Development of a SCARA robot arm for palletizing applications based on computer vision

This paper develops a computer vision system integrated with a SCARA robot arm to pick and place objects. A novel method to calculate the 3D coordinates of the objects from a camera is proposed. This method helps simplify the camera calibration process. It requires no knowledge of camera modeling and mathematical knowledge of coordinate transformations. The least square method will predate the Equation describing the relationship between pixel coordinates and 3D coordinates. An image processing algorithm is presented to detect objects by color or pixel intensity (thresholding method). The pixel coordinates of the objects are then converted to 3D coordinates. The inverse kinematic Equation is applied to find the joint angles of the SCARA robot. A palletizing application is implemented to test the accuracy of the proposed method. The kinematic Equation of the robot arm is presented to convert the 3D position of the objects to the robot joint angles. So, the robot moves exactly to the required positions by providing suitable rotational movements for each robot joint. The experiment results show that the robot can pick and place 27 boxes on the conveyor to the pallet with an average time of 2.8s per box. The positions of the boxes were determined with an average error of 0.5112mm and 0.6838mm in the X and Y directions, respectively.