GigaWorld-Policy-0.5: A Faster and Stronger WAM Empowered by AutoResearch

2026-07-15

AI Summary

This work addresses the high computational cost of existing World Action Models (WAMs), which require explicit future video generation during inference and thus hinder real-time robotic control. The authors propose an action-centric world modeling framework that leverages future visual dynamics to provide dense supervision during training but decodes only actions at inference time, substantially improving efficiency. Their approach innovatively integrates Action-Conditioned World Modeling (AC-WM) with WAM through a hybrid pretraining strategy to strengthen the coupling between actions and visual dynamics. Additionally, they introduce a Mixture-of-Transformers architecture to disentangle task-specific modeling and employ an agent-based AutoResearch system to automatically optimize training configurations. The resulting model achieves an inference latency of 85 milliseconds on an RTX 4090, significantly enhancing policy deployment efficiency while preserving the benefits of rich training-time supervision.

Abstract

World Action Models (WAMs) improve robot policy learning by jointly modeling actions and future visual observations, using future scene evolution as dense supervision for physically grounded action generation. However, a common design in existing WAMs is to explicitly generate future videos at inference time, incurring substantial computational overhead and hindering real-time closed-loop deployment. GigaWorld-Policy addresses this issue with an action-centered formulation, where future visual dynamics are used during training while action-only decoding is used at inference time. Building upon this framework, we present GigaWorld-Policy-0.5, an enhanced action-centered WAM designed for more efficient robot control. During pretraining, GigaWorld-Policy-0.5 adopts a mixed Action-Conditioned World Modeling (AC-WM) and WAM training strategy. This strengthens the coupling between visual dynamics and robot actions and improves the transferability of action representations for downstream policy learning. For efficient inference, GigaWorld-Policy-0.5 introduces a Mixture-of-Transformers architecture that separates visual dynamics modeling and action generation into specialized experts, reducing active computation during action-only inference and achieving 85 ms inference latency on a local RTX 4090 setup. In addition, we employ an agent-based AutoResearch pipeline to systematically search training configurations, enabling more efficient identification of optimal experimental setups while reducing the time and manual intervention required for hyperparameter tuning. Experiments and ablations show that GigaWorld-Policy-0.5 preserves the training benefits of future visual dynamics while improving inference efficiency for robot control.

Problem

• World Action Models
• robot policy learning
• inference efficiency
• real-time control
• future video generation

Innovation

• Action-Centered WAM
• Mixture-of-Transformers
• AutoResearch
• Efficient Robot Policy
• Visual Dynamics Modeling