Biomimetic Robotics Mechanisms and Control 1st Edition Ranjan Vepa 2024 Scribd Download
Biomimetic Robotics Mechanisms and Control 1st Edition Ranjan Vepa 2024 Scribd Download
com
https://ebookgate.com/product/biomimetic-robotics-
mechanisms-and-control-1st-edition-ranjan-vepa/
https://ebookgate.com/product/flight-dynamics-simulation-and-
control-for-rigid-and-flexible-aircraft-1st-edition-ranjan-vepa/
https://ebookgate.com/product/automatic-control-of-bioprocesses-
control-systems-robotics-and-manufacturing-1st-edition-denis-
dochain/
https://ebookgate.com/product/production-scheduling-control-
systems-robotics-and-manufacturing-1st-edition-pierre-lopez/
https://ebookgate.com/product/cell-cycle-control-mechanisms-and-
protocols-1st-edition-tim-humphrey/
Flexibility and Robustness in Scheduling Control
Systems Robotics and Manufacturing 1st Edition Jean-
Charles Billaut
https://ebookgate.com/product/flexibility-and-robustness-in-
scheduling-control-systems-robotics-and-manufacturing-1st-
edition-jean-charles-billaut/
https://ebookgate.com/product/introduction-to-robotics-analysis-
control-applications-2nd-edition-saeed-b-niku/
https://ebookgate.com/product/new-mechanisms-in-glucose-
control-1st-edition-anthony-h-barnett/
https://ebookgate.com/product/transseptal-catheterization-and-
interventions-1st-edition-ranjan-thakur/
https://ebookgate.com/product/aqueous-lubrication-natural-and-
biomimetic-approaches-1st-edition-nicholas-spencer/
This page intentionally left blank
BIOM IM ETIC ROBOTICS
Ranjan Vepa has been with Queen Mary and Westfield College in
the Department of Engineering since 1984, after receiving his Ph.D.
from Stanford University. Dr. Vepa’s research covers, in addition to
biomimetic robotics, aspects of the design, analysis, simulation, and
implementation of avionics and avionics-related systems, including
nonlinear aerospace vehicle kinematics, dynamics, vibration, control,
and filtering. He has published numerous journal and conference
papers and has contributed to a textbook titled Application of Arti-
ficial Intelligence in Process Control. He is active in learning-based
design education and the utilization of modern technology in engineer-
ing education. Dr. Vepa is a Chartered Engineer, a Member of the
Royal Aeronautical Society (MRAeS) London, and a Fellow of the
Higher Education Academy.
Biomimetic Robotics
Ranjan Vepa
Queen Mary, University of London
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo
www.cambridge.org
Information on this title: www.cambridge.org/9780521895941
© Ranjan Vepa 2009
1. The Robot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Robotics: An Introduction 1
1.2 Robot-Manipulator Fundamentals and Components 5
1.3 From Kinematic Pairs to the Kinematics of Mechanisms 12
1.4 Novel Mechanisms 13
1.4.1 Rack-and-Pinion Mechanism 14
1.4.2 Pawl-and-Ratchet Mechanism 14
1.4.3 Pantograph 15
1.4.4 Quick-Return Mechanisms 15
1.4.5 Ackermann Steering Gear 16
1.4.6 Sun and Planet Epicyclic Gear Train 17
1.4.7 Universal Joints 17
1.5 Spatial Mechanisms and Manipulators 18
1.6 Meet Professor da Vinci the Surgeon, PUMA, and SCARA 20
1.7 Back to the Future 23
exercises 24
2. Biomimetic Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1 Introduction 25
2.2 Principles of Legged Locomotion 27
2.2.1 Inchworm Locomotion 29
2.2.2 Walking Machines 30
2.2.3 Autonomous Footstep Planning 31
2.3 Imitating Animals 31
2.3.1 Principles of Bird Flight 33
2.3.2 Mechanisms Based on Bird Flight 34
2.3.3 Swimming Like a Fish 37
vii
viii Contents
7. Jacobians . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
7.1 Differential Motion 120
7.1.1 Velocity Kinematics 123
7.1.2 Translational Velocities and Acceleration 124
7.1.3 Angular Velocities 127
7.2 Definition of a Screw Vector: Instantaneous Screws 127
7.2.1 Duality with the Wrench 129
7.2.2 Transformation of a Compliant Body Wrench 130
7.3 The Jacobian and the Inverse Jacobian 131
7.3.1 The Mobility Criterion: Overconstrained Mechanisms 133
7.3.2 Singularities: Physical Interpretation 134
7.3.3 Manipulability: Putting Redundant Mechanisms to Work 136
7.3.4 Computing the Inverse Kinematics: The Lyapunov
Approach 137
exercises 140
Bibliography 335
Index 339
Preface
This book is intended for a first course in a robotics sequence. There are indeed a
large number of books in the field of robotics, but for the engineer or student inter-
ested in unmanned aerial or underwater vehicles there are very few books. From the
point of view of an aerospace engineer (flight simulation, unmanned aerial vehicles,
and space robotics) three-dimensional and parallel mechanisms take on an added
significance.
The fundamentals of robotics, from an aerospace perspective, can be very well
presented by considering just the field of robot mechanisms. Biomimetic robot
mechanisms are fundamental to manipulators and to walking, mobile, and flying
robots. A distinguishing feature of the book is that a unified and integrated treat-
ment of biomimetic robot mechanisms is offered.
This book is intended to prepare a student for the next logical module in a
course on robotics: practical robot-control design. Throughout the book, in every
chapter, are introduced the most important and recent developments relevant to
the subject matter. Although the book’s primary focus is on understanding princi-
ples, computational procedures are also given due importance as a matter of course.
Students are encouraged to use whatever computational tools they consider appro-
priate to solve the examples in the exercises.
When writing this text, I included the topics in my course and some more for
the enthusiastic reader with the initiative for self-learning.
