Application of analogical reasoning to pursuit of targets in robotics

The authors suggest a new approach of target tracking by a robot. Robot joints are independently controlled. This research improves other research carried on decentralized control of automated systems in the Laboratory of Automatics and Computer Science of Marseille (LAIM) (Brun-Picard (1988), Carmona (1990). Previous research showed the feasibility of the decentralization of control algorithms. The authors propose to decentralize intelligence to improve the decentralized unit autonomy. The solution consists in associating an agent with a robot joint. Each agent is considered as an intelligent agent and is able to take account of previous trial experience. The authors show the feasibility of the implementation of learning in control systems. The learning process will improve the control system efficiency.<
>