AnchorVLA4D: an Anchor-Based Spatial-Temporal Vision-Language-Action Model for Robotic Manipulation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhu, Juan, Shao, Zhanying, Li, Xiaoqi, Morgan, Ethan, Xu, Jiadong, Fan, Hongwei, Dong, Hao
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908883808681984
author Zhu, Juan
Shao, Zhanying
Li, Xiaoqi
Morgan, Ethan
Xu, Jiadong
Fan, Hongwei
Dong, Hao
author_facet Zhu, Juan
Shao, Zhanying
Li, Xiaoqi
Morgan, Ethan
Xu, Jiadong
Fan, Hongwei
Dong, Hao
contents Since current Vision-Language-Action (VLA) systems suffer from limited spatial perception and the absence of memory throughout manipulation, we investigate visual anchors as a means to enhance spatial and temporal reasoning within VLA policies for robotic manipulation. Conventional VLAs generate actions by conditioning on a single current frame together with a language instruction. However, since the frame is encoded as a 2D image, it does not contain detailed spatial information, and the VLA similarly lacks any means to incorporate past context. As a result, it frequently forgets objects under occlusion and becomes spatially disoriented during the manipulation process. Thus, we propose AnchorVLA4D, a simple spatial-temporal VLA that augments the visual input with an anchor image to preserve the initial scene context throughout execution, and adds a lightweight spatial encoder that jointly processes the anchor and current frames to expose geometric relationships within an episode. Built on a Qwen2.5-VL backbone with a diffusion-based action head, AnchorVLA4D requires no additional sensing modalities (e.g., depth or point clouds) and introduces negligible inference overhead. Combining anchoring with a frozen pretrained spatial encoder yields further gains, realizing a 13.6% improvement on the Simpler WidowX benchmark and confirming the approach on real-world tasks, where it achieved an average success rate of 80%.
format Preprint
id arxiv_https___arxiv_org_abs_2603_12730
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle AnchorVLA4D: an Anchor-Based Spatial-Temporal Vision-Language-Action Model for Robotic Manipulation
Zhu, Juan
Shao, Zhanying
Li, Xiaoqi
Morgan, Ethan
Xu, Jiadong
Fan, Hongwei
Dong, Hao
Robotics
Since current Vision-Language-Action (VLA) systems suffer from limited spatial perception and the absence of memory throughout manipulation, we investigate visual anchors as a means to enhance spatial and temporal reasoning within VLA policies for robotic manipulation. Conventional VLAs generate actions by conditioning on a single current frame together with a language instruction. However, since the frame is encoded as a 2D image, it does not contain detailed spatial information, and the VLA similarly lacks any means to incorporate past context. As a result, it frequently forgets objects under occlusion and becomes spatially disoriented during the manipulation process. Thus, we propose AnchorVLA4D, a simple spatial-temporal VLA that augments the visual input with an anchor image to preserve the initial scene context throughout execution, and adds a lightweight spatial encoder that jointly processes the anchor and current frames to expose geometric relationships within an episode. Built on a Qwen2.5-VL backbone with a diffusion-based action head, AnchorVLA4D requires no additional sensing modalities (e.g., depth or point clouds) and introduces negligible inference overhead. Combining anchoring with a frozen pretrained spatial encoder yields further gains, realizing a 13.6% improvement on the Simpler WidowX benchmark and confirming the approach on real-world tasks, where it achieved an average success rate of 80%.
title AnchorVLA4D: an Anchor-Based Spatial-Temporal Vision-Language-Action Model for Robotic Manipulation
topic Robotics
url https://arxiv.org/abs/2603.12730