Decentralized Reconfigurable Multi-Robot Coordination from Local Connectivity and Collision Avoidance Specifications

Abstract

Multi-robot motion planning from local connectivity specifications in unknown workspaces necessitates for reconfigurability, since real-time obstacle detection is likely to raise contradictions between collision avoidance and local connectivity specifications. Thus, a key issue in control synthesis is to allow topological changes in the communication graph in case of infeasibility, while maintaining the global connectivity of the robotic network and guaranteeing collision-free motion. Towards this direction, we propose: i) a distributed algorithm that coordinates robots, whenever feasible, based on local connectivity as well as collision avoidance specifications and ii) a novel decentralized algorithm that dynamically reconfigures the communication topology in case of conflicts with the collision avoidance specifications, without compromising the global connectivity of the communication graph. In the continuous control system, provably correct collision avoidance and robust synchronized motion are secured, employing the prescribed performance control methodology that ensures predefined transient and steady state response. Finally, simulated examples clarify the proposed methodology and verify its efficiency.