High-reliability Automatic Collision Avoidance Strategy for Unmanned Aerial Vehicles Based on Sensing Reconstruction

Addressing the safety flight requirements of complex airspace by unmanned aerial vehicles in the context of low-altitude economy development, this paper systematically considers the failure issue of atmospheric sensors under strong wind interference and proposes a high-reliability automatic collision avoidance strategy based on sensing reconstruction. Firstly, an aircraft dynamics model incorporating turbulence disturbances is established, and an adaptive cubature Kalman filter is employed to fuse navigation measurements and control signals, achieving robust online reconstruction of states such as true airspeed and airflow angles. Secondly, to address model mismatch and noise disturbances during the escape phase, an intelligent learning-based adaptive control law is designed to compensate for state estimation errors, enabling stable tracking of escape maneuver commands. Finally, a dynamic collision envelope driven by the filter covariance is constructed, and trajectory prediction uncertainty is quantified by integrating the control system model to complete terrain collision detection. This facilitates the generation of optimal obstacle avoidance commands by evaluating multiple escape trajectories. Simulation results show that accurate airflow angle reconstruction and robust collision warning and recovery control are achieved under gust and severe turbulence conditions. The related techniques can provide a reliable solution for the design of collision avoidance systems in low-altitude unmanned aerial vehicles.