On the Minimal Cooperative Attitude in Multi-robotics Systems

Abstract

A cooperative strategy for solving conflicting situations on mutirobotics motion planning is presented. The strategy is performed with a function with minimal sharing information, which controls that such movements really assists to both robots. We show that a simple function with cooperative sense can retrieve good paths in conflicting situations. As a rule, finest movements are neccessary to avoid collisions in hard environments such as inside a tunnel. We developed a novel technique inspirated in regularisation and Markovian processes. This technique is currently implemented in a randomized algorithm with a cooperative strategy for solving conflicting situations that often arrive in multi-robot manipulator environments. Randomized Algorithms (RA) have proven their capacity to escape from local minima. On the other hand, if collaborative attitudes are carried-out by the participants in the task, the time to solve conflicting situations can be considerably reduced. Moreover, when conflicting situations take long time to be solved, high risk of exploring not good free spaces can arrive and hence the search does not often drive directly the robots to their goals. Based on the above considerations, we built a framework of cooperative RA’s aiming to facilitate the coordination of robots by establishing commitments to cooperate when conflicting situations appear, and by interacting through the exchange of data to go as directly as possible towards their respective goals.