Distributed asynchronous rendezvous planning on the line for multi-agent systems

Abstract

Multi-agent systems have become increasingly significant in various application areas such as search-andrescue, exploration, surveillance, and assembly. In this study, we focus on the asynchronous autonomous rendezvous planning in multi-robot (i.e. multi-agent) systems. The objective is that the robots located in linear environments to gather rapidly at a previously unknown rendezvous location. We consider that no robot knows the positions of the other robots and its own global position. Furthermore, the robot does not know its initial distance to any other robot. Our focus is on the asynchronous case where it is not required the robots to start executing the algorithm simultaneously. We propose and develop a rendezvous planning algorithm, namely MAR, that combines distributed coordination and online motion planning. We theoretically analyze the performance of our algorithm and show that it has a constant competitive ratio. Our extensive simulations demonstrate the performance and scalability through the analysis of the key performance metrics of interest including competitive ratio, distance traveled, total time, number of rounds, and number of meetings. Additionally, we demonstrate the performance and applicability of our algorithm MAR through experimental analysis in a realistic robotic simulator.