100% found this document useful (1 vote)
2K views292 pagesRichard P. Paul Robot Manipulators Mathematics, Programming, and Control Artificial Intelligence PDF
Uploaded by
நந்த_குமார்Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
100% found this document useful (1 vote)
2K views292 pagesRichard P. Paul Robot Manipulators Mathematics, Programming, and Control Artificial Intelligence PDF
Uploaded by
நந்த_குமார்Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 292
ROBOT MANIPULATORS:
MATHEMATICS, PROGRAMMING, AND CONTROL
va, England ‘The MIT Press Series in Artificial Intelligence ae
Artificial Intelligence: An MIT Perspective, Volume I: Expert Problem Solving,
Natural Language Understanding, Intelligent Computer Coaches, Representation
and Learning edited by Pattick Henry Winston and Richard Henry Brown, 1979
Artificial Intelligence: An MIT Perspective, Volume Il: Understanding Vision
Manipulation, Computer Design, Symbol Manipulation edited by Patrick Henry
Winston and Richard Henry Brown, 1979
NETL: A System for Representing and Using Real-World Knowledge by Scott
Fahlman, 1979
The Interpretation of Visual Motion by Shimon Ullman, 1979
A Theory of Syntactic Recognition for Natural Language by Mitchell P. Marcus,
1980
Turtle Geometry: The Computer as a Medium for Exploring Mathematics by
Harold Abelson and Andrea diSessa, 1981
From Images to Surfaces: A Computational Study of the Human Early Visual
‘System by William Eric Leifur Grimson, 1981
Robot Manipulators: Mathematics, Programming and Control by Richard P, Paul,
1981 ROBOT. MANIPULATORS:
MATHEMATICS, PROGRAMMING, AND CONTROL
The Computer Control of Robot Manipulators
Richard P. Paul
By
a
“THE MIT Press
‘Cambridge, Massachusetts and London, England TSIM.
P38
PUBLISHER'S NOTE
‘This format is intended to reduce the cost of publishing certain works in book
form and to shorten the gap between editorial preparation and final publication.
‘The time and expense of detailed editing and composition in’ print have been
avoided by photographing the text of this book directly from the author's com-
puter printout,
Sixth printing, 1984
Copyright © 1981 by
‘The Massachusetts Institute of Technology
All rights reserved, No part of this book may be reproduced in any form, or by
any means, electronic or mechanical, including photocopying, recording, or by
any information storage and reirieval system, without permission in writing from
the publisher.
Printed in the United States of America,
Library of Congress Cataloging in Publication Data
Paul, Richard P. \
Robot manipulators.
(fhe MIT Press series in artificial intelligence)
Bibliography: p.
Includes index.
1. Robots, Industrial. 2. Manipulators
(Mechanism) I. Title. IL. Series.
TS1i91.8.P38 629.8'92 81-13689
IsaN 0-262-16082-X AACR
2AST ES
781918 p39 eset
Robot; “eT ay
{
| XK
Z
x
CONTENTS
Preface ix
Inuodueion 1
Chapter]
Homogeneous Transformations. 9
Chapter?
Kinematic Equations 41
Chapter3
Solving Kinematic Equations 65
Chapter 4
Differential Relationships 85,
Chapter $
Motion Trajectories a9
Chapter 6
Dynamics 157
Chapter 7 H
Control 197 {
Chapter 8 é
Static Forces 217 i
Chapter 9
Compliance 231
‘Chapter 10 |
Programming 245
Index 273 SERIES FOREWORD
Artificial intelligence is the study of intelligence using the ideas and methods of
computation, Unfortunately, a definition of intelligence seéms impossible at the
‘moment because intelligence appears to be an amalgam of so many information-
processing and information-representation abilities,
Of course psychology, philosophy, linguistics, and related disciplines offer
various perspectives and methodologies for studying intelligence, For the most
Part, however, the theories proposed in these ficlds are too incomplete and too
vaguely stated to be realized in computational terms. Something more is needed,
even though valuable ideas, relationships, and constraints can be gleaned from
traditional studies of what are, after all, impressive existence proofs that intel-
ligence is in fact possible,
Antificial intelligence offers a new perspective and a new methodology. Its
central goal is to make computers intelligent, both to make them more useful
‘and to understand the principles that make intelligence possible, That intelligent
computers will be extremely useful is obvious. The more profound point is that
artificial intelligence aims to understand intelligence using the ideas and methods
‘of computation, thus offering 2 radically new and different basis for theory formax
tion. Most of the people doing artifical intelligence believe that these theories will
apply to any intelligent information processor, whether biological or solid state
‘There are side effects that deserve attention, too. Any program that will sue-
cessfully model even a small part of intelligence will be inherently massive and
complex, Consequently, artificial intctligence continually confronts the limits of
‘computer science technology. The problems encountered have been hard enough
and interesting enough to seduce artificial intelligence people into working on
them with enthusiasm, It is natural, then, that there has been a steady flow of
ideas from artificial intelligence to computer science, and the flow shows no sign,
of abating.
