A practical grasping system using HSV modelled Kinect2 vision and YouBot

In order to reduce the instability of the robot experiment results based on monocular vision, this paper proposes a method of robot grasping using Kinect2 in vision part. In this method, Kinect2 is fixed on the robot. The robot and Kinect2 are connected to the same terminal to form the control system. In the coordinate transformation between Kinect2 coordinates and robots, we use checkerboard calibration method to determine the Kinect2's internal and external parameters. We provide Kinect2 with the RGB image and the depth data, and uses the Point Cloud Library to match them, thus the grasping system is formed. The RGB image is converted to HSV image, and the target object is identified through color recognition. The coordinates of the target in the robot coordinate system are calculated through the internal and external parameters calculated before. After getting the coordinates of the object, the optimal way of grasping is determined through the solution set of the motion angle of the manipulator and path planning.