FrodoBots
Collection
Robot navigation policies trained with the FrodoBots8K dataset • 12 items • Updated
PointGoal navigation policy trained with the FrodoBots8K dataset.
Presented in the paper Data scaling for navigation in unknown environments
Architecture:
NoMAD-like, Theia encoder + transformer backbone + UNet denoiser
Dataset:
1024H total, 32H from 32 distinct geographical locations
See the Github repo