Genetic Algorithm Based Multipath Optimization for Multimobile Robot Navigations

Abstract

Multimobile Robot Flow Network Problem (MM-RNP) is to find optimum navigation paths in a network without robot collisions. Very few works have been done for the multirobots path allocation problem. Most of the work concentrated on the single robot path allocation problem. This MMRNP is one of the hardest combinatorial optimization problems. MMRNP solutions can be derived using exhaustive enumeration and branch-and-bound linear programming methods. However, the computation required by these procedures is enormous even for a small size problem. In this paper, we present a heuristic approach using Genetic Algorithm (GA) to achieve the near-optimal solution. We propose a new population initialization for our GA with different operators. The proposed GA optimizes path allocation for mobile robots navigating in a static network environment. We follow the edge-level perspective to solve the MMRNP. Our experimental results show that the proposed method gives a better navigation strategy than the traditional methods.