An Integrated Approach To Solving the Real-World Multiple Traveling Robot Problem

In this study, we analyze the real-world Multiple Traveling Robot Problem (MTRP) and propose an integrated approach to solve this problem in real time. MTRP is a generalization of the well-known Multiple Traveling Salesman Problem and is solved by a multi-robot team. In the MTRP problem definition, target locations should be visited by the robots. Since the real world is beyond the control of robots, in most cases, the OR solutions are not directly applicable due to either robot hardware/software limitations or environmental dynamics. In this paper, we analyze the MTRP from real-world perspectives. In our solution, dynamic task selection, distributed task allocation and contingency handling mechanisms along with the low level robot controllers and the motor and sensory modules are integrated into each other to solve the real world MTRP. Target allocation and route construction is integrated into each other by an incremental assignment approach. Real time situations and contingencies that change the problem instance are handled at the same time. Empirical evaluations of the system performed on the WEBOTS robot simulator reveal the efficiency of the integrated components of our approach.

http://www2.itu.edu.tr/~sariel/publications.php