Available Benchmark Sets

3D Pathfinding Benchmarks

This page is part of Nathan Sturtevant's Moving AI Lab. This page is focused on benchmark maps and problems for 3D voxel grids.


There has been relatively little work on pathfinding in three dimensions, as compared to on 2D grids. Digital Extremes, a games company in London, Ontario, Canada has made voxel data freely available from their game Warframe to start a new repository of 3D voxel grids. (Pixels are picture elements; voxels are volume elements)

We encourage other companies that use voxel world representations to make their map data available for experimentation purposes.

Each benchmark set has a number of metrics associated with it. A short paper was published at SoCS 2018. These benchmarks can be cited as:

@article{brewer2018voxels,
  title={Benchmarks for Pathfinding in 3D Voxel Space},
  author={Daniel Brewer and  Nathan R. Sturtevant},
  journal={Symposium on Combinatorial Search (SoCS)},
  year={2018}
}

Information on file formats is found in the paper describing this work, which will be linked shortly, but files are intended to be as simple as possible:

Note: An error was found around in January, 2019 in how the benchmarks were originally generated. (It appears the generation used an older executable.) The benchmark problems were subsequently regenerated with the proper movement model.

Benchmark sets

Commercial Game Benchmarks
BoxSourceBrowseMapBenchmarks
Warframe (original release, March 2013; still being updated)
Data provided by Daniel Brewer in collaboration with Nathan Sturtevant
Browse data Download
all maps
Download
all benchmark problems

All data is made available under the Open Data Commons Attribution License