Project website for GemNav: Discrete-Token Visual Robot Navigation using a Multimodal Large Language Model (arXiv:2607.06882).
python3 -m http.server 8000
# open http://localhost:8000Clips in static/videos/ were cut from the master video gemnav_v5_review_720p.mp4 and re-encoded (H.264, 720p, no audio except gemnav_overview.mp4). To swap a clip, replace the file with one of the same name. A missing file shows a "coming soon" box instead.
| File | Where it appears |
|---|---|
teaser.mp4 |
Top of page |
anymal_image_goal.mp4, rover_pose_offroad.mp4, spot_workshop.mp4, detour_course.mp4 |
Rollout carousel (some reused below) |
gemnav_overview.mp4 |
Video section (full master video) |
warehouse_comparison.mp4, chemical_yard_comparison.mp4 |
Pose-conditioned navigation |
offroad_three_robots.mp4 |
Off-road navigation |
Figures in static/images/ are rendered from the paper's resources/ folder.
Adapted from the Nerfies template, licensed under CC BY-SA 4.0.