Abstract

This article presents a new approach to car-like robot control for obstacle avoidance and target tracking. The proposed approach employs cooperative algorithms including artificial immune algorithms, fuzzy logic and Q-learning denoted shortly as IFQ-learning control. The article explains the artificial immune system and the proposed algorithms. The fuzzy Q-learning algorithms are also presented. The article elaborates the control design as well as extensive experimental results. Very satisfactory robot performances are achieved via the proposed IFQ-learning control. VDO clips illustrating the experiments are available on the website http://www.sut.ac.th/engineering/electrical/carg/.

Key words: Q-learning, artificial immune, fuzzy logic, robot, obstacle avoidance.