Open-source VLAs still struggle with basic representation alignment, researcher argues
ChongZzZhang · x · 2026-09-09
Arguing that until there's a real advance in open-vocabulary zero-shot setups, all VLAs are still struggling to align basic representations — and open-source VLA claims of a "GPT-3 moment" can't be independently verified.
Related event: Tiny 540K-Parameter Robot Policy Hits 95% Success(2 posts)→
More from Embodied
- PGMT paper: robot tracks flat-ground reference on stairs and 40cm boxes — ChongZzZhang · 2026-09-09
- Humanoid robot learned a real factory task from just 6 demos, ran 4 days at 96% success — lukas_m_ziegler · 2026-09-09
- Maker converts SO-ARM100 robot arm into a biped robot — k7agar · 2026-09-09
- AMD unveils Threadripper Halo workstation for serious local AI — Apprehensive_Bar6609 · 2026-09-09
- Ex-autonomous driving e2e tech lead Zhang Ying launches Physical AI startup, raises hundreds of millions of RMB in a month — 机器之心 · 2026-09-09
- Action Map Policy turns robot actions into pixel classification, beating Diffusion Policy by 50-70% on real-world tasks — 机器之心 · 2026-09-09