Improved artificial potential field route planning method based on stochastic perturbation

Abstract

As a typical route planning algorithm, the APF method has been widely used in robotics, autopilot and other fields. However, it also has weakness in local optimal point. In computer science and mathematics, stochastic perturbation is an accepted approach that can assist optimization algorithms leave the local optimal solution and achieve the global optimal solution as well as improve robustness and adaptability in specific application situations. therefore, the problem can be successfully solved by using stochastic perturbation. In this article, the traditional APF improved through stochastic perturbation. When the robot is running in an APF, the distance between the robot and the target point is calculated at all times. If the distance between the robot and the target point remains unchanged for a period of time or if the stop point oscillates, it is determined whether the robot has reached the target point at this time. If not, it is determined that the robot has fallen into a local optimal point at this time. After completing the judgment, stochastic perturbation is applied to the robot, causing it to jump out of the local optimal point and continue to navigate to the target point. The results of the comparison between the paths generated by the traditional method and the improved method validate the effectiveness of this method.