Vision guiding system aims to act as the “eyes” of robots. By capturing pictures of the spatial 3D poses of workpieces with camera, the system identifies, extracts and matches features such as holes or angle points to realize the 3D positioning of workpieces under sensor coordinate system according to the relative position information of the features, so as to guide robots to grab workpieces. This system may grab workpieces from different appliances such as workbin, EMS or AGV by installing the visual sensor on the end flange of gripper or robot. Compared to traditional manual feeding mode, this system saves labor cost and improves automation level; compared with ultra-high precise mechanical positioning method, this system lowers the requirement for mechanical positioning precision significantly and saves the maintenance cost of mechanical positioning device. Besides, this system can be compatible with several vehicle models and designed with visual error-proofing function.?