Improved Genetic Algorithm in a Static Environment for the Robotic Path Planning Problem

Robotic path planning is an important problem in mobile robot applications. It targets finding an optimal path for a mobile robot from a start node to a destination node in an environment with obstacles. We consider a Genetic Algorithm (GA) which is used widely to compute an optimal path and obtain promising search results. A classical GA, however, performs more of a “blind” search and evolves without any previous knowledge, which leads to the GA being inefficient at times. In this paper, to improve the performance of the GA in path planning, a random domain inversion method, inspired by the Double-Domain Inversion, is proposed and implemented as a modified GA. This modification effectively improves the searchability of the local space and increases the probability of generating excellent individuals. The experiments conducted show that the optimal path can mostly be found, and the efficiency of the genetic algorithm is significantly improved. In comparison with the conventional GA, the suggested approach achieves better results although it consumes significantly less computation time.