GNSS, 高混凝土坝, 表面裂缝, 无人机, 定位导航, 图像检测

To address the issue of poor positioning accuracy and difficulty in achieving automatic positioning of drone for high concrete dams in mountain and canyon areas with weak or no GNSS signals, a dual-camera visual panoramic perception system, high-precision inertial navigation sensor, and high-performance onboard processor are installed on the DJI M300 drone. The SLAM algorithm for real-time map construction and positioning is adopted to integrate the static coordinates determined by the dual-camera visual panoramic perception system and the dynamic motion parameters determined by the high-precision inertial navigation sensor to obtain the high- precision three- dimensional coordinate data of the drone. The results show that the drone can still follow the set flight path when entering a GNSS-free environment, with an overlap accuracy of the flight path above 95%, achieving positioning accuracy at the centimeter level, crack width recognition accuracy of 0.1 mm, and length error less than 5%. The results indicate that this method has solved the problem of positioning and inspection by drones under GNSS-free conditions. The high-precision positioning combined with the autonomous flight path planning software enables the drone to navigate and inspect along the set path autonomously, which has realized efficient and intelligent full coverage inspection of the entire surface of high concrete dams. Moreover, this technology has been applied in Geheyan hydropower dam project and has obtained good results, therefore can be widely promoted and applied in the daily inspection of high concrete dams.

GNSS, high concrete dam, surface crack, drone, positioning and navigation, image detection