WoW: Towards a World omniscient World model Through Embodied Interaction

2025-09-26

AI Summary

Current video models (e.g., Sora) rely on passive observation and struggle to model physical causality, whereas humans develop intuitive physics understanding through active interaction. To address this, we propose an embodied learning paradigm driven by large-scale real-world robot interactions—2 million diverse trajectories—and introduce WoW, a 14B-parameter generative world model. Methodologically, WoW employs a DiT architecture to synthesize physically consistent videos and integrates SOPHIA: a vision-language-model-based mechanism that dynamically constrains and refines generation via verifiable physical reasoning. Additionally, an inverse dynamics model is co-trained to close the loop among “imagination,” “planning,” and “action.” Evaluated on our newly constructed WoWBench benchmark, WoW achieves substantial gains in collision dynamics modeling, object permanence, and causal reasoning—marking the first successful large-scale, interaction-driven instantiation of intuitive physical reasoning.

Technology Category

• Intelligent Robots: Embodied AI
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Computer Vision: Low Level & Physics-based Vision

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Humans develop an understanding of intuitive physics through active interaction with the world. This approach is in stark contrast to current video models, such as Sora, which rely on passive observation and therefore struggle with grasping physical causality. This observation leads to our central hypothesis: authentic physical intuition of the world model must be grounded in extensive, causally rich interactions with the real world. To test this hypothesis, we present WoW, a 14-billion-parameter generative world model trained on 2 million robot interaction trajectories. Our findings reveal that the model's understanding of physics is a probabilistic distribution of plausible outcomes, leading to stochastic instabilities and physical hallucinations. Furthermore, we demonstrate that this emergent capability can be actively constrained toward physical realism by SOPHIA, where vision-language model agents evaluate the DiT-generated output and guide its refinement by iteratively evolving the language instructions. In addition, a co-trained Inverse Dynamics Model translates these refined plans into executable robotic actions, thus closing the imagination-to-action loop. We establish WoWBench, a new benchmark focused on physical consistency and causal reasoning in video, where WoW achieves state-of-the-art performance in both human and autonomous evaluation, demonstrating strong ability in physical causality, collision dynamics, and object permanence. Our work provides systematic evidence that large-scale, real-world interaction is a cornerstone for developing physical intuition in AI. Models, data, and benchmarks will be open-sourced.

Problem

• Developing AI world models through embodied robot interactions
• Addressing physical causality gaps in passive video models
• Creating physically consistent video generation with causal reasoning

Innovation

• Training world model with robot interaction trajectories
• Using vision-language agents to refine generated outputs
• Closing imagination-action loop via inverse dynamics model