Grasping moving objects with incomplete information in a low-cost robot production line using contour matching based on the Hu moments

Abstract

In a low-cost system with limited hardware capacity, due to the fast speed of the conveyor and the delay in shooting and processing time, a part of the workpieces may be out of view of the camera. As a result, the information received is incomplete. To solve this challenge, this paper proposed a method to accurately estimate the position of workpieces with incomplete information. First, a complete image of the workpiece in the horizontal position is captured and stored. In the case that a part of the workpiece lies outside the image, the contour matching based on the Hu moments between the contour of the incomplete part and different parts of the complete workpiece is conducted to determine which part of the workpiece has been taken by the camera (called the matching part). Then, the distance between the centroid of the matching part and the entire workpiece is calculated. The difference between the orientation angle of the current incomplete part and the matching part is the orientation angle of the entire current workpiece (with a part out of the image boundary). From the centroid distance and orientation angle, the actual centroid of the entire current workpiece can be calculated. Finally, the calibration method is used to transform the pose of the workpiece into the robot base frame. The experimental results demonstrate that the accuracy of the centroid estimation is 2.2 pixels and 1.8 mm in the image and the 3D space, respectively.

Keywords

• Industrial robot
• Computer vision
• contour matching
• Calibration
• Hu moments