基于封闭曲线拼接的AUV群智能协同围捕机理与方法

研究基于封闭曲线拼接的AUV群智能围捕问题，与指挥或协商不同，群智能围捕是依靠个体行为规则，在群体层面出现的复杂、智能行为，目前对其实现机理及控制等基本问题尚不清楚。由于个体具有与全局一致性相背离的歧离趋势，导致群智能围捕难以实现，因此封闭曲线拼接的机理与实现对围捕具有关键作用。但个体歧离性与全局协同性存在根本矛盾，且封闭曲线的自组织拼接会受到本体、队友、猎物、环境、信息等多种因素制约，具有很大的研究难度。本项目主要根据隐函数理论，采用基因调控分布式控制、动态网络稳定性分析等，研究：1)面向群智能围捕的封闭曲线生成机理；2)基于封闭曲线拼接的围捕控制算法；3)个体感知能力约束下的围捕控制等。通过理论分析、数字仿真及实验验证，揭示个体歧离、全局协作之矛盾作用下的围捕机理；掌握围捕曲线拼接的控制方法；明确感知距离、感知时延作用于围捕的边界条件；实现群智能涌现的协同围捕，为工程实践提供理论依据。

The project investigates the mechanisms and approaches in multi-target trapping problem with a swarm of AUVs based on closed curve splicing. In contrast, there is no central controller for the system, novel collective behaviors emerge from individual’s local interaction. It is still not clear how to control a swarm of AUVs to entrap target. The individual level and the group level have a fundamental conflict because of individual-interest and group-interest. This makes the target hunting a difficult problem which needs to consider the number of neighbors, target and external environment. In this project, a distributed method based on radius basis implicit function and gene regulatory network is proposed to investigate 1)the mechanism of closed curve splicing with regards to target hunting; 2)the distributed method of target hunting based on closed curve splicing; and 3)target hunting with the perception constraints of detection distance and detection delay, etc. By means of the theoretical analysis, digital simulations and experimental verification, it is expected to reveal the mechanisms of closed curve splicing, distributed target hunting method and the boundaries of perception range for the target hunting, which is able to provide theoretical foundation for the future of the practical engineering practice.