A Distributed Coordination Model for Multi-Robot Box Pushing

Abstract

In this work, we approach the problem of a box transport by the robots that are requested to bring a box from an arbitrary initial to some preassigned target position. The robots are modeled as rational collaborative autonomous agents working on their movement coordination in a discrete state space. The proposed method for coordinating the multi-robot system is a two-level control model in which on the upper level, a Virtual Structure (VS) agent calculates offline the box trajectory from the initial to the goal position with incomplete obstacles data and informs the robots of the nominal route. The robots on the lower level, on the basis of the preassigned positions and the data about the unforeseen obstacles sensed by limited range sensors and exchanged among all the robots, individually optimize their local movement objective functions to mutually push the box following a preassigned trajectory in a safe manner. When the robots, due to the unforeseen obstacles, move from the trajectory defined by the VS agent too far away, more than a predefined distance, they contact the VS agent for the recalculation of the same. The results of the proposed coordination model's simulation in 2D environment are described.