perfectly-balanced / PathBench

Benchmarking Platform for Classic and Learned Path Planning Algorithms.
Other
91 stars 18 forks source link

Working with external (3D) maps #77

Open ThomasNobesMonash opened 1 year ago

ThomasNobesMonash commented 1 year ago

Hi, I am interested in working with PathBench for 3D pathfinding visualisation.

In your paper (https://doi.org/10.48550/arXiv.2105.01777) you mention that maps from MovingAI's Warframe voxel benchmarks are able to be easily integrated. However, I am struggling to understand the expected workflow for loading the external maps. I have had trouble attempting a direct conversion of the grid to a pickle file also .

I would appreciate any pointers in the right direction. Perhaps there is an old script laying around for parsing the Warframe benchmarks into a compatible format?

Cheers, Thomas

h2jaafar commented 1 year ago

@uncobruce Please take a look at this when you get a chance

uncobruce commented 1 year ago

Hi @ThomasNobesMonash,

For external maps, you can write scripts to parse them into .json format. The json file will have a dictionary that contains a nested list that specify the composition of the map ( 0.0 = free space, 1.0 = obstacle, 2.0 = agent start location, 3.0 = goal location) , and the agent start location/goal location for the map. Please see the screenshot below for a 2D map example. You would need to add another dimension in the nested grid list for 3D maps. The start/goal positions would then become [1, 23, 12] etc. Please try it out and let us know if it works.

Thanks, Bruce

Screenshot from 2023-05-26 17-23-02