A NOVEL PATH PLANNING APPROACH FOR MULTI-ROBOT BASED TRANSPORTATION

Ting Wang, Cristophe Sabourin, and Kurosh Madani

Keywords

Multi-robot team, rigid formation, path planning, logistic transportation, machine-learning, Q-learning

Abstract

A path planning approach has been proposed for a team of robots which formed a rigid geometry formation to navigate in the unspecified environment. It may benefit for the transportation in the logistic application. In the proposed approach, the integral rigid formation is considered as a virtual structure with virtual rigid links between robots. The path planning is based on a two-level control strategy, where the higher level controls the formation and the lower one is in charge of holding the shape of the formation. The demonstration of the proposed concept’s viability has been done on the basis of both simulation and experimentation involving real robots. The simulation is performed accordingly to a scenario involving a multi-robot team including three two-wheeled robots. In the considered scenario, the three robots form a rigid formation and navigate in the constrained structure in an unmatched environment (unknown by the group of robots at the initial time step). The real robots’ based demonstrator involves two two-wheeled robots within a “rigid object’s transportation scenario. The considered scenario assumes that the two involved robots transport a voluminous rigid object (materialized by a long paper box in our experimental demonstration) from an initial location to a final location. Both the simulation and the experimental evaluations show that our path planning approach can be successfully applied to the logistic transportation.

Important Links:

Go Back