Image Processing based UR5E Manipulator Robot Control in Pick and Place Application for Random Position and Orientation of Object

Abstract

Many industries are required to increase output and productivity. Manipulator Robot is chosen as a solution, especially for pick and place applications. An intelligent system is needed to reduce uncertainty and decision-making for some varying object or workpieces. In this paper, the UR5e manipulator robot can automatically detect the object using computer vision through the camera as a sensor and OpenCV computer program. Geometrical transformations and statistical approaches transform picture data into position and orientation data for robot control. The system detects the center point coordinates and the orientation angle of the detected object. The coordinates obtained by the camera perception are transformed into robot perception. At the same time, the value of the orientation degree is converted into a quaternion form which becomes the rotational movement parameter for the end effector of the robot. Position and orientation data are sent to the robot controller from the ROS framework computer program via ethernet. The robotic arm can move towards the object, and the end effector can rotate accordingly. The experimental results prove successful recognition of object position and orientation. The system has an accuracy in detecting object positions at x-axis direction = 99.0%, at y-axis direction = 98.05%, and the system orientation accuracy is 99.11%.