Rethinking Video Generation Model for the Embodied World

TL;DR

This paper introduces RBench, a comprehensive benchmark for evaluating robot-oriented video generation, and RoVid-X, a large-scale dataset, to improve the realism and assessment of embodied AI models.

Contribution

It presents a new standardized benchmark and a large annotated dataset to advance the evaluation and training of high-quality robot video generation models.

Findings

Significant deficiencies in current models' physical realism.

High correlation (0.96) between RBench scores and human judgments.

The new dataset enables scalable training of embodied AI models.

Abstract

Video generation models have significantly advanced embodied intelligence, unlocking new possibilities for generating diverse robot data that capture perception, reasoning, and action in the physical world. However, synthesizing high-quality videos that accurately reflect real-world robotic interactions remains challenging, and the lack of a standardized benchmark limits fair comparisons and progress. To address this gap, we introduce a comprehensive robotics benchmark, RBench, designed to evaluate robot-oriented video generation across five task domains and four distinct embodiments. It assesses both task-level correctness and visual fidelity through reproducible sub-metrics, including structural consistency, physical plausibility, and action completeness. Evaluation of 25 representative models highlights significant deficiencies in generating physically realistic robot behaviors. Furthermore, the benchmark achieves a Spearman correlation coefficient of 0.96 with human evaluations, validating its effectiveness. While RBench provides the necessary lens to identify these deficiencies, achieving physical realism requires moving beyond evaluation to address the critical shortage of high-quality training data. Driven by these insights, we introduce a refined four-stage data pipeline, resulting in RoVid-X, the largest open-source robotic dataset for video generation with 4 million annotated video clips, covering thousands of tasks and enriched with comprehensive physical property annotations. Collectively, this synergistic ecosystem of evaluation and data establishes a robust foundation for rigorous assessment and scalable training of video models, accelerating the evolution of embodied AI toward general intelligence.