MessyNav at CoRL 2026: robots clear clutter out of the way with zero-shot VLM navigation
leto__jean · x · 2026-11-10
Stanford and UC Berkeley researchers present MessyNav, a zero-shot interactive navigation framework accepted to CoRL 2026.
- Core idea: Real-world paths are often blocked by everyday objects. When no collision-free path exists, the robot uses explicit VLM-based affordance reasoning before planning to decide whether to push, pick up, or avoid an object — e.g. avoid a cart of fragile glassware but push an empty one.
- Results: 85.7% affordance accuracy, 67% real-world success rate, and a 27pp higher success rate than baselines on a mobile manipulator, with no task-specific training.
Paper is available; code and video coming soon.
Related event: MessyNav teaches robots to push obstacles aside for zero-shot navigation(2 posts)→
More from Embodied
- Developer uses a 1.9B decision model on a laptop to flag dirty restaurant tables in real time — JosephJacks_ · 2026-11-10
- Fable 5.2 beats specialized robotics AI on dual-arm RoboDojo benchmark — FinanceYF5 · 2026-11-10
- LingBot-Touch: Open-Source Tactile Foundation Model Trained on ~5,000 Hours of Data — dair_ai · 2026-11-10
- Norway runs 2,000+ underwater AI laser robots delousing 50 million salmon around the clock — cephaloform · 2026-11-10
- Anthropic robotics model drop: bitter lesson may hold in robotics — beffjezos · 2026-11-10
- Rider shares first-person Cybercab experience: skipping downtown driving in Austin — yunta_tsai · 2026-11-10