Fuzzy actor-critic learning automaton algorithm for the pursuit-evasion differential game

摘要

This paper presents an efficient learning algorithm to autonomously tune the parameters of a fuzzy logic controller (FLC) of a mobile robot playing a pursuit-evasion (PE) differential game. The proposed algorithm is a modified version of the fuzzy-actor critic learning (FACL) algorithm, in which both the critic and the actor employ a fuzzy inference systems (FIS). It uses the continuous actor-critic learning Automaton (CACLA) algorithm to tune the parameters of the FIS. It is called fuzzy actor-critic learning Automaton (FACLA) algorithm. FACLA is applied to two versions of the PE games. The first version considers that the pursuer interacts with the evader and will learn its default control strategy and the evader has a fixed strategy. The second version assumes both the pursuer and the evader are learning their default strategies. FACLA is compared through simulation with the FACL, and the PSO-based FLC+QFIS algorithms. Simulation results demonstrate that the performance of FACLA quantified by the learning time outperforms that of the FACL and PSO-based FLC+QFIS algorithms.