Thus, Chapter 1 gives an overview of robotics, with a focus on generic robotic
mechanisms, whereas Chapter 2 is about biomimetic mechanisms. Chapter 3 draws
attention to the integrated representation of displacement and rotation. Chapters 4
and 5 are about direct and inverse kinematics. Chapter 6 is about the mechanics of
grasping, and Chapter 7 provides a gentle introduction to Jacobians. Dynamics of
mechanisms is introduced in Chapter 8, and path planning and trajectory following
are discussed in Chapter 9. Chapter 10 is dedicated to the techniques of transform-
ing the dynamics to simpler forms that are useful for such tasks as control synthesis.
Chapter 10 also considers the ever-important problem of the dynamics of nonholo-
nomic systems with the archetypal example of the bicycle.
xiii
xiv Preface
R. Vepa
London, UK, 2008
Acronyms
xv
xvi Acronyms
The Robot
1
2 The Robot
elbow
(can rotate about
shoulder
a horizontal axis)
(can rotate about
a horizontal axis)
lower arm
upper arm
Body or capstan
(can rotate about a vertical axis)
Bergerac must be credited both for first employing the rocket in space travel and
for inventing the ramjet in the 1650s. These inventions, however, became a reality
300 years later.
What then is a robot? An industrial robot in its simplest form is a programmable
multifunctional manipulator designed to move materials, parts, tools, or specialized
devices through a sequence of programmed motions to perform a variety of tasks.
Some of the earliest developments in the field of robotics were concerned with car-
rying out tasks in environments that were either patently dangerous for humans to
enter or simply inaccessible. A classic example is the handling of nuclear fuel rods
in the environment of a nuclear reactor. Another, the American program to land on
the Moon, was supported in no small measure by developments in robotics, particu-
larly the Apollo Lunar Module. It was a remarkable vehicle that allowed the astro-
nauts to descend softly onto the lunar surface, carry out a number of experiments
on it, and return to the command and service modules in lunar orbit. Examples of
robots today include the Space Shuttle’s satellite manipulator, the Martian rover,
unmanned aerial vehicles (UAVs), and autonomously operated underwater vehi-
cles for ocean exploration. A broad taxonomy of robots is introduced in Figure 1.2.
Although a true robot would be an intelligent, autonomous mobile entity, capable of
interacting with its environment in a meaningful, adaptive, and intelligent manner,
no such entities exist at the present time.
An intelligent autonomous vehicle does indeed possess a number of capabilities
not found in a common machine, however ingenious it may be. Yet such intelligent
machines are only evolving at the present time. Considering the impressive range
of human achievements in the last century, it is fair to say that the development of
1.1 Robotics: An Introduction 3
these machines may be effected in the not-so-distant future. For our part, the study
of industrial robotics is best initiated by considering the simplest of these creatures:
the robot manipulator. The manipulator is indeed the basic element in a walking
robot, which in turn could be considered to be a primary component of an intelligent
autonomous robot.
Machine
Dexterity Manipulator
Nullary
Unary
Binary
Ternary
Quaternary
N-ary
.
two other links by two distinct and separate joints. In a similar manner we define
a ternary link as one that is connected to three other links at three distinct points
on it by separate joints and a quaternary link as one that is connected to four other
links at four distinct points on it by separate joints. In general, an N-ary link has N
independent joints associated with it. Each link could be of arbitrary shape and the
joints not only located at arbitrary points but also of arbitrary shape.
A joint itself is then defined as a connection between links that allows certain
relative motion between links. At the outset such joints as welds, rigid connections,
or bonded joints are excluded. Joints may then be classified, irrespective of their
nature, based on the number of links associated with each. A binary joint connects
two links together, ternary joints connect three links together, and in general an
N-ary joint connects N links together.
In defining an N-ary link we associate N joints with it. It is sometimes more con-
venient to define an N-ary link as one that is connected to N other links. The two
definitions are equivalent when every joint is a binary joint. A kinematic system is
then a collection of kinematic links connected together by an associated set of kine-
matic joints. A typical robot manipulator is such an assemblage or kinematic chain,
transmitting a definite motion. When one element of such a kinematic chain is fixed,
the kinematic chain is a mechanism. From the point of view of robot manipulators,
the objective is to investigate and synthesize kinematic chains to produce mecha-
nisms that can transmit, constrain, or control various relative motions. This is the
notion of kinematic structure.
The kinematic structure of a robot manipulator must satisfy two essential
requirements:
1. First, it must have the ability to displace objects in three dimensions, operate
within a finite workspace, have a reasonable amount of dexterity in avoiding
obstacles, be able to reach into confined spaces and thus be able to arbitrarily
alter an object’s position and orientation from one to another;
2. Second, once positioned, the manipulator must have the ability to grasp with
optimum force arbitrary or custom objects and hold them stationary irrespective
of the other forces and moments they are subjected to.
Considering the first of these requirements, the primary design problem re-
duces to one of kinematic design. In general, to operate in three-dimensional space,
a kinematic mechanism must possess at least six DOFs in order to function without
any constraints. The DOFs are removed as one applies kinematic constraints to the
assemblage.
To understand the kinematic design problem, the synthesis of a component sys-
tem in a constrained kinematic mechanism so that it has a finite number of relative
DOFs may first be considered. To ensure that the complete sets of feasible designs
are finite in number, we need to consider the constraints that must be imposed. First,
we consider the number of connected planar systems that can be put together from
purely topological considerations. Considering four different links, we may identify
1.2 Robot-Manipulator Fundamentals and Components 7
limited. As a result, the workspace, which is the space of positions and orientations
reachable by the robot’s end-effector, is still severely limited. But within this limited
workspace it is the wrist that gives the end-effector the capability to point toward
an arbitrary direction. The most common types of wrists are quite analogous to the
human wrist, with three or more spatial (nonplanar) DOFs. The latter single-loop
chain is the classic four-link kinematic chain, and when one of its links is fixed, one
obtains the well-known four-bar mechanism shown in Figure 1.4. Link OA is the
driving link or the crank, and link AB is the coupler link or coupler.
