Intelligent Control Using Dynamic Neural Networks with Robotic Applications

Abstract

The intelligent control of systems with complex, unknown, and high nonlinear dynamics, such as robotic systems, chemical engineering processes and space systems, has become a topic of considerable importance during recent years. The most concurrent advances in the area of artificial neural networks (ANNs) have provided the potential for dealing with such a challenging task. In this chapter, some new schemes of dynamic recurrent neural networks (DRNNs) are proposed to design robust learning control systems for a general class of multi-input and multi-output (MIMO) nonlinear systems with unknown dynamics. The detailed structure of the DRNNs and their learning capability are first discussed. The synthesis and design methods for the purposes of regulation, tracking and model reference control are then conducted. Based on the DRNNs approaches presented in this chapter, a new torque control scheme for robot manipulators is developed. The potentials of this scheme are demonstrated extensively by simulation studies.