Experimental investigation of distributed navigation and collision avoidance for a robotic swarm

Abstract

In this study, we focus on the navigation of a robotic swarm and perform an experimental investigation. A decentralized control method using only local information has been proposed in many previous studies. However, because global coordinates are used and control input is calculated on a single PC in experiments, the system is regarded as partially centralized control. Thus, we first realize machines that are suitable for the distributed navigation control method. In conducting experiments using these machines, we encountered two problems. The first was loss of connectivity caused by the target’s coordinates being momentarily out of the sensing range. The second was collisions between robots caused by the target being in the blind spot of the camera’s field of view and loss of sight of the target. Note that each robot had a target and moved by following it. To solve these problems, we propose a method that removes unexpectedly output coordinates and adds rotational motion to each robot. The experimental results demonstrated that we achieved distributed navigation control with collision avoidance in an actual machine experiment.