Expert Fuzzy-Neuro Controller Design for Wall Climbing Robot for Decontamination of Nuclear-Power Station

Abstract

The arrangement principles and design methodology for complex control framework of Al control systems are introduced. The notions of intelligence levels with top boundary (""intelligence in large"") and the bottom boundary (""intelligence in small"") are defined. Special methodology of Al control system design for decontamination of nuclear-power station (IMPS) on the base of a wall-climbing robot (WCR) with various intelligence levels is considered. The basis of this methodology is computer simulation of dynamics for mechanical systems with the help of qualitative physics and search for possible solutions by genetic algorithm (GA). On artificial neural networks, optimal solutions are obtained and a knowledge base of fuzzy controller on WARP (Weight Associative Rule Processor) is formed. Strategy for planning, environment recognition using two types of sensors, and locomotion control to realize autonomous locomotion of the mobile robot are described. The WCR and the mobile robot for horizontal displacement with manipulators are moved in unstructured environments. Fuzzy qualitative simulation, GA and hierarchical node map, and fuzzy neural network (FNN) have demonstrated their effectiveness for path planning of the mobile robots. The results of fuzzy robot control simulation, monitoring, and experimental investigations are presented. The application of WARP to design automatic fuzzy controller for fuzzy correction motion of manipulator and WCR is examined.