Raw Algorithm & Calibration Dataset for Robot In-Situ Geometric Reconstruction

This repository provides the source codes and original data for the paper: "Accuracy Enhancement for In-Situ Geometry Model Reconstruction for Robot-Based Measurement by Targeted Calibration Data" The code implements a task-oriented robot calibration framework for improving the accuracy of robot-based in-situ measurement of complex free-form surfaces. The proposed framework integrates: (1) a tri-laser measurement device (TLMD) calibration method, (2) a hybrid position-orientation error calibration model based on LPOE, (3) a clustering-based calibration pose selection strategy, and (4) mirror surface measurement and reconstruction accuracy evaluation. ## Folder Description ### 1. Calibration of the TLMD This folder contains the calibration algorithms for the tri-laser measurement device. The sphere-center collinearity constraint is used to identify the intrinsic geometric parameters of the three laser sensors. ### 2. Hybrid error calibration This folder contains the proposed hybrid robot calibration framework. The implementation includes: - Three calibration models (M1–M3) for different robot calibration strategies; - Sphere center position and normal vector pointing error evaluation for calibration validation. ### 3. Cluster-based point selection This folder contains the task-oriented calibration data selection method. Surface normal distribution and geometric entropy are used to cluster surface orientations and select representative calibration points. ### 4. Mirror surface measurement This folder contains the experimental data processing and reconstruction accuracy evaluation for the mirror surface measurement task. ## Requirements MATLAB R2022 or later