CVT-based swarming techniques for convex and non-convex cases

In the multi-robot swarm control algorithm, the traditional rigid formation algorithm has low formation redundancy, and it is difficult to flexibly adapt to the external environment. In this study, centroidal Voronoi tessellation (CVT) was used to control a group of robots for forming the required flexible formations and increasing the structural redundancy and internal carrying capacity of the multi-robot system. When the CVT area is divided in a non-convex environment, the Voronoi and non-convex areas will overlap, which will cause the robot to collide with the obstacles and hinder the optimization of the cost function. In this study, constrained CVT was used to solve the problem of region overlap in non-convex environments. The proposed path planning algorithm ensured the safe movement of the robot and explored the applicability of the algorithm in different scenarios. Finally, the performance of the algorithm in the non-convex area was verified and discussed by analyzing the simulation and experimental results for the convex and non-convex environments. The results showed that the constrained CVT could control the robot to generate the formation as quickly as possible in the non-convex area, and it was easy to dynamically switch the formation according to the external environment.