Triggered Communication for Reciprocal Collision Avoidance in Multi-agent Navigation
Abdul Hanif Zaini, Lihua Xie · 2018
This paper proposes a target-relative measure-based triggered communication protocol for use with a modified velocity obstacle reciprocal collision avoidance algorithm in a system of multiple single-integrator agents with constrained velocity. Each agent is limited to communicating its position and velocity with only one other agent at a time to approximate real world systems. An event-triggered protocol is proposed which reduces the rate of communication and enables agents to reach their target positions while ensuring non-collision at all times. Additionally, our approach includes a modification to allow agents to reach zero velocity at their target positions, and methods for reducing deadlock in constraint switching and selection. The effectiveness of the method is demonstrated in simulation of five agents. It is observed that the agents can reach their targets within finite time with much reduced communication rates.