Self-rearrangement in a group of differentially-driven mobile robots changing shape of the formation

This paper presents control for a large group of differentially driven mobile robots that is able to efficiently self-rearrange if desired shape of the formation changes. Distributed Goal Assignment (DGA) is used to speed up the convergence and reduce number of inter-robot collision avoidance interactions. In this approach, pairs of agents uses an analytical criterion to give a verdict on whether swapping goals between them will result in a better system performance. Peer-to-peer communication is used to allow decision-making process and possible goal swapping. Limited range of the both on-board sensing system and the inter-agent communication is considered. This is the main novelty of the paper. The effectiveness of the proposed algorithm is illustrated by numerical simulations conducted on a large group of non-holonomic mobile platforms. Formation shapes were designed to ensure that the new assignment of agents to goals is non-trivial. In addition, a scenario is shown in which a significant number of agents, performing steady-state tracking, change goals assignments due to the appearance of a new agent in the close vicinity of the reference trajectory.