Understanding Physical Dynamics with Counterfactual World Modeling
This work addresses the challenge of physical dynamics understanding for agents, representing an incremental improvement with a novel method for a known bottleneck.
The paper tackles the problem of understanding physical dynamics by using Counterfactual World Modeling (CWM) to extract visual structures from video data without annotations, achieving state-of-the-art performance on the Physion benchmark.
The ability to understand physical dynamics is critical for agents to act in the world. Here, we use Counterfactual World Modeling (CWM) to extract vision structures for dynamics understanding. CWM uses a temporally-factored masking policy for masked prediction of video data without annotations. This policy enables highly effective "counterfactual prompting" of the predictor, allowing a spectrum of visual structures to be extracted from a single pre-trained predictor without finetuning on annotated datasets. We demonstrate that these structures are useful for physical dynamics understanding, allowing CWM to achieve the state-of-the-art performance on the Physion benchmark.