‘The purpose of this MIT Press Series in Artfictal Intelligence is to provide
People in many areas, both professionals and students, with timely, detailed infor-
‘mation about what is happening on the frontiers in research centers all over the
world,
Patrick Henry Winston
Mike Brady Pe,
Alin PREFACE
‘This book on robot manipulators brings together theories from a number of
fields: computer graphics, kinematics, dynamics, control, and programming, The
book is written in the style of a text and has been used to teach robotics to
‘graduate and undergraduate students in engineering and computer science, The
field of robotics is new and interdisciplinary, with an almost constant flow of
theories into robotics from many areas. This book presents only one theory in
each area, These theories are selected because they can be brought together
to provide a working and consistent approach to the overall problem of robot
manipulator control, Most of the material in the book is atthe level of trigonome-
ity but is conceptually new and different. ‘The chapter on dynamics assumes that
the studeat is prepared to accept the Lagrangian formulation of dynamics. The
chapter on control makes use of the Laplace transformation, and the final chapter
‘on programming assumes a knowledge of PASCAL.
‘This work represents ten years experience in robotic
presented has been experimentally tested and verified and is in use today running
robot manipulators. Some material clarifies past systems or simplifies and extends
earlier theories. All the theory is developed with a digital computer implementa-
tion in mind. Computing speed, roundoff, and memory requirements have guided
the selection of many of the theoretical approaches, With the advent of VLSI,
however, more computationally intensive approaches may be appropriate.
‘The book is based on course notes developed at Purdue University over the
rast four years and has been completed using the facilities of the Massachusetts
Institute of Technology Artificial Intelligence Laboratory. The camera-ready copy
.was produced using Knuth's TEX system, Much of the theory presented in the
book was developed at Stanford University Artificial Intelligence Laboratory, and
the author wishes to thank John McCarthy and Jerome Feldman, who provided
him with the possibilty of performing the research. The author also wishes to
thank Antal Bejczy, upon whose work the dynamics chapter is based, and Bruce
Shimano, who has been part of the development of much of the theory. The
support of Mike Brady and Eric Grimson mado the final transition into print a
pleasant experience. John Birk, Mike Brady, William Fisher, Tomas Lozano-
Perez, Johnson Luh, William Perzley, and Bruce Shimano helped by reading
parts of the manuscript and correcting many errors; the author is entirely respon-
sible for any that remain, Met Paul assisted with revisions and editorial changes,
most of the material ‘This material is based partially upon research supported by the National
Science Foundation under Grant Numbers APR77-14533, APRTS-13074, and
‘APR74-01390. Any opinions, findings, and conclusions or recommendations ex-
pressed in this publication are those of the author and do not necessarily reflect,
the views of the National Science Foundation, ROBOT MANIPULATORS:
MATHEMATICS, PROGRAMMING, AND CONTROL INTRODUCTION
“This book is about industrial robot manipulators, their programming and their
interaction with sensors. We will cover robot manipulator programming and con-
trol in detail but will treat external sensors, such as vision, only in so far as they
interact with the manipulator. In order to gain an overall understanding of the
‘topics to be addressed in later chapters, let us first look at the origins of industrial
robots and the problems that arose as their capabilities were extended.
‘The present day industrial robot had its origins in both the teleoperator and the
‘numerically controlled machine tool, The teleoperator, or telecheric, is a device
to allow an operator to perform a task at a distance, ‘The numerically controlled
machine tool shapes metal automatically, based on digitally. encoded cutting data.
‘The teleoperator was developed during the second world war to handle
radioactive materials [Goertz). An operator was separated from a radioactive task
‘by aconcrete wall with one or more viewing ports through which the task could
be observed, The teleoperator was to substitute for the operator's hands; it con-
sisted of a pair of tongs on the inside (the slave), and two handles on the outside
(the master). Both tongs and handles were connected together by six degree of
freedom mechanisms to provide for the arbitrary positioning and orienting of the
master and slave, The mechanism was provided to control the slave in order to
replicate the motion of the master.
