Adaptive GA-based Potential Field Algorithm for Collision-free Path Planning of Mobile Robots in Dynamic Environments

While detecting the obstacles is the first step in the proper operation of an autonomous robot, the most vital part is the path planning. Many path planning methods like potential field rely only on the position of the obstacles and the target to determine a valid path. Since the obstacles may move in the environment, the generated path based on these algorithms won't be optimum. Recent algorithms tend to solve this problem by adding the dynamic information of the objects to the path planning method. While this approach seems to solve the problem completely, the lack of adaptiveness in many cases will cause problems if the mobile robot moves in various environment setups. This paper proposes an adaptive GA-based potential field algorithm for collision-free path planning. The modification is done by adding a term to the negative field based on the obstacle dynamics. To enhance the quality of the potential field in different situations, the required coefficients are calculated online with an adaptive genetic algorithm. The adaptiveness of the algorithm is achieved by changing the impact of population generation methods of genetic algorithm in each iteration. To validate the method, a series of simulations and experiments are conducted and the results confirm the effectiveness of the algorithm.