ArtiBench and ArtiBrain: Benchmarking Generalizable Vision-Language Articulated Object Manipulation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wu, Yuhan, Wei, Tiantian, Wang, Shuo, Wang, ZhiChao, Zhang, Yanyong, Cremers, Daniel, Xia, Yan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915641866321920
author Wu, Yuhan
Wei, Tiantian
Wang, Shuo
Wang, ZhiChao
Zhang, Yanyong
Cremers, Daniel
Xia, Yan
author_facet Wu, Yuhan
Wei, Tiantian
Wang, Shuo
Wang, ZhiChao
Zhang, Yanyong
Cremers, Daniel
Xia, Yan
contents Interactive articulated manipulation requires long-horizon, multi-step interactions with appliances while maintaining physical consistency. Existing vision-language and diffusion-based policies struggle to generalize across parts, instances, and categories. We first introduce ArtiBench, a five-level benchmark covering kitchen, storage, office, and tool environments. ArtiBench enables structured evaluation from cross-part and cross-instance variation to long-horizon multi-object tasks, revealing the core generalization challenges of articulated object manipulation. Building on this benchmark, we propose ArtiBrain, a modular framework that unifies high-level reasoning with adaptive low-level control. ArtiBrain uses a VLM-based Task Reasoner (GPT-4.1) to decompose and validate subgoals, and employs a Hybrid Controller that combines geometry-aware keyframe execution with affordance-guided diffusion for precise and interpretable manipulation. An Affordance Memory Bank continually accumulates successful execution episodes and propagates part-level actionable affordances to unseen articulated parts and configurations. Extensive experiments on ArtiBench show that our ArtiBrain significantly outperforms state-of-the-art multimodal and diffusion-based methods in robustness and generalization. Code and dataset will be released upon acceptance.
format Preprint
id arxiv_https___arxiv_org_abs_2511_20330
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ArtiBench and ArtiBrain: Benchmarking Generalizable Vision-Language Articulated Object Manipulation
Wu, Yuhan
Wei, Tiantian
Wang, Shuo
Wang, ZhiChao
Zhang, Yanyong
Cremers, Daniel
Xia, Yan
Robotics
Computer Vision and Pattern Recognition
Interactive articulated manipulation requires long-horizon, multi-step interactions with appliances while maintaining physical consistency. Existing vision-language and diffusion-based policies struggle to generalize across parts, instances, and categories. We first introduce ArtiBench, a five-level benchmark covering kitchen, storage, office, and tool environments. ArtiBench enables structured evaluation from cross-part and cross-instance variation to long-horizon multi-object tasks, revealing the core generalization challenges of articulated object manipulation. Building on this benchmark, we propose ArtiBrain, a modular framework that unifies high-level reasoning with adaptive low-level control. ArtiBrain uses a VLM-based Task Reasoner (GPT-4.1) to decompose and validate subgoals, and employs a Hybrid Controller that combines geometry-aware keyframe execution with affordance-guided diffusion for precise and interpretable manipulation. An Affordance Memory Bank continually accumulates successful execution episodes and propagates part-level actionable affordances to unseen articulated parts and configurations. Extensive experiments on ArtiBench show that our ArtiBrain significantly outperforms state-of-the-art multimodal and diffusion-based methods in robustness and generalization. Code and dataset will be released upon acceptance.
title ArtiBench and ArtiBrain: Benchmarking Generalizable Vision-Language Articulated Object Manipulation
topic Robotics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.20330