Research on autonomous UAV-based inspection of cable tunnels using visual SLAM

This paper investigates an autonomous inspection method for cable tunnels using unmanned aerial vehicles (UAVs) based on visual SLAM technology. The proposed approach addresses the challenge of localization in tunnels with weak GPS signals. By improving the ORB feature extraction algorithm, adaptive feature matching, and loop closure detection, the UAV's pose estimation is optimized. High-precision 3D semantic maps are constructed using RGB-D data from depth cameras. A path planning scheme combining APF and RRT algorithms is proposed, enabling global path optimization and local dynamic adjustments. Experiments conducted in the ROS and Gazebo simulation environments validate the reliability and effectiveness of the proposed method in complex tunnel environments, providing an efficient and intelligent solution for UAV-based inspection in enclosed spaces.