Overall, this series of books is impressive, although not all the books have the same value. The books were written in French between 1981 and 1986 and later translated into English. Some are more outdated, or better written, than others. The authors vary, but some wrote or co-wrote more than one book in the series.
The series attempts to cover all aspects of robotics while devoting enough attention to both basic foundations and research problems. A volume of indices and bibliographies helps to unite the series; this is especially useful since some topics are covered in two or more books written by different authors at different times. The books are not designed to be textbooks, though some of them could be used for specialized courses. Since they form a series, it is unclear why there are a lot of repetitions (most of the books begin by defining a robot) and why no attempt was made to use a consistent and unified notation.
The books cover most of the relevant topics, but not always in sufficient depth. The balance between depth and breadth of coverage is not easy to maintain, and the authors often prefer breadth. I have the impression that when the series was planned, the size of the material was not clear. For instance, the original plan included only one book on telemanipulation; the two thick books that now exist make it one of the best-covered topics.
Writing about robotics is difficult because people working in the field often have quite different backgrounds. One problem with this series is that it makes no attempt to specify the intended audience for the books or the background needed to understand them. Throughout the series, the authors present an unusual combination of fundamental principles and specific technology, which sometimes looks odd. References to specific technologies and research projects are often outdated, but this is to be expected from a collection of this size that was written over a number of years.
Volume 1: Modelling and Control
This book was one of the first two volumes in the series (it was originally published in 1981). It is quite short and disappointing.
The book covers the kinematics and dynamics of stationary articulated robots and includes a short description of robot programming. It contains the following chapters:
:9N(1)Definitions and Objectives
(2)Structure and Specification of Articulated Robots
(3)Articulated Mechanical Systems: Determination of Kinematic Elements
(4)Calculation of Robot Articulation Variables
(5)Positional Control of Articulated Robots
(6)Speed Control of Articulated Robots
(7)Articulated Mechanical Systems: The Dynamic Model
(8)Dynamics and Control of Articulated Robots
(9)Learning and Trajectory Generation
(10)Task and Performance of Articulated Robots
The book also contains conclusions, references, and an index. The notation is nonstandard and unnecessarily complicated. The author uses many abbreviations, such as S1 for sin &thgr;1, which he defines within the text. These definitions are hard to find, which makes it difficult to use this book as a reference. The problem is especially critical since the notation is very complex--most symbols have a subscript, a superscript, or both, and many have additional appendages such as underscores and stars. The descriptions are very formal in the mathematical sense, so the reader needs a clear understanding of the meaning of the notation.
Chapter 2 presents several systems for giving a formal description of a robot. These systems are not consistent, yet they are sufficiently complicated to require detailed study. It appears that the notation used in the rest of the book is not based directly on any of these descriptions, though it does use the graphical elements in Figure 11.
It is often hard to see what point the author is trying to make. For example, in Table B on page 110, many of the “methods to reduce the time” are generic to any problem in any scientific or computational domain. The author discusses a few of the remaining methods, but they do not seem to accomplish much. For example, the conclusion to chapter 7 (p. 105) says only that it is difficult to produce a dynamic model of a robot. This does not speak very highly of all the mathematics developed in that chapter.
Some of the statements in the book are less than helpful. For example, chapter 1 opens with a “definition” of robotics: “the theory and practice of automation of tasks which, because of their nature, were previously thought to be reserved for man alone.” Since the author does not define “tasks” until four pages later, this definition does not even specify the domain about which robotics is concerned. Every advanced monograph in computer science talks about some development that was previously “thought to be reserved for man alone,” so this phrase conveys no meaning.
The mathematical descriptions are hard to follow, not only because of the complex notation but also because of confusion in the text. For example, the author states that the underdetermined linear system (6-63) has m − n + 1 linearly independent solutions. The following page states that, for the specific example being discussed on pages 86–87, the three solutions listed on page 87 are independent and hence “form a basis for the solution of equation (6-17).” Since the generalized inverse is a basic concept of linear algebra that the reader is likely to have seen, it would be useful to use terms that are consistent with linear algebra terms. The use of the terms “linearly independent” and “basis” seems to imply that the solution space (6-53) is a linear subspace, which it is not; rather, it can be expressed as a constant vector (particular solution) plus any vector in the nullspace of G.
Volume 2: Interaction with the environment
This book, originally published in 1981, shows its age. The overall quality is disappointing. The artificial vision part, which takes up more than half the book, is often confusing and incomplete. A footnote (p. 103) tells the reader that the area of vision has changed a lot since the book was published, but my main problem with the book is the lack of material that was known when it was originally written.
