Travel Control for Robot Vehicle Using Neural Networks in Dynamic Environment

Abstract

This paper proposes a more flexible and effective obstacle avoidance travel control system based on the switching of the thought in dynamic environment where many autonomous robot vehiecles move. The proposed system plans various traveling paths by the Cascaded Neural Networks (CNN) which operates by a traffic rule or knowledge, so that robot vehiecle can smoothly move the best suited route by evaluating the energy function value of those paths. To show the effectiveness of the control system, we carried out a computer simulation of the obstacle avoidance by using four vehiecle models, and confirmed that each vehiecle can avoid other vehiecles without falling into the deadlock.