A novel multi-sensor fusion method for geometric tolerance measurement of large automotive structural components

Abstract

Abstract This study proposes an innovative measurement method using multi-sensor data fusion and binocular camera point cloud fusion technology to meet the precision demands for geometric tolerance measurements in large automotive components. Spatial location information is collected using robotic arms, inertial measurement units (IMUs), IMUs, and global cameras, with small spheres used as reference objects to obtain depth information. A dedicated workpiece coordinate system is established to integrate multi-source coordinate data into a unified framework. The spatial pose data from 3D cameras are fused using the extended Kalman filter (EKF), EKF algorithm, and features are accurately measured through the reconstruction of the workpiece. The experimental data indicates that the multi-sensor fusion method outperforms traditional coordinate measuring machines(CMMs), CMM, with a maximum error of 0.667 mm and a standard deviation of less than 0.1 mm. Notably, the method demonstrates superior accuracy in handling complex surfaces and large-scale components, whereas CMM requires frequent adjustments and encounters larger errors in such scenarios. This highlights the advantage of using 3D cameras and multi-view point cloud fusion for automated, high-precision measurements.