Control of robot manipulators in joint space, R. Kelly, V. Santibáñez and A. Loria, Springer, London, U.K., 2005, 426pp. ISBN: 1‐85233‐994‐2

Abstract

Robotic systems research is a relatively new discipline that has rapidly matured over the past several decades. The success of robotics has resulted in over 800 000 worldwide industrial robots and over 600 000 household robots 1 currently in operation. Advances in areas such as controls, path planning, and intelligent autonomous decision making have enabled robotics to become a pervasive technology with impacts in many disciplines. As robotics continues to expand into new regimes, robotics researchers are required to both broaden the emphasis of current research topics and to continue to gain a deeper understanding of existing underlying principles. In particular, robot control is a research area of continued emphasis because it is an enabling technology that provides a foundation for other technologies to build upon. The dynamics and control of serial link robots is the focus of the recent textbook Control of Robot Manipulators in Joint Space 2. This textbook is contradictory to other textbooks on this topic because the authors intentionally limit the scope and generality as a means of providing a deep understanding of how conventional controllers can be used to achieve global control objectives in the joint space. The textbook has a high pedagogical value because it provides students and robotic engineers with necessary tools to analyse the stability and performance of conventional control methods using nonlinear methods through a formal theorem, lemma, and proof exposition. The textbook is divided into four parts. Part 1 includes Chapters 1–5 and provides the requisite preliminaries required for the remainder of the text. Chapter 1 provides a sample of applications for robotics with a useful reference to a number of mainstream texts and journals on robotic concepts. Chapter 2 includes a succinct presentation of fundamental nonlinear Lyapunov-based control tools that are used in later chapters. Chapters 3–5 introduce robot dynamics and associated properties. The development in Chapters 3 and 4 is generic to n-degrees of freedom (n-DOF) robot manipulators. The development in Chapter 5 provides a specific example of a two-link revolute robot testbed that is used for experimental demonstration of the concepts developed in the subsequent chapters. This specific example provides insight that is beneficial for beginning students. Parts 2 (Chapters 6–9) and 3 (Chapters 10–12) focus on setpoint regulation and tracking control, respectively, for n-DOF robot manipulators with rigid links, no friction, and ideal actuators. The development in these chapters is not intended to provide a comprehensive overview of modern control methods. The authors limit the scope of the chapters to the development of conventional proportional derivative (PD) and proportional integral derivative (PID) controllers. Exact model knowledge feedforward compensation is included with the feedback control elements for several of the controllers developed in these chapters. These chapters illustrate how familiar linear control concepts (i.e. PD and PID controllers) can be used to achieve robotic control objectives in a nonlinear control framework without linearization assumptions. By using familiar linear control concepts, the book is accessible for students and practising robotics engineers who do not have a background in nonlinear control methods. In addition, the use of conventional controllers in a nonlinear architecture helps to mitigate misconceptions and apprehensions about the use of nonlinear control methods in industrial robotic applications. That is, the textbook uses the industrial workhorse of PD and PID controllers as a framework to introduce the use of feedforward compensation for improved stability and performance. Part 4 (Chapters 13–16) builds on the development in Parts 2 and 3 by introducing adaptive model-based feedforward compensation. Development is also provided that illustrates how the velocity dependence of the derivative controller can b