Probabilistic Quadtrees for Variable-resolution Mapping of Large Environments

Probabilistic occupancy grids are a common and well-proven spatial representation used in robot mapping. However, representing very large environments with high resolution can still impose prohibitive memory requirements. A quadtree is a well-known data structure able to achieve compact representations of large two-dimensional binary arrays. We extend this idea and present probabilistic quadtrees for efficient representation and storage of probabilistic occupancy maps. We also discuss the implementation of probabilistic quadtrees and their integration into our robotic middleware system Miro.