Distributed Task Planner for a Set of Holonic Mobile Robots

Abstract

Most distributed systems use a central supervisor computer to select the task that each robot must accomplish. This central computer has all the information needed to obtain an optimal assignation of tasks. In this article a method for task distributed selection over a multiple mobile robot system is presented. The proposed planning system lets each robot select a task to be accomplished and it has to negotiate with the others in order to finish different tasks. In this sense, each mobile robot may be considered a holon. A supervisor computer centralises the reception of new targets to reach and shows the manoeuvre of the system on the computer screen. Nevertheless the decision of which robot goes to each target is taken by the mobile robots themselves. In the mobile robots lab at the U.P.M. we have two mobile robots although simulation results are also shown for a set of a higher number of robots. To study the performance of a multirobot system more than two mobile robots are needed. Simulation of each robot is carried out on a different computer. We simulate the sensor measurements and the mobile robot movement. The path planning, task selection algorithm and task negotiation process are the same in the simulation and in the real robots.