UP-VLA: A Unified Understanding and Prediction Model for Embodied Agent

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Jianke, Guo, Yanjiang, Hu, Yucheng, Chen, Xiaoyu, Zhu, Xiang, Chen, Jianyu
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911022384676864
author Zhang, Jianke
Guo, Yanjiang
Hu, Yucheng
Chen, Xiaoyu
Zhu, Xiang
Chen, Jianyu
author_facet Zhang, Jianke
Guo, Yanjiang
Hu, Yucheng
Chen, Xiaoyu
Zhu, Xiang
Chen, Jianyu
contents Recent advancements in Vision-Language-Action (VLA) models have leveraged pre-trained Vision-Language Models (VLMs) to improve the generalization capabilities. VLMs, typically pre-trained on vision-language understanding tasks, provide rich semantic knowledge and reasoning abilities. However, prior research has shown that VLMs often focus on high-level semantic content and neglect low-level features, limiting their ability to capture detailed spatial information and understand physical dynamics. These aspects, which are crucial for embodied control tasks, remain underexplored in existing pre-training paradigms. In this paper, we investigate the training paradigm for VLAs, and introduce \textbf{UP-VLA}, a \textbf{U}nified VLA model training with both multi-modal \textbf{U}nderstanding and future \textbf{P}rediction objectives, enhancing both high-level semantic comprehension and low-level spatial understanding. Experimental results show that UP-VLA achieves a 33% improvement on the Calvin ABC-D benchmark compared to the previous state-of-the-art method. Additionally, UP-VLA demonstrates improved success rates in real-world manipulation tasks, particularly those requiring precise spatial information.
format Preprint
id arxiv_https___arxiv_org_abs_2501_18867
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle UP-VLA: A Unified Understanding and Prediction Model for Embodied Agent
Zhang, Jianke
Guo, Yanjiang
Hu, Yucheng
Chen, Xiaoyu
Zhu, Xiang
Chen, Jianyu
Computer Vision and Pattern Recognition
Artificial Intelligence
Recent advancements in Vision-Language-Action (VLA) models have leveraged pre-trained Vision-Language Models (VLMs) to improve the generalization capabilities. VLMs, typically pre-trained on vision-language understanding tasks, provide rich semantic knowledge and reasoning abilities. However, prior research has shown that VLMs often focus on high-level semantic content and neglect low-level features, limiting their ability to capture detailed spatial information and understand physical dynamics. These aspects, which are crucial for embodied control tasks, remain underexplored in existing pre-training paradigms. In this paper, we investigate the training paradigm for VLAs, and introduce \textbf{UP-VLA}, a \textbf{U}nified VLA model training with both multi-modal \textbf{U}nderstanding and future \textbf{P}rediction objectives, enhancing both high-level semantic comprehension and low-level spatial understanding. Experimental results show that UP-VLA achieves a 33% improvement on the Calvin ABC-D benchmark compared to the previous state-of-the-art method. Additionally, UP-VLA demonstrates improved success rates in real-world manipulation tasks, particularly those requiring precise spatial information.
title UP-VLA: A Unified Understanding and Prediction Model for Embodied Agent
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2501.18867