异构智能群体跟踪控制

Tracking control of non-uniform swarm

对智能群体(swarm)跟踪领航者(leader)进行了研究.提出了一个混杂控制方法,使智能群体中各个智能体的运动状态渐近地收敛到时变的参考轨迹.用图论模型表示智能体之间的相互作用或通信关系,基于异构的智能群体,建立了动态的有向网络模型.运用矩阵分析法及传统的控制理论进行了稳定性分析,从理论上证明了系统的跟踪误差渐近收敛到零.给出了智能群体跟踪时变参考轨迹及到达集合点(rendezvous)的计算机仿真示例,仿真结果证实了该方法的有效性.

Tracking a leader agent by swarm was studied. A class of hybrid control laws for swarm was proposed to make the motion states of all agents in the swarm asymptotically converge to the time variant reference trajectory. The interaction or communication relationship among agents was modeled by graph theory. Based on non-uniform swarm, a dynamic model in directed networks was constructed. The stability was analyzed by using matrix analysis and classical control theory. The main conclusions that the tracking error of the whole system would converge to zero asymptotically were proved theoretically. The simulation examples were also given, showing that the swarm could track a dynamic trajectory and reach a rendezvous respectively.