aiqbench analysis
This paper introduces EWAM, a zero-shot adaptation framework for embodied intelligence that enhances the frozen Cosmos3 backbone via inference-time reasoning modules rather than fine-tuning. By integrating memory, anomaly detection, policy routing, and action correction layers directly into the Diffusion Transformer forward path, it significantly reduces deployment data requirements without altering the pretrained model weights. The approach represents a notable advancement in efficient online adaptation for robotic agents operating under zero-shot protocols.
EWAM: An Enhanced World Action Model for Closed-Loop Online Adaptation in Embodied Intelligence
arXiv:2606.12690v1 Announce Type: cross Abstract: In this paper, we propose the Enhanced World Action Model (EWAM), a closed-loop online adaptation architecture built upon a pretrained and fully frozen Cosmos3 backbone network. Evaluated entirely under a zero-shot task protocol, EWAM is centrally focused on reducing the amount of additional deployment data required to adapt to new task layouts. Notably, no extra task-specific demonstration sets were introduced in any of the evaluations, and no fine-tuning was performed on the backbone network. Its performance gains stem entirely from an inference-time co-reasoning mechanism composed of four inserted lightweight neural layers: the Neural Experience Memory Layer located in the intermediate layers of the Diffusion Transformer (DiT) provides task-relevant execution context; the Neural Anomaly Detection Layer after the state prediction head monitors the divergence between predicted and actual states in real time; the Neural Policy Routing Layer dynamically selects direct execution, conservative replanning, or rollback recovery based on the anomaly severity; and the Neural Action Correction Layer refines the generated action chunks using execution diagnostics. Unlike naive feature fusion, the memory, anomaly detection, and correction modules are deeply integrated into the Cosmos3 forward path in a differentiable manner, with only the final routing decision being a discrete supervised one.
Abstract:In this paper, we propose the Enhanced World Action Model (EWAM), a closed-loop online adaptation architecture built upon a pretrained and fully frozen Cosmos3 backbone network. Evaluated entirely under a zero-shot task protocol, EWAM is centrally focused on reducing the amount of additional deployment data required to adapt to new task layouts. Notably, no extra task-specific demonstration sets were introduced in any of the evaluations, and no fine-tuning was performed on the backbone network. Its performance gains stem entirely from an inference-time co-reasoning mechanism composed of four inserted lightweight neural layers: the Neural Experience Memory Layer located in the intermediate layers of the Diffusion Transformer (DiT) provides task-relevant execution context; the Neural Anomaly Detection Layer after the state prediction head monitors the divergence between predicted and actual states in real time; the Neural Policy Routing Layer dynamically selects direct execution, conservative replanning, or rollback recovery based on the anomaly severity; and the Neural Action Correction Layer refines the generated action chunks using execution diagnostics. Unlike naive feature fusion, the memory, anomaly detection, and correction modules are deeply integrated into the Cosmos3 forward path in a differentiable manner, with only the final routing decision being a discrete supervised one.
Robotics (cs.RO); Artificial Intelligence (cs.AI)
arXiv:2606.12690 [cs.RO]
(or arXiv:2606.12690v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.12690 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 294 words
Related coverage
Links #3: 2026/06 Part 1
LessWrong · Jun 15, 2026
OnlineWM: Causality-Aware Active Online Learning for Effective World Modeling
arXiv · Sep 22, 2026
NVIDIA GTC Taipei at COMPUTEX: Live Updates on What’s Next in AI
NVIDIA · May 21, 2026
Vertical Advantage: Transforming Industries with Lakebase and Agentic AI
Databricks · Aug 27, 2026