The first chapter is a short introduction to what a robot is and how it can be used. This chapter is fine, but some of the material has already been covered in the first book.
The second chapter covers interaction with gravity and load. The author mentions underwater applications when he describes the interaction of a robot with its medium, but he says very little. The material is an extension of what was covered in the first book and could have been placed there.
The three chapters on stress, tactile sensing, and proximity sensing are aged but still valid even though they are somewhat short. The chapter on proximity sensing is disappointing because it is so short that it contains little useful information.
The remaining four chapters are devoted to computer vision with an emphasis on industrial applications. Overall this is the most disappointing part of the book, perhaps because so many better books on the subject are available. The discussion of methods for depth reconstitution in chapter 6 does not even mention stereo vision and vision from motion. Range finding is mentioned only in passing; the most important topic seems to be where to position the camera. Chapter 7 covers filtering, coding, and extraction of features, but unfortunately does not even mention zero-crossings, Gaussian filtering, convolutions, or “Mexican hat.” The feature extraction portion is modeled after industrial vision systems suited to 2D processing of isolated objects. Chapter 8 covers various template matching and pattern recognition approaches to classification that are clearly applicable only to 2D objects and to images where objects are not occluding each other or overlapping. This is the case in most industrial vision systems, but a separate chapter covers vision systems for industrial robots. Assuming that all the problems in using vision for robotics can be solved with the simple approaches taken by early industrial vision systems is a disservice to the robotics community.
Strangely, the chapter on industrial applications includes the Waltz labeling methods for line drawings and a little bit of stereo vision. I do not know of any industrial application of the contour labeling methods. The chapter describes various vision experiments, most of which were performed by research groups and not used in production. There are notable omissions, such as GM’s CONSIGHT system. Clearly missing is a discussion on the importance of controlling the environment (light conditions, reflectance, and colors).
Volume 3A: Teleoperations and robotics: evolution and development
This volume is subtitled “Part I--Introduction” and consists of the following chapters:
(1.1)Background
(1.2)Classification of Teleoperation Systems
(1.3)Mechanized Master-Slave Telemanipulators for Radioactive Materials
(1.4)Motorized Telemanipulators with Open-Loop Control
(1.5)Bilateral Servo Manipulators
(1.6)Transporters and Vehicles
(1.7)The State of the Art in Teleoperation at the Time of the Introduction of Computer Science
It also contains references and an index.Volume 3B: Teleoperations and robotics: applications and technology
Part II, “The Contribution of Computer Science,” contains three chapters:
(2.1)A Description of Teleoperation Systems
(2.2)The Operator Substitution Function by Computer
(2.3)The Use of Computer Feedback to the Operator
Part III, “Performance and Man-machine Interface,” is divided into two chapters: Part IV, “Applications of Teleoperation,” covers the following:(4.1)Nuclear Applications
(4.2)Underwater Applications
(4.3)Space Applications of Teleoperation
(4.4)Medical Applications of Teleoperation
(4.5)Industrial Applications of Teleoperation
(4.6)Applications in Security and Civil Protection
(4.7)Conclusion
This volume also contains a bibliography and an index. The two books are comparatively recent (the first was originally written in 1984, the second in 1985). They are long and detailed and provide excellent coverage of the history of manipulators and teleoperators. An incredible collection of pictures and drawings completes the books, which are a pleasure to read.
Volume 4: Robot components and systems
This book was written in 1983. It is relatively long and contains a good collection of material on robot components.
The book consists of nine chapters:
(1)Introduction
(2)Arm Structures
(3)Direct Current Motors
(4)Stepping Motors
(5)Pneumatic and Hydraulic Actuators
(6)Transmission Systems
(7)Robot Control
(8)End Effectors
(9)Energy Sources, Effects of Gravity and Vibration, and Operational Peripherals
The authors provide references and an index. Chapters 2 and 7 contain material already covered in the first book of the series. The other chapters are long and will certainly appeal to anybody interested in designing robots.
Volume 5: Logic and programming
This short book starts (again) with some definitions about robots. It was written in 1983, after most of the interesting robot programming languages were designed.