‘In 1947, the first servoed electric-powered teleoperator was developed. The
slave was servo-contrclled to follow the position of the master. No force infor-
mation was available to the operator, and tasks requiring that parts be brought
into contact were difficult to perform, even though the operator could view the
task. Teleoperator tasks are well described by Goertz: “The general-purpose
‘manipulator may be used for moving objects, moving levers or knobs, assembling
parts, and manipulating wrenches. In all these operations ‘the manipulator must
come into physical contact the object before the desired force and moment
can be made on it, A collision occurs when the manipulator makes this contact.
‘Generak-purpose manipulation consists essentially of a series of collisions with
unwanted forces, the application of wanted forces, and the application of desired
motions, The collision forces should be low, and any other unwanted forces
should also be small.” (Goertz}
In 1948, one year after the position-controlled teleoperator was introduced, a
new teleoperator system was developed in which the force exerted by the tongs Introduetion 2
could be relayed to the operator by backdriving the master; the operator could
once again feel what was going on.
In 1949, faced with the need to procure advanced aircraft whose parts were
designed to be machined rather than riveted, the Air Force sponsored research
in the development of a numerically controtled milling machine {Rosenberg}.
This research was to combine sophisticated servo system expertise with the newly
developing digital computer techniques. The pattern to be cut was stored in digi-
tal form on a punched tape and then a servo-controlled milling machine cut the
metal. The MIT Radiation Laboratory demonstrated such a machine in 1953,
In the 1960's, George Devol demonstrated what was to become the first Uni-
‘mate industrial robot (Engelberger], a device combining the articulated linkage
of the teleoperator with the servoed axes of the numerically controffed milling
machine. The industrial robot could be taught to perform any simple job by
driving it by hand through the sequence of task positions, which were recorded in
digital memory. Task execution consisted in replaying these positions by servoing
the individual joint axes of the the robot. Task interaction was limited to opening
and closing of the tongs or end effector, and to signaling external equipment
or waiting for a synchronizing signal. ‘The industrial robot was ideal for pick
and place jobs such as unloading a diecasting machine. ‘The part would appear
in a precise position, defined with respect to the robot; it would be grasped,
moved out of the die, and dropped on a conveyor, The success of the industrial
robot, like the numerically controlled (NC) milling machine, relied on precise,
repeatable digital servo loops. There was no interaction between the robot and its
work. If the diecast machine were moved, the robot could in no way adapt to the
new position, any more than an NC milling machine could successfully cut a part
if the stock were arbitrarily relocated during cutting. If the diecasting machine
were moved, the robot could, however, be retaught. The success of the industrial
robot lay in its application to jobs in which task positions were absolutely defined,
and in its reliability and positioning repeatability in lieu of adaptation.
‘Another development took place in MIT's Lincoln Laboratory in 1961 (Ernst)
A teleoperator slave arm, equipped with touch sensors, was connected to a com-
puter. In this case, touch feedback from the tongs, not absolute position, guided
the robot in the performance of the task. Although the absolute position of the
hand in space could be obtained, this approach was not used, since humans do
not perform the trigonometric calculations necessary to obtain a hand position,
Instead, tasks were defined as a sequence of touchdefined states to be attained.
‘The lack of any global idea of the position of objects limited this robot as much
as the complete lack of task information limited the position-controlled industrial
robot.
In 1963 Roberts [Roberts63] demonstrated the feasibility of processing a digit
zed halftone picture of a scene to obtain a mathematical description of the block- like objects which comprised the scene, expressing their location and orientation
by homogeneous coordinate transformations [Robers65]. This work was impor-
tant for two reasons: it demonstrated that objects could be identified and located
in a digitized halftone image, and it introduced homogeneous transformations as
a suitable data structure for the description of the relative position and orientation
between objects. If the relative position and orientation between objects is repre-
sented by homogeneous transformations, the operation of matrix multiplication
of the transformations establishes the overall relationship between objects. In
Chapter 1 we introduce homogeneous transformations, which we use throughout
the book to describe both position and orientation between objects and between.
objects and the manipulator.
Touch feedback, because of its slow, groping nature, was dropped in favor of
vision as an input mechanism, By 1967, a computer equipped with a television
camera as an input mechanism could, in real time, identify objects and their
location [Wichman]. The manipulator, stripped of its touch sensors, relied on
position-servoed joint axes. Homogencous transformations, however, expressed
the position-and orientation of the end effector in Cartesian coordinates, not
as the angles between a series of unorthogonal manipulator joints. Pieper ap-
plied the theories of closed-link chains to obtain a solution to this problem,
and the manipulator could then be commanded to move to Cartesian positions
in the workspace [Pieper]. In Chapter 2 we develop the methods of obtaining
the Cartesian coordinate position and orientation of the end effector for any
manipulator, given the joint coordinates as input. The position and orientation
are represented by a homogeneous transformation. In Chapter 3 we investigate
the inverse problem of obiaining the joint coordinates for a manipulator given
‘the Cartesian coordinate description of the end effector. This problem is difficult,
requires intuition, and can be solved analytically for only a limited number of
simple manipulators. Fortunately, most commercially available manipulators are
of this form and are solvable (see Figures 32, 4.3, 62, and 9.1).
