Distributed Gossip-Triggered Control for Robot Swarms With Limited Communication Range

Robot swarms usually work in uncertain and unstructured environments. It is challenging to control the robot swarm to accomplish multiple tasks in the shortest possible time when the task operation duration is unknown, and the communication network is time varying and possibly disconnected. To this end, a new distributed gossip-triggered control (DGTC) method is proposed. Robots mimic human gossip behavior to disseminate and aggregate local information while triggering decisions to reduce task conflicts and system costs. The swarm dynamics, including discrete decisions and continuous trajectory, are formulated as a giant switched system. On this basis, the system stability and convergence of the DGTC are proved, and a criterion to balance the convergence rate and communication burden is given. More importantly, robot swarms can balance the workload between individuals by self-regulating the number of active agents to match the remaining tasks, thus ensuring a tight schedule without any time knowledge. Simulation results demonstrate the advantages of DGTC in generating tight schedules, and experiments conducted with Mecanum cars validate the effectiveness in dealing with time uncertainties.