The path traced by a point on the coupler link BA is referred to as a coupler
curve. Generally moving the point to different locations on the coupler would gen-
erate different coupler curves. A basic task is to be able to predict the geometry
of the coupler curve generated by the point on the coupler. Coupler curves can be
surprisingly complex. The corresponding design problem consists of synthesizing a
mechanism for generating a specific coupler curve (for example, the shape of an
aerofoil or a ship’s hull). In the generic case, a coupler curve of a four-bar mecha-
nism is an algebraic curve of degree six. Different shapes of coupler curves that can
arise from four-bar and other mechanisms have been cataloged by several authors.
It is interesting to observe that there exist at least two other four-bar mechanisms
that can generate the same coupler curve as the one generated by a four-bar mech-
anism.
B
A
Figure 1.4. A four-bar linkage, with an exam-
ple of a coupler curve generated by a point on
the coupler.
O
O'
1.2 Robot-Manipulator Fundamentals and Components 9
Table 1.5. (a) Kinematic chains with L links and J joints (shaded links are ternary
links); (b) kinematic mechanisms derived from the chains in (a)
(a)
Watt (1)
Stephenson (3)
(b)
The basic four-link kinematic chain and the chains made more complex by the
further addition of links are illustrated in Table 1.5(a). To transmit motion, a prac-
tical mechanism may be obtained by fixing one of the links in a kinematic chain.
Thus from an n-link chain one may derive as many as n mechanisms. Some of the
10 The Robot
(a) (b)
(c) (d)
mechanisms derived may be identical to each other in a kinematic sense. Such mech-
anisms are said to be isomorphic. To find the number of distinct mechanisms that
may be derived from a kinematic chain, it is essential that isomorphic mechanisms
be counted just once. The mechanisms in Table 1.5(b) are obtained by fixing one of
the links in the kinematic chains in Table 1.5(a).
Graph-theoretic representations of kinematic mechanisms are used to detect
isomorphic mechanisms. Although these representations are by no means unique,
they are useful in the analysis of the kinematic geometry of a mechanism. One
method, for example, is based on clearly incorporating the effect of fixing a link
in the corresponding graph representation of the mechanisms. Thus one obtains the
graphical pattern by substituting a fixed link in a kinematic chain by a direct pin con-
nection. Such substituted pin connections are indicated with an additional rectangle
overlaid on the pin. The resulting diagram is known as a velocity graph. Thus the
problem of detecting isomorphic mechanisms is reduced to one of detecting isomor-
phisms in the graph-theoretic representations.
A graph is characterized by a certain number of vertices or nodes, a certain
number of edges, the adjacency matrix, with entries of ones and zeros, and the dis-
tance matrix with integer entries. The matrices are both square matrices with one
row and one column for each node. A nonzero element of the former, the adjacency
matrix, indicates that the vertices corresponding to the row and column numbers are
connected. It records information about every edge except when there are parallel
edges. An element of the latter, the distance matrix, indicates the minimum number
of edges between the vertices corresponding to the row and column numbers. The
distance matrix may be computed from the adjacency matrix. Thus two graphs may
be considered to be isomorphic, provided the numbers of nodes and edges and the
adjacency matrices are identical.
Considering, for example, the seven-link chain in Table 1.6(a), there are seven
possible theoretical mechanisms one could associate it with. However, considering
the velocity graphs, there are only three possible nonisomorphic velocity graphs.
1.2 Robot-Manipulator Fundamentals and Components 11
The three nonisomorphic velocity graphs are also shown in Table 1.6, thus indicat-
ing that only three nonisomorphic mechanisms may be derived from the seven-link
kinematic chain in Table 1.6(a).
Often one may fix an alternative link in the same kinematic chain and thus
obtain a different nonisomorphic mechanism. This process is referred to as the pro-
cess of kinematic inversion. The slider–crank mechanism in Table 1.7 is obtained by
the replacement of one of the links in the four-bar mechanism with a slider. The joint
between the fixed link and the slider is a sliding joint. Alternatively the slider may
be a collar that is pinned to the fixed link and permitted to slide over the coupler.
Other mechanisms obtained by this process of kinematic inversion or by replacing a
link with a slider are shown in Table 1.7.
When a mechanism is designed to have a single relative DOF, its motion must
be completely determined by prescribing a single input variable. For design pur-
poses, it is essential that all the distinct systems possible and their structures are
identified so an optimum choice may be made. The general solution to the preced-
ing problem results in an infinite number of possibilities. Yet, when certain restric-
tions are imposed, the number of possibilities reduces to a finite set. If one restricts
one’s attention to planar kinematic systems with a maximum of eight components
or links, so that all components move in parallel planes, and imposes the follow-
ing four constraints, the number of possibilities reduces to only 19 essential distinct
systems:
1. all kinematic joints are of the same type (for example a pin joint);
2. the system should be kinematically closed loop so disconnecting any one joint
will not separate the system into two parts;
3. the system should not have a separating component whose removal would
reduce the system to two independent and uncoupled subsystems;
4. no connected subsystem consisting of two or more components should have
zero DOFs (i.e., immobile).
12 The Robot
4 4 0 0 1
7 4 2 0 2
10 4 4 0 9
10 5 2 1 5
10 6 0 2 2
When these conditions, which are not really restrictive from a practical stand-
point, are imposed on a kinematic system, the number of possible systems satisfying
all of the constraints is finite in number (Table 1.8). A number of properties of these
subsystems can be deduced from their interchange graphs. Moreover, it is possible
to synthesize large spatial kinematic systems by assembling basic modules whose
interchange graphs are not only known but also satisfy certain specific properties.
additional parameters are required for describing the motion of the body linked to a
free body in space by such a joint. The planar pair also introduces three constraints
whereas the cylindrical pair introduces four constraints. All other lower kinematic
pairs introduce five constraints on the motion of the rigid body so only one addi-
tional independent parameter is required for describing the relative motion of the
rigid body linked to another by such a pair. However, the joints with maximal prac-
tical import are the one-DOF binary joints, the revolute joint with one rotational
DOF, and the screw and parallel joints with one translational DOF each, which
could be actuated by in-parallel rotating (the first two) and translational motors,
respectively. Whereas the revolute permits a rotation only through an angle and the
prismatic permits translation only through a distance, the screw allows a screw dis-
placement involving a simultaneous coupled translation only through a distance and
rotation through an angle that are related by the pitch of the screw.
