LLaDA-VLA: Vision Language Diffusion Action Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wen, Yuqing, Li, Hebei, Gu, Kefan, Zhao, Yucheng, Wang, Tiancai, Sun, Xiaoyan
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909780115718144
author Wen, Yuqing
Li, Hebei
Gu, Kefan
Zhao, Yucheng
Wang, Tiancai
Sun, Xiaoyan
author_facet Wen, Yuqing
Li, Hebei
Gu, Kefan
Zhao, Yucheng
Wang, Tiancai
Sun, Xiaoyan
contents The rapid progress of auto-regressive vision-language models (VLMs) has inspired growing interest in vision-language-action models (VLA) for robotic manipulation. Recently, masked diffusion models, a paradigm distinct from autoregressive models, have begun to demonstrate competitive performance in text generation and multimodal applications, leading to the development of a series of diffusion-based VLMs (d-VLMs). However, leveraging such models for robot policy learning remains largely unexplored. In this work, we present LLaDA-VLA, the first Vision-Language-Diffusion-Action model built upon pretrained d-VLMs for robotic manipulation. To effectively adapt d-VLMs to robotic domain, we introduce two key designs: (1) a localized special-token classification strategy that replaces full-vocabulary classification with special action token classification, reducing adaptation difficulty; (2) a hierarchical action-structured decoding strategy that decodes action sequences hierarchically considering the dependencies within and across actions. Extensive experiments demonstrate that LLaDA-VLA significantly outperforms state-of-the-art VLAs on both simulation and real-world robots.
format Preprint
id arxiv_https___arxiv_org_abs_2509_06932
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LLaDA-VLA: Vision Language Diffusion Action Models
Wen, Yuqing
Li, Hebei
Gu, Kefan
Zhao, Yucheng
Wang, Tiancai
Sun, Xiaoyan
Robotics
Computer Vision and Pattern Recognition
The rapid progress of auto-regressive vision-language models (VLMs) has inspired growing interest in vision-language-action models (VLA) for robotic manipulation. Recently, masked diffusion models, a paradigm distinct from autoregressive models, have begun to demonstrate competitive performance in text generation and multimodal applications, leading to the development of a series of diffusion-based VLMs (d-VLMs). However, leveraging such models for robot policy learning remains largely unexplored. In this work, we present LLaDA-VLA, the first Vision-Language-Diffusion-Action model built upon pretrained d-VLMs for robotic manipulation. To effectively adapt d-VLMs to robotic domain, we introduce two key designs: (1) a localized special-token classification strategy that replaces full-vocabulary classification with special action token classification, reducing adaptation difficulty; (2) a hierarchical action-structured decoding strategy that decodes action sequences hierarchically considering the dependencies within and across actions. Extensive experiments demonstrate that LLaDA-VLA significantly outperforms state-of-the-art VLAs on both simulation and real-world robots.
title LLaDA-VLA: Vision Language Diffusion Action Models
topic Robotics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2509.06932