An unmanned aerial vehicle navigation obstacle avoidance method and system based on adaptive airflow perception
By using adaptive airflow perception and hybrid model prediction, combined with scene feature library and improved algorithms, the problem of insufficient airflow perception of quadcopter UAVs during high-speed outdoor flight is solved, achieving high-precision obstacle avoidance and stable navigation, and improving the adaptability of UAVs in multiple scenarios.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGZHOU HOLLEY COLLEGE
- Filing Date
- 2026-02-11
- Publication Date
- 2026-06-02
AI Technical Summary
Quadrone drones have low airflow perception accuracy and passive obstacle avoidance mode during high-speed outdoor flight, which lacks foresight and leads to delayed obstacle avoidance decisions that can easily cause collision risks. In addition, they have weak adaptability to different environments and cannot meet the obstacle avoidance needs of multiple flight scenarios.
An adaptive airflow perception-based UAV navigation and obstacle avoidance method is adopted. The method acquires multi-dimensional airflow parameters, obstacle information and UAV attitude data through the perception layer, uses the LSTM-Transformer hybrid model to predict airflow change trends, combines the improved A* algorithm to plan the path, and determines the airflow response strategy based on the scene feature library to achieve attitude calibration and obstacle avoidance.
It improves obstacle avoidance accuracy and stability, enabling reliable navigation and obstacle avoidance in different environments, enhancing the drone's adaptability to different scenarios, and ensuring the stability and safety of its flight trajectory.
Smart Images
Figure CN122130076A_ABST