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

Description

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

Steps to reproduce

This repository includes full MATLAB source codes and raw experimental data for robot in-situ geometric reconstruction measurement. Instruments: Industrial robot integrated with a tri-laser measurement device (TLMD). Methods & workflows: TLMD intrinsic calibration using sphere-center collinearity constraint; Hybrid robot position-orientation error calibration based on LPOE with three calibration models; Clustering-based calibration pose selection via surface normal distribution and geometric entropy; Mirror surface measurement, geometry reconstruction and accuracy evaluation. Software: All algorithms are implemented in MATLAB R2022 or newer. All scripts and raw calibration/measurement data are categorized into separate folders, enabling full reproduction of the experiments presented in the paper.

Institutions

• Soochow University

  Jiangsu, Suzhou

Categories

Licence