Inspired by a former professor at Karlsruhe, F. J. Redtenbacher, Franz
Reuleaux embarked on a systematic method to synthesize kinematic mechanisms.
To this end, he created in Berlin a collection of over 800 models of mechanisms
based on his book. These were cataloged and classified by another German engi-
neer, Gustav Voigt. Apart from the lower and higher kinematic pairs and simple
kinematic chains, they include a variety of pump mechanisms, screw mechanisms,
slider–crank mechanisms, escapement mechanisms, cam mechanisms, chamber and
chamber-wheel mechanisms, ratchet mechanisms, planetary gear trains, jointed cou-
plings, straight-line mechanisms, guide mechanisms, coupling mechanisms, gyro-
scopic mechanisms, and a host of other kinematic mechanisms.
An industrial robot manipulator may be modeled as an open-loop articulated
chain with several rigid links connected in series by revolute or prismatic joints,
driven by suitable actuators, Motion of the joints results in the motion of the links
that position and orient the manipulator’s end-effector. The kinematics of the mech-
anism facilitates the analytical description of the position and orientation of the end-
effector with reference to a fixed reference coordinate system in terms of relations
between the joint coordinates. Two fundamental questions are of interest. First, it
is desirable to obtain the position and orientation of the end-effector in the refer-
ence coordinate system in terms of the joint coordinates. This is known as the direct
or forward kinematics of the manipulator. Second, it is often essential to be able to
express the joint coordinates in terms of the position and orientation of the end-
effector in the reference coordinate system. This is known as inverse kinematics. In
most robotic applications it is relatively easy to find the one, usually the forward
kinematics, but it is the other, the inverse kinematics, that must be known because
the task is usually stated in the reference coordinates and the variables that could
be controlled are the joint coordinates.
Pinion
Rack
Pawl
Ratchet
C D
Figure 1.7. The planograph.
Q
B A O
1.4.3 Pantograph
The pantograph was used by James Watt in his steam engine to exactly enlarge the
reciprocating motion of the piston in the cylinder. One form of the pantograph, the
planograph, is illustrated in Figure 1.7. The links are pin jointed at A, B, C, and
D. The links AB and DC are parallel to each other, and the links AD and BC are
parallel to each other. The link BA is extended to the fixed pin O. The point Q on
the link AD and the point P on the extension of the link BC both lie on a straight
line that passes through O. It can be shown that under these circumstances the point
P will reproduce the motion of point Q to an enlarged scale; alternatively, point Q
will reproduce the motion of the point P to a reduced scale.
It is often necessary to convert rotary motion into reciprocating motion, and the
slider–crank mechanism, driven by the crank, is commonly used. Moreover, what is
often necessary in practice is for the forward stroke of the reciprocating motion to
be slow and controlled because of the need for the mechanism to do work during
this stroke. On the return stroke no work is done, and therefore it is desirable that it
be completed as quickly as is possible. Thus the reciprocating motion is necessarily
asymmetric.
A suitable mechanism to meet these requirements is the crank and slotted lever
quick-return mechanism. In this mechanism, instead of the connecting rod being
driven directly by the crank pin, it is attached to one end of a slotted link, which in
turn is driven by the crank. In Figure 1.8 the crank pin O rotates at uniform speed
about the center of rotation, C, and as it slides over the link PQ, it causes this link
Q
R
O
Figure 1.8. The crank and slotted lever quick-return
mechanism. C
P
16 The Robot
Q
R
O
Figure 1.9. Whitworth quick-return
mechanism.
C
to rock from side to side about the pivot at P. The connecting rod QR, which is
attached to the rocker PQ at Q, transmits the rocking motion to the slider at R.
The forward stroke begins with OC being perpendicular to PQ, with the slider at
O located to the right of C and rotating counterclockwise. The forward stroke is
completed when OC is again perpendicular to PQ, with the slider at O located to
the left of C. During the forward stroke the slider at O traces the arc of a circle,
which in turn subtends an angle greater than 180◦ at the center of rotation C. Thus
the forward stroke takes a relatively longer time to complete within one cycle of
rotation of the crank pin O, in comparison with that of the return stroke.
Another type of quick-return mechanism is the Whitworth quick-return mecha-
nism, as illustrated in Figure 1.9. In this mechanism, the crank pin P also slides over a
link, although in this case the link does not merely rock but completes a full rotation
about a center, O, that is different from the center of rotation, C, of the crank. Thus
the link OQ will be slowed down and speeded up alternately as the driving crank
exerts first a large moment and then a small moment about the center O. Thus the
quick-return motion is a direct consequence of the eccentricity between C and O.
The Ackermann steering gear is one of the simplest mechanisms invented for the
purposes of exerting directional control over the motor car; it is illustrated in Fig-
ure 1.10. This steering gear is based on the four-bar mechanism in which the fixed
A C
B D
K L
P
1.4 Novel Mechanisms 17
Planet Annulus
link, AC, and link KL are unequal but the links AK and LC are equal. In the sym-
metric state the parallelogram AKLC is a trapezium. In this configuration the fixed
link, AC, and link KL are parallel, whereas links AK and LC are inclined to the
vertical at same angle in magnitude to the longitudinal axis.
To steer the car, link CL is turned so the angle it makes to the longitudinal
axis increases while link KL rotates and translates such that the angle that link AK
makes to the longitudinal axis decreases. As a consequence, link AB turns through
an angle that is less than the angle of turn of the link CD. That is, the left front axle
turns through a smaller angle than the right front axle, and consequently the center
of rotation is located at point to the right of the car and not at infinity. Thus the car
is able to turn smoothly without slipping.
