Collecting a Flock With Multiple Sub-Groups by Using Multi-Robot System

Abstract

In this letter, we present a distributed approach to automatically collect a flock with a group of robots. This collecting problem is challenging due to the fact that the flock size is usually much greater than that of robots, meanwhile the local sensing range of robots may not ensure a globally successful collecting. Existing literature for collecting flocks assume that the flocks are coherent, however, this limits the practicality of these approaches especially in the case where there are multiple sub-groups. To address these issues, we relax the assumption and propose a density-based strategy, which can drive the robots to move towards the edge of the flock in an edge-following behavior and ultimately encircle the flock. Once encircled, a shrink mechanism is used for squeezing the flock into a tight cluster. Collecting a single cluster and collecting a flock with multiple sub-groups can be achieved by using such edge-following behavior in combination with shrink mechanism. The lower bound on the minimum number of robots required for successfully collecting a given flock size is also theoretically investigated. We validate the performance of our approach via numerical simulations and the number of robots required for a successful collecting operation is validated by using statistical simulation results for a range of flock sizes.