Robot Tool Calibration in Precise Glass Handling

In the cell phone protective glass manufacturing industry, glass need to be first ground to a desired thickness, which requires human workers to place the glass pieces precisely into the grinder. We propose to use a 6 DOF industrial robot equipped with vision sensors to automate the process by the “pick and place” task. The precision of the placing depends not only on the vision detection, but also on the calibration of the camera and the glass plane. In this paper, a Maximum a Posteriori (MAP) method is proposed to increase the calibration accuracy. A nominal calibration is first obtained with standard method, then it is corrected with observations. Experimental results shows the increased accuracy of placing.