Figure 1.11 illustrates a typical epicyclic gear train incorporating the Sun and planet-
type gears. In the figure, the central gear wheel is the Sun, and the gear wheel mesh-
ing with it, which not only rotates about a pin mounted on the carrier link but also
revolves about the central gear wheel in a circular orbit, is the planet. The planet
also meshes with the internal annulus. When the planet carrier is held stationary
and not allowed to rotate, the planet wheel acts as an idler and the gears form a
simple gear train with the Sun wheel rotating the annulus. On the other hand, when
the Sun wheel is rotated, say, clockwise, the planet wheel is driven about its axis in a
counterclockwise direction while rolling inside the annulus and carrying the planet
carrier in the clockwise direction. Between these two extreme modes of operation
there are an infinite number of other modes in which this epicyclical train could be
operated, depending on the speed of the Sun and the planet carrier. Thus the con-
cept may be applied to a host of applications in bicycles, aircraft engines, automobile
rear-wheel drives, and a number of machine tools.
The universal joint is essentially a binary link comprising two revolute joints, with
the axes of these joints being two skewed, nonintersecting, and nonparallel lines. A
universal joint is used to couple two shafts in different planes. Universal joints have
various forms and are used extensively in robot-manipulator-based equipment. An
elementary universal joint, sometimes called a Hooke’s joint (Figure 1.12) after its
18 The Robot
inventor Robert Hooke, consists of two U-shaped yokes fastened to the ends of the
shafts to be connected (Figure 1.13). Within these yokes is a cross-shaped part that
couples the yokes together and allows each yoke to bend, or pivot, in relation to
the other. With this arrangement, one shaft can drive the other even when the two
shafts are not in alignment.
A variant of the universal joint is a gyroscope-like suspension illustrated in Fig-
ure 1.14. This is a two-DOF system in which the twin-gimbaled system takes the
form of a Hooke’s joint that couples the wheel to the motor driveshaft. The axis of
rotation of the wheel need not be aligned to the axis of the driveshaft, which could
be flexible.
(a) (b)
Figure 1.15. (a) The Stewart platform and its variants, including the hexapod; (b) a cable-
actuated version of the Stewart platform.
20 The Robot
(b)
supported at three locations on a fixed rigid base on a ground surface by six addi-
tional universal joints. No two struts are jointed to the platform or to the fixed rigid
base at more than one location. Thus the struts form a kinematic mechanism con-
sisting of a variable triangulated frame, which controls the position of the platform
with six DOFs. Although there are a number of advantages as all the actuators are
essentially linear, the kinematics and the algorithms for controlling the posture of
the platform are essentially nonlinear.
Figure 1.16 illustrates a typical cable-actuated suspension system of an aircraft
model in a wind tunnel and the spatial mechanism being developed for industrial
applications. The latter is the Eclipse II manipulator with six actuated joints, three
of which are double joints (revolute and slider) while the other three are prismatic
joints moving along a fixed, horizontal circular guide ring. The vertical guide ring is
free to rotate about a vertical axis and the platform is mounted on ball joints.
elbow
(can rotate about
shoulder
a horizontal axis)
(can rotate about
a horizontal axis)
wrist
Body or capstan
(can rotate about a vertical axis)
greater precision and flexibility. Furthermore, the robot’s pencil-thin arms and del-
icate grippers may be inserted through holes as narrow as 8 mm. Two of the hands
are used to wield surgical tools that follow finger and wrist movements made by an
operator sitting at a nearby console. A third arm carrying a miniature camera is also
inserted to track the motions of the surgical tools wielded by the other two.
The da Vinci manipulator has been used, among other tasks, to assist in a suc-
cessful kidney transplant. Its primary role was in removing the existing kidney from
the patient’s body, and conventional surgery was then used to implant the new
kidney.
Another robot, known as the Penelope Surgical Instrument Server, assisted in
the removal of a benign tumor on the forearm of a patient on June 16, 2005. Pene-
lope is now giving doctors at New York-Presbyterian Hospital an extra hand, as the
doctors might put it.
Penelope is really a robotic arm equipped with voice-recognition technology
that understands commands from doctors and nurses in the operating room. A sur-
geon can ask for any instrument, a scalpel for example, and Penelope then hands it
over.
The Unimation PUMA 560 (PUMA stands for Programmable Universal
Machine for Assembly) is a robot manipulator whose joints are all revolute joints
and it has six DOFs. It is shown in Figure 1.17 and should be compared with the
manipulator illustrated in Figure 1.1. There are three DOFs in the wrist, and a gear-
ing arrangement couples these three DOFs. The nature of the coupling plays an
important role in the derivation of the kinematics. Thus the number of joints is not
the same as the number of independent actuators. The brief technical specifications
of the PUMA 562, a member of the PUMA 560 family, are given in Table 1.9.
22 The Robot
General Axes 6
Drives DC motors
Control Numerical
Positional control Incremental encoders
Coordinates Cartesian
Configuration Revolute
Cables 5m
Work area Reach at wrist 878 mm (between joints 1 and 5)
Work volume 360◦ working volume in left-arm or
right-arm configuration
Limit joints – 1 to 6 320◦ , 250◦ , 270◦ , 280◦ , 200◦ , 532◦
Load capacity Nominal payload 4 kg
Permitted load at wrist 4 kg at 127 mm from joint 5 and
37.6 mm from joint 6
Performance Repeatability 762 ± 0.1 mm, 761 ± 0.1 mm
Maximum speed 1.0 m/s
Universal Controller Programming and Teach pendant; Virtual Device
language (VAL II) Table (VDT) using VAL II
VAL II processor Motorola 68000 32 bit
Auxiliary processor Torque processor 68000
Interface 16 input/output exp 128 Analog I/O
Serial I/O
Serial interface RS 232 RS 423
Memory buffer 64 Kb, 512 Kb, 1 Mb, 2 Mb
Battery buffer 1 year
Environment Temperature/humidity 5–40◦ C (EN 60204-1)
Interface suppression 50% at 40◦ C, 90% at 20◦ C
Power supply Incorporated 208, 240, or 380 V
Option 50–60 Hz 0.6 kW – 1 phase
clean-room version (class 10)
Weights Arm 63 kg
Controller 200 kg
SCARA (which stands for Selectively Compliant Assembly Robot Arm) shown
in Figure 1.18 has three parallel revolute joints that allow it to move and ori-
ent within a plane, and a fourth prismatic joint moves the end-effector normal to
the plane. This is an example of a manipulator that is not fully articulated and is
not anthropomorphic. Rather, the combination of the revolute and prismatic joints
allows this manipulator to operate in a cylindrical workspace. An interesting feature
is the fact that only a minimum number of joints are load bearing, depending on the
nature of the application.
