Research on the System of Patrol Unmanned Aerial Vehicle (UAV) Docking on Charging Pile based on Autonomous Identification and Tracking

This paper studies the system of patrol unmanned aerial vehicle (UAV) autonomous identifying, tracking the charging pile and distance calculation in the process of achieving innovative autonomous charging target. This paper firstly proposes an SRDCF-based (Spatially Regularized the Correlation Filters) identification and tracking algorithm. The algorithm extracts sift features within the scope of real-time image to match the existing template. Then, a minimum circumscribed rectangle around the target is created as the initial tracking box to determine target area. Afterwards, the target is tracked through training and detecting the location of area. Lastly, camera ranging module with monocular camera measures the distance between the camera and the target. Thus UAV body position relative to charging pile label target can be obtained. The real UAV experimental results shows that the target detection and tracking algorithm can accurately recognize and track the charging pile docking label under the condition of camera movement in the UAV normal flight. Compared with the STC (Spatio—Temporal Context) tracking algorithm, the improved SRDCF algorithm has improved accuracy and robustness obviously. In addition, camera ranging module can accurately measure the distance between camera to the target and the requirement of real-time performance and reliability is reached.