World models learn clear action rules from videos for better prediction
VIDEAS: Distilling Explicit Action Semantics from Demonstration Videos for World Models via Prior-Guided Simulation
Robotics
Summary
Predicting how things will change when actions happen is hard, especially when machines need to understand the rules behind those changes. The authors created VIDEAS, a method that turns videos of actions into clear, step-by-step descriptions of what actions need to happen and what effects they cause. They use smart vision and language tools with a system that checks if these descriptions make sense physically. This helps build better world models that can plan and understand new situations more reliably.
What this means in practice
- •For robotics developers: Create robots that understand and plan physical actions by learning action rules directly from demonstration videos.
- •For game designers: Develop game AI that predicts game world changes from player actions using distilled action semantics from gameplay videos.
Authors
Jianan Wang, Haoquan Zhai, Siyang Zhang, Bin Li, Juan Chen, Jingtao Qi, Zhuo Zhang, Enze Wang, Haoxiang Jin, Chen Qian
Abstract
World models learn internal representations of environment dynamics to predict future states, enabling agents to optimize action plans without physical interactions. However, developing world models that genuinely internalize underlying causal physical laws to explicitly reason about action preconditions and subsequent state transitions remains an open challenge. In this paper, we propose VIDEAS, a data distillation framework that transforms continuous physical dynamics from operational videos into explicit action semantics for foundation models. Specifically, it deconstructs visual demonstrations into discrete action trajectories and utilizes advanced vision-language models (VLMs) to extract structured knowledge encapsulating action preconditions and effects. To ensure physical consistency, we introduce a prior-guided trajectory simulation mechanism grounded within a text-based environment to rigorously validate the extracted knowledge. Notably, we incorporate negative trajectories to enrich knowledge completeness and enhance data diversity to mitigate cognitive bias. Furthermore, we present VIDEAS-WM, an 8B/9B-parameter suite of language-based world models trained on 34K high-quality samples derived from AgiBot-World dataset. Extensive experiments demonstrate that VIDEAS-WM establishes state-of-the-art performance in high-level embodied action semantic reasoning, exhibiting profound physical understanding and robust generalization across unseen scenarios.