An Improved Artificial Potential Field UAV Path Planning Algorithm Guided by RRT Under Environment-Aware Modeling: Theory and Simulation

Abstract

Unmanned Aerial Vehicles (UAVs) have been extensively researched and used in civil and military applications due to their effectiveness and flexibility. However, when identifying obstacles and avoiding them, most of the existing path planning methods fail to accurately perceive the environment, such as without considering the differences between obstacles, which leads to low timeliness and easy fall into a local minimum. In this work, an improved artificial potential field UAV path planning algorithm (G-APF) guided by the rapidly-exploring random tree (RRT) based on an environment-aware model is designed to overcome the limitations of traditional methods. The model can perceive different objects in the environment through the addition of supervised environment modeling to traditional unsupervised path planning. Specifically, an environment-aware model based on YOLOv8 is used to establish the UAV flight environment model, and an adaptive optimal threat distance calculation module is used to construct the repulsive potential field. Secondly, to improve the timeliness of path planning and the global awareness of the model, we first use the G-APF algorithm to plan the rough flight path based on the UAV flight environment. Then, the initially generated trajectory is replanned by building an attractive potential field and combining it with a repulsive potential field. Finally, the problems of local minimum and target unreachability and local trajectory oscillation generated by the artificial potential field (APF) algorithm are solved by G-APF. Experiments with generated regions are performed to demonstrate the efficiency and effectiveness of the proposed approach.

Keywords

• UAV
• path planning
• object detection
• perception
• supervised learning
• modeling