Improving Efficiency and Effectiveness of Wheeled Mobile Robot Pathfinding in Grid Space Using a Genetic Algorithm with Dynamic Crossover and Mutation Rates

Incorrect parameter tuning of crossover and mutation rates in Genetic Algorithms (GA) can negatively impact their effectiveness and efficiency in mobile robot pathfinding. This study focuses on improving the performance of wheeled mobile robots in grid-based environments by introducing a Dynamic Crossover and Mutation Rates (DCMR) strategy within the GA framework. The primary contribution of this research is enhancing the efficiency and effectiveness of mobile robot pathfinding, resulting in shorter average path lengths and faster convergence times. Additionally, this method addresses the challenge of selecting appropriate GA parameters while increasing the algorithm's adaptability to different phases of the search process. The DCMR approach involves linearly increasing the crossover rate by 10% (from 0% to 100%) and decreasing the mutation rate by 10% (from 100% to 0%) over every 10 generations during the GA's evolution. Unlike fixed parameter tuning or exponential and sigmoid parameter tuning—both of which require trial and error to determine optimal values—the DCMR method provides a systematic and efficient solution without additional computational cost. Experiments were conducted across eight scenarios featuring varying distances between the start and target points, with two obstacles randomly placed in the robot's environment. The results showed that implementing the DCMR method consistently identified the optimal path, reduced average path lengths by 0.99%, and accelerated algorithm convergence by 48.39% compared to fixed parameter tuning. These findings demonstrate that the DCMR method significantly enhances the performance of GAs for mobile robot pathfinding, offering a reliable and efficient approach for navigating complex environments.