Paper Detail
Guiyu Zhao, Longteng Guo, Yanghong Mei, Zilin Zhu, Yu Zhang, Bin Cao, Mingming Yu, Xingjian He, Jie Jiang, Jing Liu
While Vision-Language-Action (VLA) models have advanced embodied AI, their fundamentally reactive paradigm severely limits performance in partially observable and long-horizon tasks. When restricted to a single wrist-mounted camera, they inevitably suffer from perception forgetting as objects exit the field of view, and temporal task-progress forgetting} during multi-step execution. To overcome these bottlenecks, we propose AtlasVLA, a novel framework that transitions from direct reactive manipulation to proactive reasoning through a persistent world-ego state. AtlasVLA features a dual-memory architecture: a 4D Persistent World State Memory that lifts transient 2D observations into a globally updated, voxel-hashed spatial state to resolve visual blind spots, and an Ego-Working State Memory that tracks historical ego state and task progress. By conditioning a diffusion transformer (DiT) on this joint World-Ego state, AtlasVLA enables robust spatial reasoning. Extensive evaluations across LIBERO, RLBench, and real-world benchmarks demonstrate that AtlasVLA achieves state-of-the-art performance using solely a wrist camera. Remarkably, it decisively outperforms multi-view baselines, yielding absolute success rate improvements of 9.4% on LIBERO-Long and 17.5% in real-world long-horizon tasks.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{zhao2026atlasvla,
title = {AtlasVLA: Persistent World-Ego State Modeling for Vision-Language-Action Models},
author = {Guiyu Zhao and Longteng Guo and Yanghong Mei and Zilin Zhu and Yu Zhang and Bin Cao and Mingming Yu and Xingjian He and Jie Jiang and Jing Liu},
year = {2026},
abstract = {While Vision-Language-Action (VLA) models have advanced embodied AI, their fundamentally reactive paradigm severely limits performance in partially observable and long-horizon tasks. When restricted to a single wrist-mounted camera, they inevitably suffer from perception forgetting as objects exit the field of view, and temporal task-progress forgetting\} during multi-step execution. To overcome these bottlenecks, we propose AtlasVLA, a novel framework that transitions from direct reactive manipu},
url = {https://huggingface.co/papers/2608.06729},
keywords = {Vision-Language-Action models, 4D Persistent World State Memory, voxel-hashed spatial state, Ego-Working State Memory, diffusion transformer, DiT, world-ego state, huggingface daily},
eprint = {2608.06729},
archiveprefix = {arXiv},
}
{}