HumanCLAW: best frontier VLM completes only 16.8% of embodied interaction tasks
jiqizhixin · x · 2026-09-13
Meta, NTU, UW, Brown, and Northwestern present HumanCLAW, testing whether vision-language models can actually act through a body rather than just describe what they see.
Key points:
- Separates high-level action selection from low-level motor control while preserving continuous motion, collision, contact, and gravity
- A frozen off-the-shelf VLM makes decisions from a first-person view, choosing each step from parameterized atomic skills (move forward, turn, sit down)
- The framework studies "what to do", leaving joint-level control to a reliable low-level controller
Result: nine frontier VLMs tested under the same protocol — the best completes only 16.8% of full interactions, a stark finding that understanding the world is not the same as acting in it. Fully open-sourced.
More from Embodied
- NVIDIA's MotionBricks robotics motion-generation model trends on Hugging Face — nvidia · 2026-10-31
- Agility Robotics to raise $465M ahead of Nasdaq debut as AGLT on Nov 5, Foxconn leads with $201M — xmercury_one · 2026-10-31
- Tesla to open Optimus humanoid robot reservations in early December, per source — Scobleizer · 2026-10-31
- New MolmoSpaces-CommonSense benchmark: no robot policy tops 30% on manipulation tasks — RanjayKrishna · 2026-10-31
- Neuralink Unveils Next-Gen R1 Surgical Robot That Implants Brain Chips — elonmusk · 2026-10-31
- Google demos Sprout, a semi-autonomous rover that hunts sunlight to care for office plants — AI_Andrew · 2026-10-31