Mobile Robot Control Based on Dynamic Potential

An efficient method is presented for implementing mobile robot perception-action behaviors, based on time-varying environment potential field approach. First, in this paper, the concept of dynamic potential U(x,y,z,t) is proposed for representing the environment of a mobile robot, and the form of U(x,y,z,t) is deduced, and then the velocity control of the vehicle with two wheel is directly calculated by transition function T of U(x,y,z,t). Finally, the perception-action layer is successfully implemented for avoiding collision, wandering, and integrating path planning and steering control on THMR-Ⅱ (TsingHua university Mobile Robot system). Based on sonar array signals, the experimental results are given to show that THMR-Ⅱ has better reflexive function, real-time obstacle avoidance, adaptablity and robustness for complex environments.