感知约束的多智能体系统动态编队控制与协同机制研究

多智能体系统动态编队是多无人平台协同完成复杂任务的关键，对于海域协同监测等领域意义重大。弱GPS、视距遮挡等环境下智能体位置信息难以直接持续感知，感知受限下动态编队和协同机制设计面临极大挑战。本项目提出基于间歇性方位角观测的位置感知-编队控制联合设计思想，针对“定位-控制联合性能表征和协同机制设计”科学问题，开展三方面研究：①基于间歇性方位角测量，设计面向时变编队的非线性联合定位方法，避免全局坐标系下位置测量的先验要求；②基于实时相对位置估计，设计适应环境约束的感知反馈动态编队控制律，同时确保估计和控制误差收敛性；③设计面向能量受限的系统性能协同优化方法，实现闭环系统的稳定性和最优性。预期成果将为感知约束下的动态编队控制问题提供一般性解决方法，为提升海洋环境中多水下潜航器等智能体系统的自主性和智能性提供理论和技术支撑。

The dynamic formation of multi-agent systems is the key to the coordination of multiple unmanned platforms to complete complex tasks, and is of great significance to the field of coordinated monitoring of sea areas. It is difficult to directly and continuously sense the position information of the agent in GPS-denied and line-of-sight occlusion happened environments, and the design of dynamic formation and coordination mechanism under limited sensing ability is facing great challenges. This project proposes a co-design idea of localization-formation control based on intermittent bearing measurements, and conducts three researches on the scientific problem of "localization-control joint performance characterization and collaborative mechanism design": ①Based on intermittent bearing measurements, time-varying formation oriented nonlinear joint localization method is proposed, where the prior requirements of position measurement in the global coordinate system is avoided; ②Based on real-time relative position estimation, design a dynamic formation control law with sensing feedback that adapts to environmental constraints, while ensuring the convergence of estimation and control errors; ③Design an energy-constrained system performance collaborative optimization method to achieve the stability and optimality of the closed-loop system. The expected results will provide a general solution to the problem of dynamic formation control of sensing limited multi-agent systems, and provide theoretical and technical support for improving the autonomy and intelligence of intelligent body systems such as multiple underwater vehicles in the marine environment.