A robust distributed convergence algorithm for autonomous mobile robots

A common x-y coordinate system shared by the robots of a multi-robot system can function as a central mechanism for solving many multi-robot motion coordination problems. Under the assumption that the robots have infinite visibility, Suzuki and Yamashita (1994) considered the problem of generating such a common coordinate system using autonomous mobile robots. The idea was to reduce the problem to a motion coordination problem of forming certain geometric patterns such as a point, line segment and circle. Motivated by possible applications to physical robots, Ando et al. (1995) then proposed a point formation algorithm for the robots under the assumption that the robots have only a limited visibility range, but the correctness proof of the algorithm was still based on the following strong, unrealistic assumptions: (1) each robot is represented as a point, and (2) there are no sensor and control errors. The goal of this paper is to analyze the performance of the algorithm using computer simulation for the case where these assumptions are not satisfied, and demonstrate that the algorithm is expected to work sufficiently well even for physical robots that cannot be modeled by a point and that are subject to sensor and control errors.