Michael Lindroth

Smart Point Cloud is a research and development project focused on converting raw LiDAR point clouds into structured 3D models for digital twin applications.

The pipeline uses advanced point cloud processing techniques, custom algorithms, and spatial indexing structures to process large-scale 3D datasets. It performs tasks such as ground and object segmentation, surface and edge detection, and transformation of point cloud data into optimized representations suitable for AI and machine learning workflows. The objective is to automate the extraction of infrastructure components from LiDAR scans to support digital reconstruction, analysis, and engineering applications.

Part of the project was the creation of a flexible, node-based processing framework that enables users to create and modify point cloud workflows through a visual interface. As in visual programming environments, processing components such as filtering, segmentation, classification, and feature extraction can be connected to form custom pipelines without changing the underlying code. This improves experimentation, reproducibility, and adaptability across different datasets and use cases.