The book is divided into five chapters and three appendices:
Chapter 1Introduction
Appendix 1 Definitions for Mechanics, Geometry, Control and Programming
Appendix 2 Designation of Geometrical Axes and Movements
Chapter 2 Maximal Effort Manipulators
Appendix Concepts and Rules for the Development of GRAFCET
Chapter 3 Servocontrolled Robots
Chapter 4 Programming Languages
Chapter 5 CAD Robot Programming
As usual, the book contains references and an index. Chapter 2 covers maximal effort manipulators (pick-and-place robots) with bang-bang control and describes GRAFCET, a method for describing the operation of robots that is popular in France. Chapter 3 includes a short description of motors and robot components even though more detailed similar material appears in volume 4.
Chapter 4 covers programming languages. The authors say very little about the fundamental issues in robot programming: what makes programming robots difficult, why no standard programming languages for robots exist, and areas of future development. Most of the material is a review and comparison of various of the most popular programming languages. This choice tends to provide the reader with unimportant details. Nice tables compare the features of different robot programming languages, but the authors do not provide examples of complete programs written in different languages. The chapter on CAD robot programming covers important issues of geometric modeling.
The index is very short and too many items are missing. For instance, the robot programming languages described in the book do not appear in the index.
Volume 6: Decision and intelligence
This volume covers the following topics:
(1)Prospects for Knowledge-Based Robots
(2)Robots and AI: Parallel Developments
(3)Expert Systems and Knowledge-Based Languages
(4)Production-Rule Expert Systems
(5)Introduction to Search Techniques
(6)Heuristic Graph Searching
(7)AND/OR Graphs
(8)First Order Predicate Logic
(9)Future Prospects for Knowledge-Based Robots
It also contains a bibliography and an index. Most of the 202 pages of the book discuss artificial intelligence (AI) theory that has nothing to do with robotics. Readers will be better served by real AI books, which are more accurate and up to date. Most of the parts of AI relevant to robotics are missing, including intelligent scheduling, planning, qualitative physics, truth-maintenance systems, blackboard-based architectures, frame-based systems, reasoning about space, reasoning about time, and reasoning about processes.
Even though the book is recent (1986), it does not mention recent applications of expert systems in manufacturing. The two chapters on expert systems are sketchy, yet they contain redundant material. The authors devote less than two pages to certainty factors in expert systems even though almost every expert system shell includes them. Some of the material is imprecise (for instance, the use of variables in MYCIN is more limited than is described here) or obsolete (many of the expert systems described are no longer in use).
Three chapters discuss search (65 pages) and one covers logic (30 pages). The chapters on search are very formal, with proofs of many of the properties of the algorithms described. The notation is confusing and unnecessarily cumbersome. These chapters include many mistakes, some of which result from poor proofreading, some from missing definitions, and some from incorrect steps in algorithms. In addition, the authors use some terms incorrectly. The names depth-first and breadth-first search are always used to describe uninformed search, rather than as the authors use them here. What the book calls depth-first search is known everywhere else as a form of best-first search.
The authors devote too many pages to AND/OR graphs, a topic of little practical importance in AI. They give lots of details of uninteresting algorithms. Some material is referenced but not really described. For instance, where does the book introduce the STRIPS type of operators? The authors mention them on page 124 but it is hard for the reader to figure out what they are.
In conclusion, I cannot recommend this poor book to anyone interested in AI, much less to anyone interested in learning what AI could do for robotics.
Volume 7: Performance and computer-aided design
This book covers the following topics:
(1)The Rigid Body: Configuration and Motion
(2)Definition and Measurement of Precision: Operation-Performance Relationship
(3)Introduction to Dynamics
(4)Structure of Robots: Geometrical and Mechanical Constraints
(5)3D Models of Static Performances
(6)Analysis, Generation, and Optimization of Movements
(7)The Dynamic Behaviour of Robots: Characterization and Use of Models
The author has provided references and an index. The book is concerned with mathematical modeling and graphic representation of robot performance. It considers three aspects of performance: geometrical and kinematic, static, and dynamic aspects.
The book is well written, well illustrated, and detailed. It compensates for some of the material missing or poorly described in the first book. Chapters 1 and 3 discuss material covered in volume 1, but handle it much more crisply and clearly. Chapter 6 gives some material on legged locomotion, a nice addition to a series that is primarily on articulated arms and not on navigation.
Volume 8: Indexes and Bibliography
This volume contains the subject bibliography, subject index, and author index for the entire series. It has some errors, such as missing references and incorrect pages, but overall it is a useful book that helps in finding topics spread across more than one volume of the series.
The subject bibliography is divided into many groups (such as active sensors, AND/OR graphs, and artificial intelligence), which appear in alphabetical order. The author index contains references to papers cited in the text or listed in the bibliography of all the books in the series. This is a very good idea, but many references are missing and some page numbers are wrong.
