Intelligent control of robotic systems. (Q1417134)

This book is a result of many years of research and teaching in automatic control, intelligent robotics and robotics in general. One unifying aspect of the work is in the book, namely its interdisciplinary nature, especially in the combination of robotics, control theory, computational intelligence and soft computing paradigms. The book contains several theoretical and application aspects of neural networks, fuzzy logic, genetic algorithms and hybrid intelligent techniques in robotics. This application requires robots to have intelligent, high sensory capabilities and a higher level of dexterity compared with humans. For all these reasons we must develop new intelligent control techniques which are capable of solving these requirements. So soft computing paradigms consisting of complementary elements of fuzzy logic, neural computing and evolutionary computation are viewed as the most promising methods towards intelligent robotic systems. The authors provide a large database of knowledge using a large bibliography composed of over 400 articles. They developed a new control method based on hybrid intelligent techniques which satisfies actual intelligent robots and they give special attention in research work to the development of efficient learning rules for robotic connectionist training and synthesis of neural learning algorithms for low-level control in the robotic compliance tasks domain. The book is organized in 8 chapters. The first chapter gives us a presentation of basic ideas of intelligent control in robotics such as: autonomy in decision making for all hierarchical control levels; robustness and great adaptability to system uncertainties and environment changes; learning and self-organizing capabilities with generalization of the acquired knowledge; skill acquisition based on acquisition of expertise and experience; implementation in real time using fast processing architectures for sensor fusion and control computation. The second chapter deals with neural networks, reviews the fundamental concepts and learning principles and provides algorithms in contemporary robotics. The structure of this chapter is as follows: connectionist models with application in robotics, learning principles and learning rules applied in robotics, neural issues in robotics and efficient learning control for manipulation robots by feedforward neural networks. In the followings chapters, fuzzy logic, genetic algorithms and hybrid intelligent techniques as well as recent applications in autonomous robotic systems are presented. The most important ideas in these chapters are: fuzzy controller, synthesis of a fuzzy controller example for robotics, synthesis of genetic algorithms, examples in robotics and neuro-fuzzy algorithms in robotics. The main research contribution of the authors is presented in chapters 6 and 7 as the synthesis of new advanced learning algorithms for robotic contact tasks by nonrecurrent and recurrent connection structures. The most important themes of these 2 chapters are: fundamentals of connectionist control synthesis, synthesis of learning stabilizing control laws by connectionist structures for contact robotic tasks; synthesis of connectionist learning impedance laws for robotic contact tasks, affecting task performance and stability in robotic compliance control, a comprehensive connectionist control algorithm based on learning and a classification for compliance robotic tasks. The proposed comprehensive algorithms include the synthesis and application of two newly proposed classifiers: a pure perceptron classifier and a wavelet network classifier. Both chapters are accompanied by simulation studies to validate the proposed algorithms. The book concludes with chapter 8, which presents some interesting examples of connectionist approaches together with some supporting intelligent techniques for special robotics applications. The examples include connectionist reactive control of the soft-sensored grippers robotic assembly tasks, a special genetic connectionist algorithm for compliant robotic tasks and a connectionist approach to robotized automobiles. The book contains many examples and case studies with suggestive plots.