CausalWAM: Learning the Causes of Actions in World Action Models
CausalWAM: Learning the Causes of Actions in World Action Models
This paper introduces CausalWAM (CWAM), a method designed to address limitations in World Action Models (WAMs) where action-relevant latent variables become entangled with nuisance variables. The authors prove that under a sparsity constraint, the latent variables used for generating actions are an invertible function of the action-relevant ones alone. CWAM incorporates this by adding a sparse routing path to the action head of a pretrained WAM while keeping the world modeling objective unchanged. Experiments on synthetic data and real-world robots demonstrate the theoretical validity and improved efficiency in closed-loop success.
The authors propose CausalWAM (CWAM), a method that modifies pretrained World Action Models by adding a sparse routing path to the action head to isolate latent variables that directly cause actions from those required for general scene prediction, validated on synthetic data and real-world robots.
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.
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
adding a sparse routing path to its action head, with the world modeling objective remaining unchanged. Experiments on both synthetic data and real-world robots validate the theory as well as showcase the improved efficiency in closed-loop success.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org