La-JEPA: Compact Robot Policies with Latent World Modeling

原文：La-JEPA: Compact Robot Policies with Latent World Modeling

作者：Mateo Clémente, George Z. Xue, Amin Abyaneh, Liheng Ma, Rui Heng Yang, Tongtong Cao, Zhanguang Zhang, Yingxue Zhang

来源：arXiv cs.LG（机器学习）

正文

Computer Science > Robotics

arXiv:2611.08156v1 (cs)

[Submitted on 5 Nov 2026]

Title:La-JEPA: Compact Robot Policies with Latent World Modeling

Authors:Mateo Clémente, George Z. Xue, Amin Abyaneh, Liheng Ma, Rui Heng Yang, Tongtong Cao, Zhanguang Zhang, Yingxue Zhang

View a PDF of the paper titled La-JEPA: Compact Robot Policies with Latent World Modeling, by Mateo Cl'emente and 7 other authors

View PDF

HTML (experimental)

Abstract:World-action models adapt pretrained video generators, and vision-language-action models build on pretrained vision-language backbones; both tie policy architecture and inference cost to large transformers designed for generation rather than control. It remains unclear how much these generation-oriented choices affect control. In this work, we introduce Latent Action JEPA (La-JEPA), a compact world-action model whose policy transformer is trained from scratch in the feature space of a pretrained V-JEPA 2.1 encoder. Following Fast-WAM, La-JEPA learns to predict future features jointly with actions during training and generates only actions at inference. Among policies without embodied pretraining, La-JEPA achieves state-of-the-art performance in clean and randomized RoboTwin 2.0 scenes and on LIBERO-Plus with correct instructions, while using far fewer parameters than the strongest competing policies. It runs substantially faster than Fast-WAM at comparable performance and outperforms the closest JEPA-based methods, LeapBot-WA and JEPA-WAM, on LIBERO-Plus and RoboTwin 2.0 without their pretrained policy backbones. On four real-robot tasks, its average success is comparable to pi0.5 and above non-embodied-pretrained baselines. Through a design study of the choices behind La-JEPA, we show that JEPA features outperform video VAE latents, most clearly out of distribution, and that training-time future prediction choices significantly impact the policy even when future prediction is skipped at inference. We correct the LIBERO-Plus instructions, as the officially released evaluator corrupts them for most variants, and re-evaluate several strong open baselines. Without a generative pretrained policy backbone, La-JEPA is better and faster than far larger policies, suggesting that robot foundation models may benefit from architectures designed for control rather than inherited from generation.

Comments:

30 pages, 6 figures. Code: this https URL

Subjects:

Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)

Cite as:

arXiv:2611.08156 [cs.RO]

(or

arXiv:2611.08156v1 [cs.RO] for this version)

https://doi.org/10.48550/arXiv.2611.08156

Focus to learn more

arXiv-issued DOI via DataCite (pending registration)

主题

机器人 · 计算机视觉 · 机器学习

由「前沿雷达」于 2026-11-05 采集。正文取自原文页面，已保留出处链接。标题与正文版权归原作者所有。