Reward Function Design Method for Long Episode Pursuit Tasks Under Polar Coordinate in Multi-Agent Reinforcement Learning

作     者：DONG Yubo CUI Tao ZHOU Yufan SONG Xun ZHU Yue DONG Peng 董玉博;崔涛;周禹帆;宋勋;祝月;董鹏

作者机构：School of Aeronautics and AstronauticsShanghai Jiao Tong UniversityShanghai200240China Beijing Institute of Electronic System EngineeringBeijing100854China 

基　　金：National Natural Science Foundation of China(Nos.61803260 61673262 and 61175028) 

出 版 物：《Journal of Shanghai Jiaotong university(Science)》 (上海交通大学学报（英文版）)

年 卷 期：2024年第29卷第4期

页      码：646-655页

摘      要：Multi-agent reinforcement learning has recently been applied to solve pursuit ***,it suffers from a large number of time steps per training episode,thus always struggling to converge effectively,resulting in low rewards and an inability for agents to learn *** paper proposes a deep reinforcement learning(DRL)training method that employs an ensemble segmented multi-reward function design approach to address the convergence problem mentioned *** ensemble reward function combines the advantages of two reward functions,which enhances the training effect of agents in long ***,we eliminate the non-monotonic behavior in reward function introduced by the trigonometric functions in the traditional 2D polar coordinates observation *** results demonstrate that this method outperforms the traditional single reward function mechanism in the pursuit scenario by enhancing agents’policy scores of the *** ideas offer a solution to the convergence challenges faced by DRL models in long episode pursuit problems,leading to an improved model training performance.

主 题 词：multi-agent reinforcement learning deep reinforcement learning(DRL) long episode reward function 

学科分类：080202[080202] 08[工学] 0804[工学-材料学] 0802[工学-机械学] 

核心收录：

D　O　I：10.1007/s12204-024-2713-4

馆 藏 号：203127424...