A Probabilistic Risk Field Model for Dynamic Obstacle Avoidance of Unmanned Aerial Vehicles under State Uncertainty

Recent developments in urban air mobility (UAM) have intensified the demand for real-time risk assessment and dynamic obstacle avoidance technologies to ensure safe and flexible unmanned aerial vehicle (UAV) operations in complex low-altitude environments. However, UAM is still in its early stages of development. UAV operations largely rely on preflight static risk assessments and conflict-free trajectory planning and do not account for the effect of environmental disturbances during flight. Therefore, dynamic risk evaluation and obstacle avoidance methods are required to adapt to real-time changes in low-altitude environments and meet the safety demands of UAM operations. This study proposes a novel probabilistic risk field (PRF) for dynamic obstacle avoidance of low-altitude UAVs. Specifically, the model constructs a repulsive risk field based on the UAV’s physical attributes and motion states and incorporates a probabilistic position deviation model to model uncertainty during flight. This framework enables dynamic risk modeling under complex and uncertain low-altitude conditions. Simulation experiments under different conflict scenarios and uncertainty levels demonstrate that the proposed PRF method reduces the maximum collision risk by more than 50% compared to conventional risk field approaches. The PRF model provides a robust framework for uncertainty-aware dynamic obstacle avoidance and holds potential for broader applications in urban air mobility systems, reasonable flight planning, and low-altitude airspace management.