面向空间探测任务的机器人群体学习进化与智能协作技术

According to practical applied background and technical requirements of space exploration, the self-management organization structure of robot swarm system and key technologies are studied in depth in order to implement learning evolution and intelligent cooperation among robots and robot teams of robot swarm so as to overcome those challenge difficulties from the environment complexity and uncertainty in space explorations. The key contents in this research project may include: the environment perception, targets recognition and tracking for robot swarm system based on multi-resource information fusion; modeling analysis and optimizing design of self-management mechanism of robot-swarm system towards cooperative missions; strategies and implementing algorithms of team learning and concurrent learning for robot swarm system; optimizing decision methods of behavior coordination and task cooperation for robot swarm system based on transfer learning; simulation analysis and experiments of learning evolution and intelligent cooperation for robot swarm system towards space exploration, and so on. Some hopeful research results will be achieved such as novel or new learning strategies and algorithms for robot swarm; some new synthetic design methods and pragmatic technologies for the self-management of robot swarm system towards behavior coordination and task cooperation in space explorations. All these research achievements will develop and enrich the intelligent control theory and new intelligent technologies of large scale robot swarm system so as to promote the development progress of robot space exploration in our country.

本项目紧密结合空间探测的实际应用背景，针对多机器人群体在探测作业过程中需要克服未知环境中诸多复杂不确定因素所导致的困难，从建立网络化多机器人系统的智能感知空间入手，研究基于多源信息融合的多机器人环境感知、目标识别与跟踪；面向合作任务的多机器人群体系统自管理机制模型化分析与优化设计；多机器人队学习和协同学习的策略与算法；基于迁移学习的多机器人群体系统行为协同与任务协作的优化决策；面向空间探测的多机器人学习进化与智能协作的仿真分析与模拟实验等内容。提出高效可行的学习策略和实现算法，以及通过感知理解与学习实现多机器人群体的功能和行为进化的分析综合方法，进而为建立有效进行多机器人群体的自管理行为协同与任务协作机制提出先进的设计方法，开发新型应用技术。本项目的研究成果将丰富和发展网络化多机器人系统的学习进化、智能协同控制理论体系，可为提高机器人群体的空间探测能力提供先进的智能理论方法和应用技术储备。

本项目针对空间探测任务的技术需求和实际应用背景，以研发面向空间探测任务的网络化多机器人群体系统的学习进化、智能行为协同与任务协作技术为主攻方向，在基于多源信息融合的环境感知、目标识别与跟踪；网络化多机器人群体系统面向合作任务的自管理机制的模型化分析与优化设计；行为协调与任务协作的优化决策方法；面向空间探测的网络化机器人群体系统的学习进化与智能协作的仿真分析与实验等方面取得了重要研究成果，在国内外重要学术期刊和重要国际会议上发表了一系列学术论文，产生了一定的学术影响。这对丰富和发展网络化多机器人群体系统的学习进化、行为协同、任务协作的智能化理论体系具有一定的科学意义和贡献性促进作用，也可为提高网络化机器人群体系统的空间探测能力提供先进的智能工程方法和应用技术储备，同时也在相关技术领域的网络化机器人群体系统的智能优化控制与管理决策方面产生应用效益，具有一定的应用前景。