LabVLA: Empowering AI to Run Real-World Lab Experiments
jiqizhixin · x · 2026-07-06
Zhejiang University, Shanghai AI Lab, and Harbin Institute of Technology proposed LabVLA to enable VLA models to autonomously conduct experiments in real scientific labs. The team built the RoboGenesis simulation engine to generate training data for lab robots, utilizing a two-stage training approach: learning motion understanding first, followed by fine-tuning with flow matching for precise control. On the LabUtopia benchmark, LabVLA achieved the highest success rates on both familiar and entirely new tasks.
More from Embodied
- NUS builds a soft force sensor that drives actuators without electronics or power — CurieuxExplorer · 2026-07-27
- Chelsea Finn says robot RL is bottlenecked by physical rollout cost, not algorithms — ycombinator · 2026-07-27
- Robot goes to the fridge and fetches a beer — Darpinian · 2026-07-27
- Researchers show digital circuits can be replicated with knitted fabric — mtizard · 2026-07-27
- Local Qwen models power a robot that tests 78 smartphones’ battery life — gappyvalley · 2026-07-27
- TechCrunch says brain-wave signals could be the next unlock for physical AI training — TechCrunch AI · 2026-07-27