This brings us to the end of our broad survey of engineering robotic systems.
The survey is by no means exhaustive, yet is most relevant to current industrial tech-
nology. Robotics plays a key role in current mechatronic systems that are the heart
of most automated engineering systems such as unmanned autonomous vehicles or
remotely operated manipulators operating in a hazardous environment, manipula-
tors designed to assist surgeons on the operating table by letting them operate with
1.7 Back to the Future 23
axes of rotation of
revolute joints
prismatic
joint
Body or capstan
(can rotate about a vertical axis)
greater precision and accuracy than ever before, and industrial automated assem-
bly lines for the mass manufacture of a range of engineering products. We are
now in position to identify four broad areas that constitute the fascinating field of
robotics: mechanisms, control, perception, and computing. In the following chapters
we restrict our attention to robot mechanisms and their control.
EXERCISES
1.1. Study the various kinematic mechanisms that were used in engineering indus-
tries in the past 50 years and indicate those that you think are exceptional or novel.
Highlight, with the aid of sketches, all the special features of these mechanisms.
1.2. Conduct a survey of industrial robot manipulators currently used and identify
at least two different
(a) anthropomorphic configurations,
(b) nonanthropomorphic configurations, and
(c) spatial mechanisms
from those discussed in this chapter. Name the salient features of each of these
systems and classify these manipulators based on their distinguishing features.
1.3. Study the various grasping and gripping mechanisms that may currently be in
use in various industrial applications and classify these on the basis of their kine-
matic geometry.
1.4. It is frequently necessary to constrain a point in a mechanism to move along a
straight line without the use of a sliding pair. Mechanisms designed for this purpose
are known as straight-line-motion mechanisms.
(a) Discuss your thoughts on how an exact straight-line-motion mechanism
may be designed. Based on your personal research, suggest two alternative
mechanisms that are able to generate straight-line motion.
(b) Often it is not necessary that the motion be an exact straight line. Dis-
cuss two or three alternative mechanisms that may be used as approximate
straight-line-motion mechanisms.
(c) Sketch three different forms of straight-line-motion mechanisms based on
the four-bar kinematic chain and the slider–crank mechanism.
1.5. You are a member of a team of engineers who have been assigned the task
of designing a walking robot, the SPD 408, capable of emulating humans in this
respect. Your first task is to establish a kinematically feasible design. Draw up a set
of requirements and establish three or four feasible kinematic architectures for a
walking robot.
1.6. Isaac Asimov’s three laws of robotics are that
(a) a robot may not injure a human being, or, through inaction, allow a human
being to come to harm;
(b) a robot must obey orders given to it by human beings except where such
orders conflict with the first law; and
(c) a robot must protect its own existence as long as such protection does not
conflict with the first or the second law.
Discuss the relevance and adequacy of Asimov’s laws, particularly in the context
of humans and robots coexisting with each other in the not-too-distant future.
2
Biomimetic Mechanisms
2.1 Introduction
The concept of biomimetic control, i.e., control systems that mimic biological ani-
mals in the way they exercise control, rather than just humans, has led to the defini-
tion of a new class of biologically inspired robots that exhibit much greater robust-
ness in performance in unstructured environments than the robots that are currently
being built. It is believed that there is a duality between engineering and nature that
is based on optimum use of energy or an equivalent scarce resource, particularly
in exercising control over actions and over interactions with the immediate envi-
ronment. Biomimesis is generally based on this concept, and it is believed that by
mimicking animals that are most capable of performing certain specialized actions,
such as the lobster on the seabed, insects and birds in flight, and cockroaches in loco-
motion, one could build robots that can surpass any other in performance, agility,
and dexterity.
A key feature of biomimetic robots is their capacity to adapt to the environ-
ment and ability to learn and react fast. However, a biomimetic robot is not just
about learning and adaptation but also involves novel mechanisms and manipula-
tor structures capable of meeting the enhanced performance requirements. Thus
biomimetic robots are being designed to be substantially more compliant and stable
than conventionally controlled robots and will take advantage of new developments
in materials, microsystems technology, as well as developments that have led to a
deeper understanding of biological behavior.
Roboticists have a lot to learn from animals. Birds have a superior flying
machine with multielement “aerofoils” capable of controlling the flow around them
quite effortlessly. They use a pair of clap-and-fling or flapping wings to execute
the modes such as flapping translation that effectively compensate for the Wagner
effect.1 The Wagner effect manifests itself as a time delay or transport lag in the
growth of lift over an impulsively or suddenly started and accelerated aerofoil.
1 Sane, S. P. (2003). Review: The aerodynamics of insect flight, Exp. Biol. 206, 4191–4208.
25
26 Biomimetic Mechanisms
To compensate for this effect, it can be shown that increasing the angle-of-attack
aerofoil by a rapid high-frequency flapping action can effectively ensure an almost
instantaneous growth in the lift developed. The rapid high-frequency flapping has
the added advantage of generating a net mean thrust. Thus a bird can effectively
take off by adopting a technique based on rapid flapping of its wings. The clap-and-
fling mechanism also has a similar consequence.
Furthermore birds use visual perception based on the concept of optic flow
for flight control. One particular control concept that is gaining importance is
biomimetic flight control based on the principles of optic flow, which is a visual
displacement flow field that can be used to explain changes in an image sequence.
The underlying principle that is used to define optic flow is that the intensity level is
constant along the visual motion trajectories. Thus there exists a displacement field
around the motion trajectories that is similar in structure to a potential flow field.
