Rollback: World Models Rewind Robot Decisions to Explore Counterfactual Actions
agibot-world · hf · 2026-10-28
AgiBot, Beihang University and NUS introduce Rollback, a policy improvement paradigm that turns world models from passive simulators into counterfactual engines.
- The method rewinds to past critical decision points, branches into alternative actions, and contrasts outcomes to attribute success to individual decisions, producing fine-grained supervision.
- Dynamics are continuously refined via real-world feedback to limit sim-to-real drift.
- Motivation: real robot interaction is costly; existing world-model methods use forward-only rollouts, leaving the counterfactual "rewind" capability untapped.
Project page: rollback-wm.github.io
More from Embodied
- Reward AI's OM-1 Learns Manipulation Skills From Human Demonstrations for Cross-Body Transfer — CyberRobooo · 2026-10-28
- 4 Stanford freshmen build assistive walker robot in 36 hours, win grand prize at CalHacks 13.0 — Scobleizer · 2026-10-28
- Just Action Model: decoupling what separates VLA from world-action models — Yonghao Yu · 2026-10-28
- NVIDIA's ThinkWorld learns robot skills from unlabeled human videos via chain-of-dynamics reasoning — nvidia · 2026-10-28
- NVIDIA Research releases Foundation4D: unified 4D perception model turns camera video into 3D scenes, no lidar needed — Scobleizer · 2026-10-28
- Dyna Robotics visits four hotel laundry rooms before starting Taku POC — JasonMa2020 · 2026-10-28