Path Planning with Improved Artificial Potential Field Method Based on Decision Tree

Abstract

Path planning is one of the key research directions in the field of mobile robots. It ensures that moving objects can reach the target point safely and without collision in a complex obstacle environment. The path planning is to search an optimal path from the starting point to the target point for the mobile robot in an environment with obstacles, according to certain evaluation criteria (such as the time, the best path, the minimum energy consumption, etc.). The path planning based on artificial potential field method has been paid more and more attention because of its advantages such as convenient calculation, simple implementation of hardware and outstanding real-time performance. However, the artificial potential field method has some limitations, such as the local minimum, the oscillation of moving objects among obstacles and so on. To solve these problems, we can introduce the idea of decision tree into the artificial potential field method for improvement. In machine learning, decision tree is usually used for classification. It is a prediction model, which represents a mapping relationship between object attributes and object values. By utilizing the advantages of decision tree in rule expression and extraction, an improved artificial potential field path planning model based on decision tree is constructed, which can realize real-time and accurate identification of current behavior and fast decision-making of next time behavior in path planning. Aiming at the dynamic path planning problem of mobile robots in indoor complex environment, based on the traditional artificial potential field method, this paper introduces the distance term into the potential field function, and proposes an improved artificial potential field method based on the idea of decision tree, to solve the local minimum, the oscillation between obstacles and concave obstacle problems. According to repulsion coefficient, deflection angle of resultant force and velocity, a reasonable classification decision is made to meet the needs of different obstacle distribution scenarios, and the effectiveness of the proposed method is verified by simulation experiments. Simulation results show that, compared with the traditional artificial potential field method, the planning time of improved algorithm is reduced by 50%, and the smoothness of path planning by the improved algorithm is increased by 43.3%.