Hybrid Learning Approach for the Collision Avoidance Behavior of a Mobile Robot

In the paper, a hybrid learning approach is proposed for a mobile robot to learn the collision avoidance behavior in a dynamic environment. In the proposed approach, a fuzzy controller based on the simplified fuzzy inference is used to control the robot's motion. Each fuzzy rule is expended to have multiple possible strategies. The selection probability of strategies is updated by the learning automaton, and output parameters of fuzzy rules are updated by the steady-state genetic algorithm. Simulation results show that after learning the robot can reach the goal while avoiding collision against moving obstacles. This proves the feasibility of the proposed learning approach