Distributed Neural Network-based Policy Gradient Reinforcement Learning for Multi-Robot Formations

摘要：

Multi-robot learning is a challenging task not only because of large and continuous state/action spaces,but also un- certainty and partial observability during learning.This paper presents a distributed policy gradient reinforcement learning (PGRL)methodology of a multi-robot system using neural net- work as the function approximator.This distributed PGRL algo- rithm enables each robot to independently decide its policy,which is,however,affected by all the other robots.Neural network is used to generalize over continuous state space as well as dis- crete/continuous action spaces.A case study on leader-follower formation application is performed to demonstrate the effective- ness of the proposed learning method.

作者: Wen Shang Dong Sun

作者单位: Dept.of Manufacturing Engg.and Engg.Management Suzhou Research Institute of City University of Hong Dept.of Manufacturing Engg.and Engg.Management City University of Hong Kong Hong Kong,China

会议类型: 国际会议

会议名称: 2008 IEEE International Conference on Onformation and Automation(IEEE 信息与自动化国际会议)

会议地点: 张家界

会议语种:英文

页码: 113-118

在线出版日期: 2008-06-20（万方平台首次上网日期，不代表论文的发表时间）