Binocular vision-based 3D method for detecting high dynamic and wide-range contouring errors of CNC machine tools

Abstract

Periodic checking for contouring error in unloading conditions can effectively evaluate the dynamic performance of machining centers. Existing measurement devices have limitations in the high accuracy three-dimensional (3D) measurement of arbitrary contouring errors (e.g. ballbar and cross-grid encoder). Thus, in this paper, a cost-effective binocular vision-based 3D method for detecting high dynamic and wide-range contouring error of computer numerical control machine tools is proposed. With this method, the strobe lighting method is first presented to suppress the blurring effect of the newly designed 1024 coded targets that were embedded in the measurement fixture. Thereafter, an encoding and decoding method based on finding the optimal initial non-encoding region is proposed to automatically identify, match, and reconstruct the time-varying targets. Then, to enhance the performance of the vision system, a calculated method based on three visible non-collinear coded targets is presented to deduce the dynamic and large-scale contouring error. Finally, contouring error detection experiments are carried out in a self-built five-axis machine tool to validate the advantages of the method. Both the proposed vision method and cross-grid encoder are used to measure contouring errors of three types of trajectories under different feed rates. By comparing the difference between the two trajectories measured by the two devices, the experimental results illustrate that the mean vision detection error at 7 m min−1 is about 4.3 µm, which verifies the accuracy and effectiveness of the proposed vision measurement method.