基于图形势能场的群体构型填充控制方法

Swarm Configuration Filling Control Method Based on Graphic Potential Field

针对群体构型集中式控制中存在的计算量大、鲁棒性差等问题,提出了基于图形势能场的群体构型填充控制方法。该方法基于群体智能的基本思想,通过个体的简单规则组合完成群体复杂的行为活动。首先利用距离图像(Distance Transform,DT)构建形状势能场,将形状势能场和个体势能场叠加以指导群体中大量个体进行分布式运动,使群体快速收敛并大致均匀分布在预定形状中。然后利用期望最大值法(Expection Maximum,EM)进行迭代寻优,得到一个更好的填充效果。相较于传统群体构型的集中式控制,该方法具有分布式、计算量小、收敛快速和鲁棒性强的特点。解决了现实情况下群体控制易受通信速率、计算能力、个体损失等限制的难题,同时解决了群体控制过程中个体之间避撞规划复杂的问题。仿真结果表明,所提方法在凸型、非凸型和线条型等形状的群体构型填充控制方面具有良好的应用效果。

Aiming at the problems of large amount of computation and poor robustness in the centralized control of swarm configuration,a swarm configuration filling control method based on graphical potential field is proposed.This method is based on the basic idea of swarm intelligence,and the complex behavior of swarm is completed through the combination of individual simple rules.Firstly,the shape potential energy field is constructed by using the distance transform(DT).The shape potential energy field and individual potential energy field are superimposed to guide the distributed movement of a large number of individuals in the swarm.The swarm converges quickly and distributes roughly uniformly in the predetermined shape.Then,the expectation maximum method(EM)is used for iterative optimization to obtain a better filling effect.Compared with the centralized control of traditional swarm configuration,this method has the characteristics of distributed,small computation,fast convergence and strong robustness.The problem is solved that swarm control is easily limited by communication rate,computing power and individual loss in reality.The problem is also solved that the collision avoidance planning between individuals is complex in the process of swarm control.The simulation results show that the proposed method has good application effect in the swarm configuration filling control of convex,non-convex and linear shapes.