Performance

Integrator

UFOMap introduced a new fast integrator. This integrator is freeing space at a higher level in the octree, making it significantly faster while still preserving the occupied space.

Method	Voxel size (cm)	Time (ms)
UFOMap OctoMap	16	4.98 ± 1.30 5.52 ± 1.66
UFOMap UFOMap fast OctoMap	8	12.3 ± 7.4 6.5 ± 2.2 16.3 ± 10.4
UFOMap UFOMap fast OctoMap	4	60.9 ± 44.7 10.9 ± 4.7 104.6 ± 82.2
UFOMap UFOMap fast OctoMap	2	371 ± 254 28 ± 15 745 ± 548

UFOMap fast is always freeing space at 16 cm voxel size in this case.

Performed on the cow dataset.

Memory Consumption

UFOMap differentiates between inner nodes and leaf nodes and is packing the nodes more densly. This means UFOMap uses less memory compared to OctoMap.

Dataset	Voxel size (cm)	UFOMap (MiB)	OctoMap (MiB)
FR-078 tidyup	5	7.42	21.49
FR-079 corridor	5	19.70	51.86
Freiburg campus	20	58.71	155.46
freiburg1_360	2	15.98	42.05
New College	20	29.41	75.40

Performed on the OctoMap 3D scan dataset.

Number of Nodes

UFOMap's explicit representation of unknown space means it has more nodes in total compared to OctoMap. However, as seen in Memory Consumption above, it still uses less memory.

Dataset	Voxel size (cm)	UFOMap	OctoMap
FR-078 tidyup	5	1 642 113	1 369 165
FR-079 corridor	5	2 823 713	1 829 134
Freiburg campus	20	8 402 193	5 515 178
freiburg1_360	2	2 161 849	1 547 112
New College	20	4 157 217	2 633 701

Performed on the OctoMap 3D scan dataset.

Exploration

Using UFOMap for RH-NBVP exploration significantly increases the exploration performance.

More in the Paper

You can read more about UFOMap in the paper.

If you use UFOMap in a scientific publication, please cite the following paper:

@article{duberg2020ufomap,
  author={Daniel Duberg and Patric Jensfelt},
  journal={IEEE Robotics and Automation Letters}, 
  title={{UFOMap}: An Efficient Probabilistic {3D} Mapping Framework That Embraces the Unknown}, 
  year={2020},
  volume={5},
  number={4},
  pages={6411-6418},
  doi={10.1109/LRA.2020.3013861}
}