LAUSR

Intelligent robots play an increasingly important role in the field of autonomous charging, such as automobiles, civil aircraft, military aircraft and so on. The traditional manual charging method not only has low efficiency, but also has high labor cost, especially in the occasions where day and night compatibility are required. This paper proposes an intelligent autonomous charging method based on visual guidance. It analyzes historical cumulative natural feature data, estimates accurate pose(position and attitude) information, solves the problem of low precision of traditional measurement methods, and improves the robustness of the algorithm. Firstly, based on the template matching algorithm, the effective area of the charging port is obtained. Secondly, according to the least squares method(LSM), the elliptical features of the effective area image are fitted. Combined with the geometric parameters of any two ellipses, the charging port is coarsely positioned to determine its relative pose. Thirdly, the cumulative natural feature data(CNFD) is used to accurately positioning of the charging port by Extended Kalman Filter(EKF) algorithm. Further, the vision-based docking method is presented. Finally, we developed an experiment system, which is composed of a charging port docking model, a UR5 robot, one monocular camera, and a high precision laser tracker (used to evaluate the vision measure accuracy). The experiment results verified the effectiveness of the proposed method.