Robot model adapts mid-task after the hand is changed during rollout
E0M · x · 2026-07-24
A robot control experiment changed the hand/tool mid-rollout and let the same model continue running.
- The model noticed the new hand configuration and found a new trajectory and contact strategy to finish the task.
- The result suggests training on mixed data helps the model condition its behavior on the specific hand in front of it, moving it closer to a more general understanding of shapes, surfaces, and actuation strategies in the physical world.
More from Embodied
- NVIDIA open-sources FLORA, a robotics data engine ~125x cheaper than GPT-6 Astra — nvidia · 2026-10-27
- Embodied AI startups need a different operational discipline, investor says — ___Mufasaa · 2026-10-27
- π0.5 study: raising demos from 10 to 438 lifts Place Ball in Bowl success from 7% to 83% — stingning · 2026-10-27
- GPT-6 Astra controls dexterous hands zero-shot at just 5.2% success in 500 trials — stingning · 2026-10-27
- HandsOn benchmark: 66 verified tasks and 18,000 human demos for dexterous manipulation — stingning · 2026-10-27
- DexPort: a handheld interface with a real robot hand for dexterous-hand data collection — pabbeel · 2026-10-27