Adaptive formation transition among a mobile robot group based on phase gradient

Abstract

We propose an algorithm for adaptive formation transitions of a robot group composing autonomous, nonlabeled robots. In the proposed system, one leader robot that knows the target position guides the other robots; the other robots do not have any global information. Each individual robot is equipped with a nonlinear oscillator and a simple communication system realized by LED flashlights and photodetectors. In order to control these robots, we have used phase gradients and phase waves in a manner similar to that of a slime mold (amoeba). By controlling only the leader robot, we can generate and erase a phase gradient. We propose an algorithm to control the heading directions of the follower robots according to the phase gradients. By using this algorithm, a robot group can change its formation adaptively in an obstacle region. The algorithm enables the robots to move in a tight formation in free space and dissolve this formation in obstacle regions in order to avoid deadlocks. We have not only formulated the algorithm but also developed real hardware and analyzed the system design. We have verified the proposed system by simulations and by real implementations of 12 autonomous mobile robots.