Our AI was trained to understand point clouds. The neural network searches for building parts like windows, doors, and many more, and transforms million of points to only a few individual objects. The result is a “Smart Point Cloud” which can be loaded, e.g., in Revit.
We worked for years on our training algorithms to teach an AI to understand laser scanners. Without an AI, point clouds can be a complicated field. There are, e.g., multiple definitions of what exactly structured or unstructured point clouds are, but structured or unstructured simply doesn’t matter for our neural network. It just “sees” the points. Customers ask where they have to configure the type of point clouds and the scanner type, but there is simply no configuration. We have one AI and it fully automatically processes point clouds from different scanner types.
So far, we received point clouds from terrestrial scanners from Faro and Leica, handheld scanners like the BLK2Go or scanners from GeoSLAM, and mobile scanners from NavVis. The results can be seen in the video above. If you would like to try other scanners, feel free to contact us for a field test.