Loop Closures in Large Datasets

For large datasets, it is important to regularly revisit zones that have been mapped already (loop closure). This reduces drift error and creates a more accurate map.

Loop closures are calculated during post-processing. That is why, especially for large datasets, the resulting map after post-processing will look better than the quality map created live during mapping.

When the mobile mapping system crosses its previous path, the software will "recognize" the previously mapped environment. It will adjust the map, reducing the drift error that occurred since the mobile mapping system first passed the area.

Take any opportunity you get to close loops. In areas with long corridors, such as factory halls, the opportunities are rarer.

When closing a loop, make a point of crossing the previous path.

Scaffolding

Loop closures give the dataset more stability. Like the knots in a net or the struts in a scaffold, they define the overall structure and keep it from "falling apart".