ReWorld: Multi-Dimensional Reward Modeling for Embodied Video Generation
ReWorld: Multi-Dimensional Reward Modeling for Embodied Video Generation
This research presents ReWorld, a framework that applies reinforcement learning to improve physical fidelity, dynamic consistency, and task logic in embodied video generation. The authors construct a large-scale (~235K) video preference dataset to train a hierarchical reward model capturing multi-dimensional human-aligned rewards. A practical alignment algorithm post-trains flow-based video generators using this reward via an efficient PPO-style method. Experiments show ReWorld significantly enhances physical fidelity, logical coherence, embodiment plausibility, and visual quality compared to prior methods.
The paper introduces ReWorld, a framework using reinforcement learning to align embodied video generation with physical realism and task logic, supported by a ~235K video preference dataset and PPO-style post-training; claims of outperformance are based on comprehensive experiments within the study but require independent verification for deployment readiness.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
video preference dataset size
≈235,000
View original evidence
o generation with physical realism, task completion capability, embodiment plausibility and visual quality. Specifically, we first construct a large-scale (~235K) video preference dataset and employ it to train a hierarchical reward model designed to capture multi-dimensional reward consistent with human preferences. We further propose a practical alignment algorithm that post-trains flow-based video generators using
Open source S3
Dataset details reported by the source
- Modalities
- video Open source S3
- paper: not_reported
What remains unknown
Dataset licensing and commercial-use rights are not established here.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
o generation with physical realism, task completion capability, embodiment plausibility and visual quality. Specifically, we first construct a large-scale (~235K) video preference dataset and employ it to train a hierarchical reward model designed to capture multi-dimensional reward consistent with human preferences. We further propose a practical alignment algorithm that post-trains flow-based video generators using
Open source S3
Implications for data suppliers
RoboSignal interpretation and collection questions, not statements of buyer demand.
- Confirm the required data type and collection setting with the buyer; this source does not establish a complete collection specification.
- Compare the reported units and scope before using these quantities in a budget. Recording hours, sensor-hours and trajectories are different measures.
- Check the exact dataset version, license, modalities and task coverage against the intended use.
- Validate demand and acceptance criteria with a buyer before scaling. Publication, popularity and a research result do not establish a purchase commitment.
Source：arXiv Robotics — research abstracts · arxiv.org