Stereo vision-based real-time localization of electric vehicle battery in the complex environment

The approach to passive stereo vision-based real-time localization is developed in this context, as it can adapt to complex factors such as the illuminations, backgrounds, viewing angles, and occlusions. After the calibration of the visual coordinates, the proposed system can jointly calibrate the robots and the vision systems flexibly by controlling the terminal positions of the robots. The fast phase-only correlation algorithm is proposed to considerably improve the matching efficiency of the image pairs. Pose normalization and regional three-dimensional vector feature detection can make the proposed system robust to the surface patterns of the cars and the battery shapes, allowing for reliable localization and detection. The proposed real-time system has been verified in the practical application.