AW-0 hybrid mode lifts RoboTwin success from 20.6% to 29.2% under visual randomization
trevordarrell · x · 2026-10-27
Part 6/8 of the AW-0 thread: code can guide and execute actions. In default mode, generated programs condition learned action chunks; in hybrid mode, program slices passing validity and progress checks execute through perception and control APIs, with the action chunk as fallback. Hybrid execution raises RoboTwin success from 20.6% to 29.2% under visual randomization and improves real towel folding/stacking rubric from 84.4 to 90.6, though direct action execution stays stronger in clean settings.
Related event: AW-0's Code Planning Sharpens Prediction and Boosts Execution(2 posts)→
More from Embodied
- DONIC hits 86.79% on bimanual dexterous tracking, crushing prior SOTA's 9.43% — liuziwei7 · 2026-10-27
- Bedrock Robotics raises $600M as its fully autonomous excavators do paid work on live job sites — Sethwinterroth · 2026-10-27
- AW-0 hits 80.8% on RoboTwin tasks, beating strongest baseline by 16 points — trevordarrell · 2026-10-27
- AW-0's code stream fixes world model prediction: picks mango where baseline grabs apple — trevordarrell · 2026-10-27
- AW-0 runs at 230ms on a single RTX 5090, 2.4x faster than autoregressive code planning — trevordarrell · 2026-10-27
- AW-0 debuts: one generative world action model plans in code, predicts, and controls robots — trevordarrell · 2026-10-27