Polarizing image fusion-based pose-measuring approach considering the measuring baseline for hand-eye calibration of a SCARA robot

•The edge sharpness is evaluated by the variance information entropy.•The polarizing image fusion is completed based on the edge sharpness.•Normal vector constraints are combined to solve the target pose.•Pose measurement accuracy is improved by polarizing image fusion and normal vector constraints. Polarized light technology can be applied to pose measurement in the hand-eye calibration procedure of a SCARA robot. However, an inappropriate polarizing angle will decrease the image sharpness, which causes errors in the extracted pixel image coordinates of calibration points. Moreover, the noise easily impacts the pose-measuring accuracy and stability. Aiming at the abovementioned issues, we proposed a new pose-measuring approach for the SCARA robot's hand-eye calibration. First, the polarizing apparatus is used to obtain the polarizing images of the target from multiple polarizing angles, and the image fusion is completed according to the edge sharpness of the calibration points. Through the fused image, the image sharpness is ensured, and the imaging quality of the calibration points is improved. Second, the normal vector constraints created by the longest measuring baselines composed of calibration points are considered extra constraints to construct the pose-measuring model. This enhances the anti-noise ability without increasing the number of calibration points. Using our approach, the pose-measuring re-projection error is reduced by 87 %. Utilizing the hand-eye calibration data acquired by our approach, the robot positioning error is reduced by 47 %. The positioning error of a target is less than 0.04 mm. The positioning error of some real objects is less than 0.05 mm.