面向多地点动态集结任务的空海无人系统智能协同控制研究

空中与水面多无人平台的协同为轮船碰撞、飞机坠海、油井爆炸等海上事故中的大范围搜救探测提供了一种理想手段，具有人员伤亡少、作业效率高等显著优点。本项目采用多地点动态集结（MLDR）的统一任务框架对空海多无人平台协同搜索打捞、无人艇与多无人机在多地点协同充电等多种动态协作任务进行数学描述，并针对单个无人艇和多个无人机在协同执行MLDR任务中的协同控制问题进行研究。首先利用图论和滚动时域优化理论建立多重约束下多机-艇动态航迹规划模型，研究空海无人系统动态航迹规划方法；其次建立基于模型预测控制和深度强化学习的智能跟踪控制方法，实现机-艇协作中对动态航迹的自主跟踪；最后融合动态航迹规划和智能跟踪控制模型，建立一套面向MLDR任务的空海多无人平台智能协同控制方法。该研究可实现多机-艇自主高效地完成多地点动态集结这一类动态协作任务，为提高空海无人系统的大范围自主协同作业能力提供理论和方法支持。

The cooperative unmanned system consisting of unmanned surface vehicles (USVs) and unmanned aerial vehicles (UAVs) provides a promising approach for large-scale maritime search and rescue (MSR) tasks in marine accidents, such as ship collision, aircraft crash, oil well explosion. This cooperative unmanned system can effectively avoid casualties and significantly improve the search and rescue efficiency. This project proposes a general task framework, called multi-location dynamic rendezvous (MLDR), to describe and formulize a variety of dynamic cooperation tasks in MSR, including the cooperative search and salvage task, the multi-location dynamic charging task for the UAVs, etc. The project will perform studies on the cooperative control of an unmanned system of multiple UAVs and a single USV in terms of MLDR tasks. Based on the graph theory and the receding horizon optimization strategy, a dynamic path planning model with multiple constraints will firstly be built for the unmanned system. The dynamic path planning method will be studied for the USV-UAV unmanned system to effectively complete the MLDR tasks. Then, this project will integrate model predictive control (MPC) theory and the deep reinforcement learning (DRL) approach to propose an intelligent path tracking method for the USV-UAV cooperation system in order to realize the autonomous tracking of the planned path. Finally, an intelligent cooperative control approach, by integrating the receding path planning model and the intelligent path tracking model, will be proposed for the unmanned system. Overall, this project would enable the USV-UAV unmanned system to effectively and autonomously complete the cooperative MLDR tasks. This project will provide theoretical and method supports for improving the autonomous and cooperative ability of the USV-UAV unmanned system in conducting large-scale MSR tasks.