Now that I’ve refined my project focus, it’s time to start working with LiDAR data. I’ve downloaded VeloView, the official software for Velodyne Puck Lite, and done a few lab test scans. These were recorded in different formats, including .pcap and .json, to examine the raw data and determine what I can extract from it.

I’m also considering working on Linux as it might offer better integration with the Velodyne driver, but I’ll see how things go.

Processing Lidar Data – The Plan

The goal is to process these raw scans into useful structural information. The planned steps are:

• Convert raw scans into point cloud data.

• Process the point cloud: Extract valuable data and remove noise (e.g., using voxel grid filtering or similar methods).

• Wall detection: Techniques such as RANSAC are used to identify flat surfaces.

• Convert walls into 2D geometry.

• Parse DXF floor plans using libraries like ezdxf.

• Align coordinate systems (manually at first, then ideally automate this step).

• Compare geometries using tools like Shapely to detect differences between the scan and the floor plan.

This is just the initial plan, and I expect the details to evolve as I test different methods. The next step is to experiment with point cloud processing and implement essential wall detection. I’ll update you on my progress in the next blog!