This study combines RGB-D SLAM with laser-based calibration on a robotic arm platform to achieve millimeter-level accuracy in 3D wall reconstruction. It addresses key challenges in achieving reliable wall geometry modeling under dynamic conditions, texture-poor surfaces, and spatial constraints. By integrating SLAM, laser-guided precise calibration, and refinement through point cloud registration, the proposed method overcomes the limitations of single-sensor systems and meets the high-precision demands of modern construction and architectural workflows.

1.Literature review on existing methods and their Limitations
2.RGB(D)-based SLAM from two cameras on a moving robotic arm
3.Calibration of the 3D reconstruction using laser-based point measurements
4.Point cloud registration and refinement of the 3D model using plane regression
5.System validation and performance evaluation

This thesis presents a complete research workflow, from problem definition to experimental validation. It begins by introducing the motivation and objectives of high-precision 3D wall reconstruction, followed by a literature review of existing methods. The system design section describes the integration of RGB-D cameras and laser sensors on a robotic arm. Subsequent parts detail the implementation of SLAM, point cloud processing, and model refinement. The study concludes with experimental validation and a discussion of future research directions.
Additionally, this work was carried out under the guidance of Bram Vanderborght as Thesis Promotor 2 and Zemerart Asani as Thesis Supervisor 2.

Good knowledge of Python (including PyTorch), machine learning, and signal processing. Knowledge of image processing, SLAM, and computer vision methods is a plus.