Optic flow can be then defined as the motion of all the surface elements, in a limited
view, as perceived by the visual system. As a body moves through the limited view,
the objects and surfaces within the visual frame flow around it. The human visual
system can determine its current direction of travel from the movement of these
surfaces.
The term “optic flow” has been associated with flight and has been used suc-
cessfully in the design of flight simulators. The initial research on optic flow not only
inspired many cognitive psychologists and neurologists2 with new ways of under-
standing how animals perceive motion around them but has also led to an increased
understanding of how flying animals maintain stability and perceive the world while
flying. Migratory birds use the magnetic field as a compass does to fly in the appro-
priate migratory direction! They seem to know where they want to go by using a
magnetic-field map. Light-absorbing molecules present in the retina of a bird’s eye
help the bird to sense the Earth’s magnetic field.
Currently roboticists are exploring ways to use optic flow to provide a robot with
visual perception and the associated sensations similar to those of a bird in flight.
Although researchers were able to integrate optic-flow-based artificially simulated
vision into robotic platforms and to use them to perform simple navigation tasks in
real time more than a decade ago, it was only recently that optic-flow sensing was
successfully integrated into small autonomously flying aircraft and used to provide it
with a basic degree of autonomous flight control. The methodology, which is based
on a synergy of techniques that have evolved in computational vision processing
and in biologically inspired research into human vision, has led to a better under-
standing of computer vision that may be implemented. Although computational
methods in vision processing have provided a sequence of processes that allows a
practical image to be transformed into one of a series of representations, the major
focus of computer vision has been the sensation and perception of relevant features
in the representations.
2 Tammero, L. and Dickinson, M. H. (2002). The influence of visual landscape on the free flight
behavior of the fruit fly Drosophila melanogaster, J. Exp. Biol. 205, 327–343.
2.2 Principles of Legged Locomotion 27
(a) (b)
(b)
1 2 3 (Extension) 4
reference axis
4 5 6 (Contraction) 7
Figure 2.3. Operation of an inchworm motor.
Combined with the ballistic action of walking and the energy-exchange mode
of running, in which energy is regularly transferred from potential to kinetic energy,
these models are capable of moving at high speeds irrespective of the nature of the
terrain. Walking robots are better than rolling robots in navigating through uneven
terrain, and hopping and running robots are even better than walking robots in this
respect.
The design and construction of several biomimetic robots have been inspired by
the principles of inchworm locomotion. The inchworm-like robots are composed of
three primary abilities provided by a corresponding set of subsystems. The ability to
stretch provided by a stretching linear actuator that is responsible for creating a step
is placed between two blocking elastic bodies that support the main trunk housing
the actuator. The robot is able to perform the succession of the six phases of the
inchworm locomotion described in Figure 2.3. In the initial position corresponding
to phase 1, the robot’s main trunk is supported by both the elastic bodies. In the next
phase, 2, the frontal body contracts vertically and is moved forward by stretching
of the linear actuator in phase 3. In phase 4 both bodies support the main trunk.
In the next phase, 5, the rear body contracts vertically and is then moved forward
by the linear actuator in the sixth phase, which now retracts to its initial unstretched
position. The inchworm robot then returns to its initial configuration in phase 7 and
in the process has inched forward by one step. Rapid cyclic execution of these phases
is the basis for the locomotion of the inchworm.
A typical inchworm mechanism for locomotion in the intestines consists of
two types of actuators: a clamper and an extender. The clamper is used to clamp
the device onto the wall of the intestine while the extender generates a forward
displacement.
The construction of an inchworm-like robot is feasible because of the avail-
ability of smart elastic materials that can be made to contract and expand by the
application of an external electric potential.
30 Biomimetic Mechanisms
Although the first and most important role of walking robots has been to understand
several biomechanical aspects of human walking, the applications envisaged in the
future is walking into hazardous environments, civil disaster relief, and the scientific
exploration of sites on Earth, in space, and on other planets and moons. Although
a few walking robots have been built, the engineering of a fully walking machine
capable of mimicking all aspects of human walking, capable of interaction with the
environment, and with the ability to deal with unexpected disturbance is still a long
way off.
As of now robots are able to deal with only preprogrammed activities and are
unable to naturally adapt and execute new activities as and when these are required
of them. Being able to interact with a community of robots and, more important,
with humans requires a degree of intelligence. A notable feature of current walk-
ing robots is the fact they offer very high impedance when faced with any exter-
nal disturbing force. A characteristic feature of humans is their ability offers a low
impedance to forces they intend to respond to and a very high impedance to other
forces when they so desire. This ability is somewhat akin to the concept of marginal
static stability in fighter aircrafts. Fighter aircrafts, which are expected to be very
agile and fast in responding to a pilot’s command, are for these reasons deliber-
ately designed to be only marginally stable or sometimes even unstable. This lack
of a margin of stability greatly enhances the pilot’s ability to control the aircraft. A
walking robot too must be designed so it can just about maintain its balance.
In the context of individual joints and component mechanisms the requirement
reduces to an ability to offer a low impedance to control forces and reject all disturb-
ing forces; this means that the same force when presented as a control force must be
offered a relatively low impedance whereas a high impedance is offered when it is
construed as a disturbance. It is clear that this means that the joint and component
controllers must be endowed with the ability of logical reasoning at a certain level
so they can distinguish between a disturbance and a control force.
Biomimicry can a play a significant role in the ability to develop walking robots.
Much human capability, particularly in relation to use of limbs, is acquired through
learning by imitation. The neural system uses the biologically based central pattern
generators and sensory reflexes for learning preferred patterns. The idea of having
walking robots learn by imitating a human is a feasible biomimetic concept. The fact
that both learning and adaptation can be modeled in a variety of ways facilitates the
design and implementation of biomimetic robots capable of learning and adapta-
tion. They are particularly suited for autonomous or semiautonomous tasks involv-
ing human and environmental interaction on a much larger scale than is currently
considered possible. Thus, based on a variety of imaging techniques, simulations of
human walking and other functions involving the use of human limbs can be used in
robot implementations.
