Research on adaptive artificial potential field obstacle avoidance technology for unmanned aerial vehicles in complex environments

Path planning for Unmanned Aerial Vehicles (UAVs) in complex environments with coexisting static and dynamic obstacles remains a critical challenge in Artificial Potential Field (APF) research. Current APF methods are constrained by limitations such as target unreachability, local minima entrapment and inefficient dynamic obstacle avoidance, hindering their practical deployment. To address these challenges, an enhanced APF algorithm is proposed, incorporating four key innovations: an adaptive repulsive potential field function to address target unreachability, a randomised directional perturbation strategy for escaping local minima, a collision risk prediction-based force field for dynamic obstacle avoidance, and fuzzy rules with adaptive safety distances to optimise avoidance velocity. Simulation experiments in hybrid static-dynamic obstacle environments demonstrate that the proposed algorithm achieves a 96.3% success rate in trajectory planning with a 3.4 s runtime, 14.79 m path length and maximum angular velocity of 27.89°/s, outperforming conventional APF, DDPG-APF, IAPF, GWO-APF. The collaborative multi-strategy optimisation effectively enhances UAV adaptability in dynamic environments, reduces collision risks and improves trajectory smoothness, providing an efficient solution for real-time path planning in obstacle-dense scenarios.