搜索

欧洲杯下单:鱼群运动启发下的微型集群机器人强化控制研究

发布时间:2024-06-03????????浏览量:10

时间:2024年6月4日(星期二)14:30-15:30

地点:经管大楼A楼 四楼第二会议室报告厅

主题:鱼群运动启发下的微型集群机器人强化控制研究(Research on Reinforcement Control of Micro-Swarm Robots Inspired by Fish School Movement)

主讲人:刘磊(欧洲杯下单欧洲杯下单)

简介:刘磊,系统科学系副教授,主要从事集群智能,复杂系统控制研究,法国国家科学研究中心访问学者,授权发明专利十余项,发表学术论文数十篇,主持上海市自然科学基金面上项目2项。

Lei Liu, Associate Professor in the Department of Systems Science, mainly engaged in research on swarm intelligence and complex systems control. He has been a visiting scholar at the French National Center for Scientific Research. He holds more than ten authorized invention patents and has published dozens of academic papers. He has also led two general projects funded by the Shanghai Natural Science Foundation.

摘要:生物集群通过个体之间的社会性交互能自组织生成宏观秩序,通过抽取鱼群运动模型能使集群机器人涌现秩序,但是自然集群秩序难以有效地被人工控制,为此引入多智能体深度强化学习方法,在鱼群硬注意力模型的基础上使用深度强化学习修正鱼群模型的决策输出,设计轨道强化网络与安全强化网络,以实现自然模型对集群机器人的可控迁移,所提控制模型在无人机群空中协作、智慧农机集群作业、物流仓储多体搬运等领域具有较大的应用潜力。

Biological swarms can self-organize into macro-scale order through social interactions among individuals. By extracting models of fish school movements, swarm robots can exhibit emergent order. However, the natural order of swarms is difficult to control artificially. To address this, a multi-agent deep reinforcement learning method is introduced. Building on the hard attention model of fish schools, deep reinforcement learning is used to adjust the decision outputs of the fish school model. The design includes a trajectory reinforcement network and a safety reinforcement network to achieve controllable transfer of the natural model to swarm robots. The proposed control model has significant application potential in areas such as drone swarm aerial cooperation, intelligent agricultural machinery cluster operations, and multi-body handling in logistics and warehousing.







 








































欧洲杯下单-搜狗指南