Multi-robot coordination based on noncooperative game approach

Abstract

The aim of this thesis is to develop a multi-robot coordination algorithm based on a noncooperative game theoretic approach. The robots are assigned the task of picking up and packaging the goods flowing in on a conveyor band. All the robots are identical and have non-overlapping workspaces on the conveyor band. The workspace of each robot is divided into a finite number of equal-sized subregions. During the course of the tasks, each robot has to first decide which subspace to pick up from and then do the picking. This cycle is repeated in a continual manner as long as the conveyor band is flowing. Optimality requires the packaging of maximum number of products within minimum amount of time. Our approach is based on noncooperative game formulation of the problem where the robots operate in a coordinated manner. Before a pickup, each robot decides on its action based on its gain estimated using its respective payoff function. Each payoff function takes its possible action and the actions of its neighboring robots into consideration. An algorithm based on this approach has been developed. The developed algorithm has been implemented and tested in a simulated environment containing a conveyor band with multiple robots. This environment has been programmed using the Webots simulation software. Its design and development has been such that different scenarios can be generated in a programmable manner via varying the initialization of the production parameters such as the number of robots, product feeding period and the speed of the robots accordingly. Our proposed algorithm has been employed in totally 27 scenarios. Results obtained from the simulations are analyzed using a variety of statistical measures.