摘要
提出在机器人足球这样的复杂、动态的环境中使用强化学习的方式与问题 ,阐述了强化学习的实现及如何处理机器学习中的常见问题 ,即延迟奖赏、探索与利用、不完整信息等 ,同时探讨了减少复杂性的若干措施 .
This paper put forward methods and problems of application of reinforcement learning (RL) in robot soccer, which is complex and dynamic. It specified the implementation of RL and that how to deal with the common problems such as delay rewards, exploration vs. exploitation, partially observability. And it also explored approaches to reduce the complexity.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2002年第3期302-305,共4页
Journal of Shanghai Jiaotong University
关键词
多智能体系统
机器人足球
复杂环境
强化学习
multi agent sysetm
robot soccer
reinforcement learning (RL)
multi agent learning
