MAP-Bench

Paper · Project · GitHub · Checkpoints

MAP-Bench evaluates instruction-conditioned navigation to a Manipulation Anchor Pose: the base position and heading from which a robot can begin the intended manipulation. Built in HM3D, it uses one forward RGB view and three distance levels: Short, Medium, and Long.

• 35,454 training trajectories with RGB observations, navigation actions, instructions, target boxes, and MAP annotations.
• 15,000 evaluation episodes with scene references, start poses, target MAPs, and instructions.
• 75,000 evaluation rollouts and their results: 15,000 each for GR00T N1.7, π₀.₅, FastWAM, UniWAM-Base, and UniWAM-Scaled.

Download

Run from your workspace directory to download the complete benchmark package (approximately 18.3 GB before extraction):

python -m pip install -U huggingface_hub
hf download Fysics-AI/MAP-Bench --repo-type dataset --local-dir ./MAP-Bench

Training videos and evaluation rollouts are stored in compressed archives. Use the GitHub download helper with --training --traces to verify and extract them. Obtain HM3D scene assets separately for simulator evaluation.

Usage

See GitHub for benchmark setup and evaluation, navigation training, and results.

HM3D-derived data are subject to the source dataset terms; see LICENSE.md.