When to Trust Imagination: Adaptive Action Execution for World Action Models

TL;DR

This paper introduces an adaptive execution framework for World Action Models in robotics, enabling robots to decide dynamically when to trust their predictions and replan, thereby improving efficiency and robustness in manipulation tasks.

Contribution

The paper proposes FFDC, a lightweight verifier for adaptive action execution, and Mixture-of-Horizon Training, enhancing long-horizon trajectory coverage for WAMs.

Findings

Reduces WAM forward passes by 69.10% on RoboTwin

Improves success rate by 2.54% over short-chunk baseline on RoboTwin

Increases real-world success rate by 35%

Abstract

World Action Models (WAMs) have recently emerged as a promising paradigm for robotic manipulation by jointly predicting future visual observations and future actions. However, current WAMs typically execute a fixed number of predicted actions after each model inference, leaving the robot blind to whether the imagined future remains consistent with the actual physical rollout. In this work, we formulate adaptive WAM execution as a future-reality verification problem: the robot should execute longer when the WAM-predicted future remains reliable, and replan earlier when reality deviates from imagination. To this end, we propose Future Forward Dynamics Causal Attention (FFDC), a lightweight verifier that jointly reasons over predicted future actions, predicted visual dynamics, real observations, and language instructions to estimate whether the remaining action rollout can still be trusted. FFDC enables adaptive action chunk sizes as an emergent consequence of prediction-observation consistency, preserving the efficiency of long-horizon execution while restoring responsiveness in contact-rich or difficult phases. We further introduce Mixture-of-Horizon Training to improve long-horizon trajectory coverage for adaptive execution. Experiments on the RoboTwin benchmark and in the real world demonstrate that our method achieves a strong robustness-efficiency trade-off: on RoboTwin, it reduces WAM forward passes by 69.10% and execution time by 34.02%, while improving success rate by 2.54% over the short-chunk baseline; in real-world experiments, it improves success rate by 35%.