Formation Control Over Delaunay Triangulation Networks With Guaranteed Collision Avoidance

Abstract

This article addresses the collision avoidance of networked swarms of mobile agents. A gradient-based control scheme with artificial potential fields guides all the agents into a distance-based formation. While, in the relevant literature, unbounded forces are used to achieve collision avoidance, this article shows a way to guarantee collision avoidance that respects control input limitations. To this end, it is proposed to use Morse potential functions, and it is shown how to parameterize them appropriately. The results are extended to the case that a switching proximity communication network in form of a Delaunay triangulation is applied to the networked system. It is shown that under which conditions the collision avoidance is preserved among the overall network. The proposed method is evaluated experimentally with mobile robots.