基于张拉整体结构的多智能体系统编队控制研究

The capability of autonomously maneuvering the overall formation has been regarded as one essential property for multi-agent systems to cope with challenges resulting from the complex, competitive and dynamic environment. Currently, most of the published results are established based on the assumptions that all the agents have to know the predefined formation parameter via network sharing, as well as the reference orientation with respect to the global coordinate system. Apart from these, it also lacks the research on rigid formation merging. Note that the tensegrity structure well studied in the field of spatial structure enjoys distinctive properties, such as self-equilibrating, easy tunability, and scalability. Motivated by these facts, we will introduce virtual tensegrity structures into dynamic formation transformation and merging problems for multi-agent systems. To be specific, we will first consider the situation when the global parameters describing the formation shape are not available in the communication network to the agents. In this circumstance, we confine the equilibrium states into the affine space of the nominal configuration by employing the self-equilibrated property of tensegrity structures, based on which we propose an integrated control scheme comprising of two globally stable observers and a controller. Furthermore, to overcome the difficulty that all of the agents have no awareness of the global reference orientation, the generic property of universal rigidity is applied with the aim of transforming the global relative localization problem into its dual problem, i.e., the local formation stabilization problem. Then we design a fully distributed controller such that the prescribed formation is achieved in a global manner. Finally, formation merging principles are developed on the basis of infinitesimal rigidity theory irrespective of the inequality constraints on the lengths with respect to cables and struts. The innovative results achieved in this project will provide new ideas for solving constrained dynamic formation transformation and merging problems, and thus will be of great significance in practical applications.

为应对复杂、对抗、多变的任务环境，多智能体系统的队形动态调节至关重要。目前多数研究还要求共享整体队形参数和全局基准朝向，缺乏刚性队形融合方法。受空间结构领域中张拉整体（Tensegrity）结构的自平衡、构型易调整、易扩展等特点的启发，本项目创新性地将张拉整体结构应用到动态编队变换和队形融合研究中。首先，在无需共享队形全局信息的条件下，利用张拉整体结构的自应力平衡体系，将队形平衡态压缩到标称位形的仿射空间，在此基础上设计出全局稳定的队形估计器-控制器；进而，在无全局基准朝向条件下，利用张拉整体结构普遍刚性一般性的性质，将全局相对定位问题转化为局部队形镇定的对偶问题，设计出依赖局部量测的全局队形镇定算法；最后，利用张拉整体结构的极小刚性理论，建立距离不等式约束下的刚性队形融合策略。本项目将为多智能体受约束条件下的队形动态变换和融合问题提供新的解决思路，具有重要实际应用价值。

机器人协同编队在多个领域作为典型应用样式受到了越来越多的关注，如无人机编队侦察、移动传感器网络、区域监视、物流运输等。在实际应用中，任务环境复杂、动态多变，因此编队系统的队形调节能力对于高效完成既定任务至关重要。目前多机器人系统整体行为调节依赖精准的队形参数传输以及无偏状态测量，缺乏高效以及受限条件下的协同机制与方法。受建筑结构领域中张拉整体结构的自平衡、构型易调整、易扩展等特点的启发，本项目创新性地将张拉整体结构应用到多机器人动态编队研究中，兼具理论与科学意义。. 本项目将张拉整体结构的代数拓扑特征映射到多机器人系统编队协同控制中，直面全局队形参数无法稳定传输，传统“广播式估计”失效时稳定协同控制问题。信息缺失下的队形参数估计形成耦合环路模式，进一步增加了收敛性分析的难度。同时，基于局部测量的控制与实现全状态镇定形成了新的矛盾，而且张拉整体普遍刚性不等式的引入都带来了很大挑战。针对以上研究背景，本项目主要进行了以下几个方面的研究：.（1）多智能体分布式协同编队控制研究.（2）受限条件下协同编队控制研究.（3）多智能体分布式事件驱动控制研究.（4）不等式约束的非光滑分布式优化问题研究.（5）时空多约束下协同规划研究.（6）空地无人系统的协同控制研究. 上述编队控制研究内容中利用了张拉整体普遍刚性的拓扑性质，并提出了多种受限条件下编队控制器，实现了高效灵活的系统协同，对于多智能体协同控制的理论发展和落地应用均具有重要意义。本项目研究取得的总体成果如下：发表SCI论文13篇，发表EI论文7篇，授权专利6项，受理专利5项，出版专著1部；特邀报告3次，获得教育部自然科学奖一等奖1项（排5），自动化学会自然科学奖一等奖1项（排2）。应用前景包括环境监测、应急救援、国防军事等。

内容获取失败，请点击重试