It is a complete sorting and depalletizing system solution based on binocular structured light camera and equipped with 3D vision software for target recognition and positioning developed from deep learning.
The binocular structured light camera acquires the RGB image and 3D point cloud data and processes the data with 3D vision software to output the segmentation result and attitude positioning of the target identified, as well as the coordinate information of the grabbing point. The information will then be converted into the grabbing coordinates for the robot to perform automatic grabbing after moving to the target object.
Three-dimensional imaging principle of structured light: Use structured light (stripe or scatter) to encode space. The camera takes images and decodes them to generate depth images. With this method, the three-dimensional contour information can be obtained in a short time without requiring the movement of objects or cameras.