Autonomous Obstacle Avoidance and Navigation Method for Unmanned Aerial Vehicles Based on Multi-Sensor Fusion Algorithm
Min Jin, Jun Tao, Zhen Qiu, Jingpo Bai, Chong Liang, Xiaoning Li · 2024
With the rapid development of smart grid technology, power inspection work is facing unprecedented challenges and opportunities. Traditional inspection methods are not only inefficient, but also difficult to cover complex power environments. As an efficient and flexible inspection tool, drones have gradually become an important force in the field of power inspection. However, in the process of power inspection, due to the complex and ever-changing environment and numerous obstacles, how to achieve autonomous obstacle avoidance for drones has become an urgent problem to be solved. This paper proposes an autonomous obstacle avoidance method for unmanned aerial vehicles based on multi-sensor fusion. By using an improved Bayesian fusion algorithm, the point cloud information obtained from two-dimensional LiDAR and depth camera is fused to compensate for the shortcomings of a single sensor in detecting complex structures of power lines. The experimental results show that this method significantly improves the perception accuracy of drones in the surrounding environment during power inspection, with a root mean square error of less than 0.05m for fused point clouds, and has good obstacle avoidance effect. During the testing process, the distance between the drone and obstacles was maintained at 0.5m or more, ensuring the safe flight of the drone during power inspection and providing effective technical support for the inspection work of the smart grid.