Scalable Vehicle Team Continuum Deformation Coordination with Eigen Decomposition (2002.03036v1)

Abstract: The continuum deformation leader-follower cooperative control strategy models vehicles in a multi-agent system as particles of a deformable body. A desired continuum deformation is defined based on leaders' trajectories and acquired by followers in real-time through local communication. The existing continuum deformation theory requires followers to be placed inside the convex simplex defined by leaders. This constraint is relaxed in this paper. We prove that under suitable assumptions any (n+1)(n=1,2,3) vehicles forming an n-D simplex can be selected as leaders while followers, arbitrarily positioned inside or outside the leading simplex, can acquire a desired continuum deformation in a decentralized fashion. The paper's second contribution is to assign a one-to-one mapping between leaders' smooth trajectories and homogeneous deformation features obtained by continuum deformation eigen-decomposition. Therefore, a safe and smooth continuum deformation coordination can be planned either by shaping homogeneous transformation features or by choosing appropriate leader trajectories. This is beneficial to efficiently plan and guarantee inter-agent collision avoidance in a large-scale group. A simulation case study is reported in which a virtual convex simplex contains a quadcopter vehicle team at any time t; A* search is applied to optimize quadcopter team continuum deformation in an obstacle-laden environment.