Research On Improved Artificial Potential Field Path Planning Algorithm

Path planning is an important problem in the navigation of mobile robot. The mobile robot autonomously searches for an optimal or sub-optimal collision path from the start state to the target state based on a particular performance measure. Many path planning algorithms can achieve good results by getting prior environmental information, but in the unknown environment these algorithms maybe not working. With the increase of mobile robot complexity and the expansion of application scope, the requirements of path planning are getting higher and higher, especially the application of local path planning is restricted by traditional planning method.The traditional artificial potential field (APF) introduces the generalized potential field concept, generates the attractive potential field by the target point and the repulsive potential field by obstacles. The mobile robot uses the potential gap to moving in the attractive potential field and its trajectory is the optimal collision-free path in the unknown environment. The APF method is adaptable for a real-time path planning and obstacle avoidance. It has high real-time performance and smooth planning trajectory, but also has two inherent defects: the goal nonreachable with obstacle nearby and local minimum problem. The traditional artificial potential field based path planning methods have a loval minimum problem, which can trap mobile robots before reaching it's goal. In this study, the relative distance between the mobile robot and the target point will attach to an improved repulsive potential field function model so that the target point is always the gobal minimum point in the potential field. The problem of local minimum caused by the force balance between the obstacles and the target point, this paper proposes a detection method based on the movement range of the mobile robot, and uses the temporary sub-target strategy to escape local minimums occurred in local path planning.This paper introduces the rapidly exploring random tree (RRT), based on the sam-pling algorithm which is probability completeness, has a good sampling and convergence speed. For combining the artificial potential field method with the RRT algorithm, the ob-jective force of the attractive potential field method is added into the search tree extension stage of RRT, and the target-oriented factor is added to the tree node growth function in the RRT algorithm, which solves the local minimum problem of the mobile robot in the path planning stage and also solves the randomness problem of the RRT algorithm.The experimental results show that this new local path planning algorithm, the APF-RRT algorithm, improves the defects of the traditional artificial potential field algorithm and also has a good effect in avoiding local minimum problem and reducing the compu-tational complexity.