Decentralized Low-Energy Avoidance Control Framework for Multiple Mobile Agents Using Irregular Observations

Reactive, provably safe collision avoidance methods for multivehicle systems commonly require continuous monitoring of other agents and tend to yield conservative, low-efficient responses where vehicles may try to avoid other agents that pose very little threat. To counteract the conservatism of traditional methods, this article presents a cooperative, asynchronous, decentralized collision avoidance strategy that uses self-triggered control to reduce the frequency at which nearby agents are observed and velocity information to decrease the distance at which obstacles are detected and avoided. The avoidance strategy triggers new observations at irregular time intervals using the last available relative position of nearby agents. Similarly, it adjusts both the detection range at each discrete sampling instance and the distance at which a vehicle starts avoiding an obstacle using the vehicle’s velocity and the relative distance. The control strategy is rigorously proven to guarantee the safety of an arbitrarily large group of agents at all times, and the observation triggering rule is shown to avoid Zeno behavior. It is also shown via simulation and experiments that the control strategy can generate smoother vehicle’s trajectories and reduce the power consumed by obstacle detection systems.