Walking robots are being developed that attempt to replicate neuroanatomical
functional architectures, at different levels of abstraction, and to use them to imitate
2.3 Imitating Animals 31
— Kuinka niin?
— Sekä minä että Boguslaw luulimme, että hän heti tekisi liiton
ruotsalaisten ja meidän kanssamme, mutta hän on kovin varovainen
mies ja ajattelee vain omaa parastaan. Nähtävästi hän odottaa miten
tulee käymään, ja odotellessaan solmii liittoja preussilaisten, Jan
Kasimirille uskollisten kaupunkien kanssa. Sillävälin ovat ruotsalaiset
saaneet kokonaisen sarjan liittoja vastaansa, ja jos he sattuvat
kompastumaan Vähässä-Puolassa, niin nousevat Suur-Puola ja
Masovia heti, preussilaiset liittyvät heihin, ja saattaa tapahtua…
— Minä hölmö!
— Siksi, että niin kauan kuin hän ei valitse, niin kauan ruotsalaiset
pitävät meitä hyvänä; sama on vaaliruhtinaan laita. Jos meidän tulee
muuttaa menettelytapaamme ja ryhtyä ruotsalaisia vastaan, silloin
hän tulee toimimaan välittäjänä minun ja Jan Kasimirin välillä ja
helpottamaan minun siirtymistäni, mitä minä en muuten voisi
tehdäkään, kun olen ilmoittanut olevani ruotsalaisten puolella. Koska
hänen Podlasiessa pitäisi pian tehdä valintansa, olisi parasta, että
hän kiireimmän kautta lähtisi Preussiin, Tilsitiin, ja odottaisi siellä,
mitä tulee tapahtumaan. Vaaliruhtinas asustaa kreivikunnassaan, ja
silloin tulee Boguslawilla olemaan ylin valta Preussissa, sopiva
tilaisuus koota joukkoja ja asettua huomattavan armeijan
etupäähän… Silloin kummatkin menettelevät niin kuin me
määräämme voittaakseen meidät puolelleen, ja meidän sukumme
arvo ylenee yhä, mikä on pääasia.
— Mutta teidän ylhäisyytenne sanoi äsken, että isänmaan paras on
pääasia?
— Olkoon tai ei, mutta minä en siitä pidä, enkä tahdo, että hän
katsoo kortteihini nähdäkseen, mitä valtteja minulla on kädessäni.
Sitäpaitsi hän henkilökohtaisesti ei suhtaudu suopeasti minuun.
Varmaan hän kirjoittaa sieltä kuninkaalle, että olen joko heikko tai
epäluotettava. Se on korjattava. Sinä jätät minun kirjeeni
kuninkaalle, ja jos hän kyselee sinulta Klawanyn kahakasta, niin
kerro niinkuin asia on. Voit mainita niistä miehistä, että minä olin
tuominnut heidät kuolemaan, mutta että sinä olit saanut minut
muuttamaan päätökseni. Mitään pahaa sinulle ei tule tapahtumaan
suoruutesi tähden. Magnus-kreiviä sinun ei pidä moittia suoraan, sillä
hän on kuninkaan sukulainen. Mutta jos kuningas noin sivumennen
utelee, mitä ihmiset täällä oikein ajattelevat, niin kerro hänelle, että
Magnus ei kylliksi kunnioita hetmania ja että ruhtinas — nimittäin
minä — suree sitä. Jos hän vielä sattuisi kysymään, ovatko kaikki
vakituiset joukot hylänneet minut, niin sano, että se ei ole totta, ja
mainitse oma lippukuntasi esimerkkinä. Huomauta, että olet eversti,
niinkuin oletkin… Selitä myöskin, että herra Gosiewskin puoluelaiset
ovat yllyttäneet joukkoja kapinaan ja että hän ja minä olemme
verivihollisia. Sano, että jos Magnus olisi lähettänyt minulle vain
jonkinverran tykkejä ja ratsuväkeä, niin minä olisin jo aikoja sitten
kukistanut kaikki konfederaatit… ja että se on yleinen mielipide.
Muuten, pidä silmällä, mitä kuninkaan ympärillä tapahtuu ja kerro
näkemästäsi ja kuulemastasi, ei minulle, vaan ruhtinas Boguslawille
Preussissa, jos suinkin saat tilaisuuden tehdä sen. Mahdollisesti saat
hänen väkensä mukana lähetetyksi hänelle uutisia. Puhutko saksaa?
— Se on hyvä.
Uskottu tehtävä oli mennyt häneltä sivu suun, mutta sen sijaan
hän oli saanut odottamattoman ylennyksen.
— Voikaa hyvin!
— Jumalan haltuun!
— Entä neiti?
— Neiti on sisällä.
— Älä puhu mitään siitä! Älä puhu mitään siitä! — huusi Kmicic
kuumeisesti, — niin ei sopu väliltämme riku. Olen eksynyt tai en,
mutta älä puhu siitä! Seuratkoon kukin omantuntonsa ääntä, ja
tuomitkoon sitten Jumala… Hyvä oli, että tulin tänne ja että en
lähtenyt sanomatta hyvästi. Ojenna minulle kätesi hyvästiksi!…
Huomenna en enää saa nähdä sinua, en ylihuomenna, en
kokonaiseen kuukauteen, mahdollisesti en koskaan… Ah, Oleńka!…
Oleńka! Emmekö ikänä enää saa nähdä toisiamme?…
Yö oli tullut.
YHDESTOISTA LUKU.
Charlamp oli oikeassa siinä, että matkat ovat paras lääke henkistä
kärsimystä ja levottomuutta lieventämään. Andrzej-herralla oli
rautainen terveys, ja hänen mielikuvituksensa ja seikkailuhalunsa
kasvoivat hetki hetkeltä. Seikkailut jo ikäänkuin viittasivat hänelle,
hän hymyili niille ja ajoi eteenpäin pysähtyen vain yöksi vähäksi
aikaa levähtämään.