Robot Behaviour – Design, Description, Analysis and Modelling

Citation

Reina, G. (2012), "Robot Behaviour – Design, Description, Analysis and Modelling", Industrial Robot, Vol. 39 No. 1. https://doi.org/10.1108/ir.2012.04939aaa.011

Robot Behaviour – Design, Description, Analysis and Modelling

Article Type: Book review From: Industrial Robot: An International Journal, Volume 39, Issue 1

Ulrich NehmzowSpringer2009$69.95252 pp.ISBN: 978-1-84800-396-5www.springer.com/computer/ai/book/978-1- 84800-396-5

Robot Behaviour: Design, Description, Analysis and Modelling is intended to introduce students at the advanced undergraduate and graduate levels to fundamental concepts in scientific research, experimental design and procedure, and system analysis. These methods and tools are applied specifically to the area of mobile robotics. Nonetheless, this textbook can be of great interest to a wider audience of PhD students and researchers in robotics, artificial intelligence, cognitive science, biology and psychology. It is written to complement and build on the author’s earlier book: Scientific Methods in Mobile

Robotics. Ulrich Nehmzow is a researcher well known in the robotics community, with many years of teaching experience focusing on robot learning, novelty detection, and robot-environment interaction. He pioneered the concept of a “scientific method for robotics” that is well explained in Chapter 3 and supported throughout the book. As robotics spreads across many different types of automation and industries, entering daily life, reliable design tools, and rigorous methods are primarily required that address some of the basic issues:

• •

  How can the performance of a mobile robot be assessed and objectively compared?

• •

  How can experiments in robotics be appropriately designed?

• •

  How can robot-environment interaction be modelled, analyzed, and faithfully simulated?

This textbook provides some of the concepts and techniques necessary to get a formal research methodology and, therefore, avoids the systematic distortions of experimental results biased bywhat it is “expected” to be achieved, or wrong guessing of the reasons behind robot misbehaviour. This approach to scientific experimentation is articulated in threemain pillars: description,modelling, and analysis.

Description refers to quantitative evaluation of experimental results; that is, measurable indicators that allow precise comparison between different experimental observations or different robotic systems. Statistical methods such as those discussed inChapters 4 and 5 can serve this purpose. Chapter 4 presents a wide range of statistical methods, commonly used in biology and psychology that are applicable to mobile robotics. These methods establish whether two distributions differ or not, whether a sequence is random or not, if there is correlation between two sets of numerical data, or if there is a correlation between two sets of categorical data. Chapter 5 deals with the theory of dynamical systems, where a quantitative description of a robot’s behaviour is described capturing the dynamics of its interaction with the environment within a framework now known as “deterministic chaos”. The analysis of the dynamics of robot-environment interaction typically begins by reconstructing the robot’s phase space through time lag embedding.When the phase space has been reconstructed, it can be described quantitatively; for 104 instance, by estimating the Lyapunov exponent, the attractor’s correlation dimension or the prediction horizon beyond which the system becomes unpredictable.

Modelling is the second pillar, because a model captures the essence of the modelled behaviour. This is done from Chapters 7 to 10 where it is shown how system identification techniques can be successfully employed to obtain highfidelity models of the robot-environment interaction. Three different approaches are explained: neural network, ARMAX, and NARMAX along with case studies where these methods are applied in practice. Analysis comprises the comparison between the behaviour of the original systemand its model and how the validity of the model can be ascertained. In Chapter 11, some static and dynamic techniques are presented that address this issue. Nonetheless, the precise meaning of “identical behaviour” needs further research discussion and it is still an open problem. In some chapters of the book, numerical examples of methods and algorithms discussed in the text are included using the mathematical programming package Scilab. This is a powerful, yet intuitive mathematical programming language that is free for personal use. However, the examples given in this book require few changes to run on other mathematical programming languages, such as Matlab. The result is an excellent textbook, which provides amodern, comprehensive, and modelling-oriented analysis of the traditional mobile robotics topic.

Giulio ReinaUniversity of Salento, Lecce, Italy