摘要
研究了基于强化学习的多机器人学习社会合作行为的问题.通过定义的分配布尔矩阵,对参与任务者进行奖励回报,并综合基础行为,生成状态到行为的新的映射,形成高级的群体合作行为,使得团队作为一个整体受益.讨论了学习社会行为的可行性和必要性,并采用强化学习方法,给出了多机器人传接合作搬运的详细算法实现.
Learning social cooperative behavior in multi-robot was introduced, which assigns reinforcement by defining a boolean matrix, to agents if they have ever participated the accomplished task, and synthesizes basic behaviors to generate a new mapping from states to behaviors, and forms higher-level group cooperation, which will benefit the group as a whole. The possibility and necessity of learning social behavior were discussed, and applying reinforcement learning and the above idea to multi-agent's learning relay cooperation in convey.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2005年第8期1331-1335,共5页
Journal of Shanghai Jiaotong University
基金
国家自然科学基金资助项目(60105005)
关键词
强化学习
多机器人
传接合作
社会行为
reinforcement learning
multi-robot
relay cooperation
social behavior