无人集群跟踪与编队中的自主协同关键技术研究

Aiming for the tracking and formation problem of unmanned swarm, the key technologies of autonomous cooperation in estimation fusion, decision planning, motion control and the combination of the above three, are studied to construct a holistic research framework and to propose a set of design strategies. For the multi-sensors of unmanned swarm subject to the mobile platforms and time-varying communication topology, the autonomously cooperative estimation fusion is studied, and the associated algorithm with convergence, consensus and optimality is proposed based on moving horizon estimation. To achieve the reconfigurable formation of the unmanned swarm according to the specific tracking task and environments, the autonomously cooperative decision planning algorithm with optimality, feasibility and uniformity is proposed based on receding horizon optimization. Considering the unmanned swarm with complex dynamics and uncertain disturbances, the autonomously cooperative control algorithm is proposed based on robust model predictive control, which can guarantee the convergence of tracking and formation, and the satisfaction of some special requirements. Finally, according to the double layered structure, the estimation fusion, decision planning and motion control are combined to construct the holistic research framework, based on which a complete combination strategy is designed. Due to providing the theoretical and technological improvements on the autonomous cooperation for unmanned swarm, and providing the solution for its practical application, the achievements of this project possess with the importance in both the theoretical and practical fields.

本项目面向无人集群的跟踪与编队问题，重点研究估计融合、决策规划、运动控制以及三者结合中的自主协同关键技术，拟构建三者综合一体的研究框架，并给出一整套设计方案。考虑无人集群中传感器平台运动且通讯拓扑时变的情况，基于滚动时域估计方法设计出保证收敛性、一致性与最优性的自主式协同估计融合算法。考虑无人集群应根据跟踪任务与环境态势而变化队形的需求，基于滚动时域优化策略设计出保证最优性、可行性与统一性的自主式协同决策规划算法。考虑无人集群的复杂动力学特性及不确定干扰影响，基于鲁棒预测控制方法设计出保证跟踪编队任务与一系列特殊需求的自主式协同运动控制算法。基于“双层结构”理念，对自主式协同估计融合、决策规划与运动控制进行一体化综合，构建出整体研究框架并给出一整套结合设计方案。项目研究成果将进一步完善无人集群自主协同的相关理论与技术，解决其集群化与自主化的实际应用问题，具有较强的理论创新性与实际应用价值。

无人集群对目标进行跟踪并相互组成编队是保障其遂行任务能力的行为基础，也直接反映着无人系统集群化与自主化的水平，它需要由各无人系统根据任务需求，通过自主协同式的信息处理、决策规划与运动控制三个功能环节来实现。本项目即面向无人集群的跟踪与编队问题，采用滤波理论、优化方法、预测控制等技术，重点研究了以下四个方面内容：1.考虑无人集群的传感器网络实时运动且通讯拓扑时变等情况，基于约束滤波估计方法，设计了能够实现传感器网络相对定位，且具有一致性、收敛性与最优性的自主协同估计融合算法；2.考虑战场环境中的一些突发事件，基于以滚动优化为主的多种方法，设计了能够实现组织结构与任务计划自适应调整，且具有可行性、统一性与最优性的自主协同决策规划算法；3. 考虑具有非完整动力学与不确定外部干扰的无人集群，基于以鲁棒预测控制为主的多种方法，设计了能够实现无人集群跟踪与编队任务，且保证集群相互之间安全避碰的自主协同运动控制算法；4.分别考虑估计与控制一体化、规划与控制一体化以及估计与规划控制一体化的方式，设计了相应的综合方案，并通过四旋翼无人机自主跟踪地面运动目标飞行进行了实飞试验验证。以上研究成果解决了无人集群跟踪与编队中的一些关键理论与技术问题，进一步推进了无人集群的实际应用，具有比较重要的理论与实用价值。.本项目共计发表学术论文27篇，其中SCI检索12篇（JCR一区5篇、JCR二区3篇）、EI检索12篇、核心期刊论文3篇，出版图书著作1部，授权国防专利2项、待授权国家发明专利1项，培养博士后1名、博士生4名、硕士生3名。