By 1970, a camera- and arm-equipped computer could play real-world games
and the “instant insanity” puzzle was successfully solved at Stanford University
[Fetdman], In this puzzle, four cubes with diferent colored faces must be stacked
so that no identical colors appear on any side of the stack. At MIT, a block
structure could be observed and copied. In Japan, research led to a hand-eye
system which could assemble block structures when presented with an assembly
drawing. The drawing was first viewed, the materials surveyed, and then the
required structure built [Fir], Potential applications of this research appeared to
‘be in deep space probes and in factory automation. Corresponding to these goals,
vision research was directed to the more difficult task of identifying castings and
rocks, and manipulation research was directed to the assembly of objects.
‘The manipulator, working in relationship with a vision sensor, is frequently Introduction 4
required to make small motions of accommodation. I Chapter 4 we consider the
[nn | (2.21)
0001 Aransformation]
2.6 Cylindrical Coordinates
‘We might, however, wish to specify the position of the hand in cylindrical coor
dinates, This corresponds to a translation r along the 2 axis, followed by a rota-
thon a about the 2 axis, and finally a translation z along the = axis (see Figure
26).
Cyl(z, a, 7) = Trans(0, 0, z) Rot(z, a) Trans(r, 0, 0)
‘cosa —sina 0 O]f1 0 0 ¥
r, sina cosa 0 010 1 0 0
Cale, 0,7) = Trans(a, 0,2" EF 9 OH) SD Of (2.22)
0 0 oO io 0 OL Cylindrical Coordinates 48
Ls —y
ot
Figure 26. Cylindrical Polar Coordinates
x
Olfeosa —sina 0 reosal
sina cosa 0 rsina
: ofa (2.23)
0 o o 8
Cyilz, a, 7) =
cosa —sina 0 reosal
: [sine cosa 0 rsina .
Mea ee nD (2.24)
o 0 o 8
If we were to postmultiply this transformation by an orientation transform, as in
Equation 2.21, then the orientation of the hand would be with respect to the sta
tion coordinates rotated a about the 2 axis. IF it were desired to specify orientation
with respect to unrotated station coordinates, then we would rotate 2.24 by —a
about its z axis or
cosa —sina 0 reosalfeos(—a) —sin(—a) 0 0)
_|sina cosa 0 rsinal|sin(—a) co(—a) 0 0
Moana “o 1 ez 0 o 10
o o ft ° 0 0
cosa —sina 0 reosalf cosa sina
Colgan) =]eine ca renal —sing cova
o o o tho o
ence
neces x
Figure 27, Spherical Polar Coordinates
10-0 reosay
{0°10 rsina
oieay=|p 4 pe (2.27)
000 1
‘This isthe form in which we will interpret Cyl(z, a, 7).
2.7 Spherical Coordinates
Finally, we will consider the method of specifying the position vector by means of
spherical coordinates, This method corresponds to a translation r along the 2 axis,
followed by a rotation B about station y, and then a rotation a about station z (see
Figure 2.7).
Sph(a, B,r) = Rot(z, a) Rot(y, 8) Trans(0, 0, r) (2.28)
cosh 0 sing Olf1 0 0 6
i o 1 0 ofo 100
Soh(e,B,7) = Rotia}] tos § eons Olle a 1 oi (2.29)
0 0 0 ilo oo 1
cosa —sina 0 Of cos 0 sinf rsing’
—|sine cosa 0 Off O 1 0 0
SMe yy OL Oll—wing 0 exes reosg| 2)
0 0 OL 0 o 0 1 samc
Specification of Ts 50
Table 24
[Transiation] Ban. [Rotation] Eqn.
PoP Pe 00,0000,
Rot(k, 8) 170
Cyr, a,7) 227 Buler( 4,0, 8) 216
Sph(a, 6, r) 33 RPY(4,8, 9} 2.20
cosacos8 —sina cosasing rcosasingy
= |sinacosé cosa sinasinB rsinasinf
See,.)=l inp 0 cosh cosh seen)
0 oO 0 1
Once again, if we do not wish the orieitation to be expressed with respect to this
rotated coordinate frame, we must postmultiply by Rot(y, —A) and Rot(z, ~
