Robot Path Planning Using an Improved Genetic Algorithm with Ordered Feasible Subpaths

Abstract

This paper proposes an improved genetic algorithm to enhance the search efficiency and robustness of robot path planning. Its three main contributions are: 1)initialization: free grids are divided into multiple sets according to the main diagonal direction, and all feasible subpaths in each set are found. From a given starting point, feasible subpaths are randomly selected in each set and orderly connected to reach the end point. This guarantees that each initial path is feasible. 2)mutation: a new mutation operator is proposed to ensure the generated paths are feasible. 3)simplification: a simplification operator is proposed to shorten the path while maintaining its feasibility. Experimental results demonstrate the speed and performance of the proposed algorithm.