive fuzzy approximation method to approximate the unknown and desired control input
signals instead of the unknown nonlinear functions, the number of adaptive design parameters
obtained in the control design process is greatly reduced. Moreover, it is shown that the designed
31
controller can guarantee all the signals in the closed-loop system to be semi-globally uniformly
ultimately bounded. The future work, based on the decentralized adaptive control method, the
tracking performance will be improved by implementing a fuzzy logic system control or neural
network control approaches [14].
In 2015, Claudio Urrea used a simulation tool to design and compare sex controller approaches
which are applied to 5DOF redundant robot. Through homogeneous transformation matrices the
inverse kinematic model of the redundant robot is obtained. Six controllers are prepared to test the
robots dynamic model: hyperbolic sinecosine; computed torque; sliding hyperbolic mode;
control with learning; and adaptive. A simulation environment is developed by MatLab/Simulink
software, which allows analyzing the dynamic performance of the robot and of the designed
controllers. This simulation environment is used to carry out different tests of the redundant
manipulator model together with each controller as they are made to follow a trajectory in space.
The results, obtained through a simulation environment, are represented by comparative curves
and RMS indices of the joint and Cartesian errors. They show that the redundant manipulator
model follows the test trajectory with less error using the adaptive controller than the other
controllers, with more homogeneous motions of the manipulator. The largest joint and Cartesian
errors generated occurred when the computed torque controller is used. Fig 2.1 shows the
comparison of joint error [4].
32
33
34
final configurations. Then a graph is generated which will be completed as new configurations are
obtained. Each node in the graph represents a possible configuration of the robot, and the arcs
between nodes represent the distance travelled by the significant points of the robot between these
configurations [7].
35
36
3. Problem Definition
Redundant robots have become widely used in industrial applications such arc and spot welding
painting, assembly and material cutting. Therefore, in an industrial application the path of endeffector is known and it will be in the dexterous space of robot, and that will lead to multi-solution
problem for inverse kinematic. Moreover, Robotic manipulators in industrial applications will
have to encounter nonlinearities and various uncertainties in their dynamic models, such as
friction, disturbance, and load changing. Due to the above factors, maintaining the end-effector
trajectory in the presence of these uncertainties will be the focus of this research (position control).
Designing a controller to achieve the acceptable speed with a minimum error is another challenge
(velocity control).
The forward kinematic problem: is to determine the position of the end effector from the
knowledge of the link positions, in other words, we know rotations angles and the translations
between them.
The inverse kinematic problem is to determine the link positions from the knowledge of the end
effector trajectory. The research will deal with solving the inverse kinematic problem of redundant
robot which is 7DOF to perform repetitive task in industrial sector such as arc welding, painting.
It will provide a new intelligent control algorithm to add more cleverness to redundant
robots and more security as well as safety in their work space.
It will combine concepts from both computer science and control theory, in order to
develop a meaningful theory for controlling the motion in redundant robots
37
3.2. Methodology
1. Stage1. Model the redundant robot with 7 DOF which will
solutions to perform the predetermined trajectory by using different parameters in the invers
kinematic matrix by Mathlab.
2. Stage 2. Simulate the uncertainties in the workspace such as obstacles and surface friction and
generate profile for each possible solution by Mathlab.
3. Stage 3. Specify control rules to generate a complete intelligent profile for all the possibility
solutions to perform the predetermined trajectory.
4. Stage 4. Implement the solution experimentally on 7 DOF ANAT robot in the ETS laboratory.
38
2015
2016
2017
2018
2019
Task 2
DGA
1031
Task 3
DGA
1032
Written
exam
Task
4
DGA
1033
Oral
exam
Task 5
First
Contribution
Task 6
Second
contribution
Task 7
Third
Contribution
Task 8
Thesis
Defense
39
Fall
Summer
Winter
Fall
Summer
Winter
Fall
Summer
Winter
fall
summer
Winter
Fall
Task 1
DGA
1005
References
[1] Boubaker Daachi, et al., Adaptive neural controller for redundant robot manipulators and
collision avoidance with mobile obstacles, Neurocomputing 79 (2012) 5060
[2] Brice Le Boudec, et al., Modeling and adaptive control of redundant robots, since direct,
Mathematics and Computers in Simulation 71 (2006) 395403
[3] Claudio Urrea, John Kern, Design, simulation and comparison of controllers for a redundant
robot, Case Studies in Mechanical Systems and Signal Processing, 2015
[4] Claudio Urrea, John Kern, Trajectory Tracking Control of a Real Redundant Manipulator of
the SCARA Type, J Electr Eng Technol.2015;
[5] Fabricio Garelli, Luis Gracia, et al., Sliding mode speed auto-regulation technique for robotic
tracking, Robotics and Autonomous Systems 59 (2011) 519529
[6] Franc Hanz. Karel Jezernik, Finite-state machine control of mechatronic systems, J Systems
and Control Engineering 226(10) 13941409, 2012
[7] Francisco Rubio, et al., Comparing the efficiency of five algorithms applied to path planning for
industrial robots, Industrial Robot: An International Journal 39/6 (2012) 580591
[8]Farzin Piltan, Design Robust Fuzzy Sliding Mode Control Technique for Robot Manipulator
Systems with
Modeling Uncertainties, I.J. Information Technology and Computer Science, 2013, 08, 123-135
[9] Hariharan S. Ananthanarayanan, Complete path planning of higher DOF of manipulators in
human like environment, PhD thesis, Copyright by Hariharan S. Ananthanarayanan All rights
reserved 2015
[10] Jing Zhao, et al., Generating human-like movements for robotic arms, since direct,
Mechanism and Machine Theory 81 (2014) 107128
[11] J.M. Selig, Introductory Robotics, Prentice Hall, London, 1992
[12] John J. Craig, Introduction to Robotics Mechanics and Control, Third edition, Pearson
Prentice Hall, 2005
[13] Jing Zhao, Biyun Xie, Chunyu Song, Generating human-like movements for robotic arms,
Mechanism and Machine Theory 81 (2014) 107128
40
[14] Qi Zhou, et al., Decentralized Adaptive Fuzzy Tracking Control for Robot Finger Dynamics,
ieee transactions on fuzzy systems, vol. 23, no. 3, june 2015
[15] Singiresu S. Rao, Engineering Optimization Theory and Practice, by John Wiley & Sons, 2009
[16] Slotine, Jean-Jacques, Applied nonlinear control, Prentice Hall, New Jersy 1991
[17] Takashi Nammoto, and Kazuhiro Kosuge, An Analytical Solution for a Redundant
Manipulator with Seven Degrees of Freedom, Copyright 2012 International Journal of
Automation and Smart Technology
[18] Tarek Madani, et al., Adaptive variable structure controller of redundant robots with
mobile/fixed obstacles avoidance, Robotics and Autonomous Systems 61 (2013) 555564
[20] Z. Doulgeri, Y. Karayiannidis, Force position control for a robot finger with a soft tip and
kinematic uncertainties, Robotics and Autonomous Systems 55 (2007) 328336
[21] Tzuu-Hseng S. Li, Yun-Cheng Huang, MIMO adaptive fuzzy terminal sliding-mode
controller for robotic manipulators, Information Sciences 180 (2010) 46414660
41