Saved in:
Bibliographic Details
Main Authors: Rasouli, Amir, Wu, Yangzheng, Li, Zhiyuan, Yang, Rui Heng, Zhao, Xuan, Eret, Charles, Pakdamansavoji, Sajjad
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.21192
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910159102541824
author Rasouli, Amir
Wu, Yangzheng
Li, Zhiyuan
Yang, Rui Heng
Zhao, Xuan
Eret, Charles
Pakdamansavoji, Sajjad
author_facet Rasouli, Amir
Wu, Yangzheng
Li, Zhiyuan
Yang, Rui Heng
Zhao, Xuan
Eret, Charles
Pakdamansavoji, Sajjad
contents Vision-language-action models (VLAs) have been extensively used in robotics applications, achieving great success in various manipulation problems. More recently, VLAs have been used in long-horizon tasks and evaluated on benchmarks, such as BEHAVIOR1K (B1K), for solving complex household chores. The common metric for measuring progress in such benchmarks is success rate or partial score based on satisfaction of progress-agnostic criteria, meaning only the final states of the objects are considered, regardless of the events that lead to such states. In this paper, we argue that using such evaluation protocols say little about safety aspects of operation and can potentially exaggerate reported performance, undermining core challenges for future real-world deployment. To this end, we conduct a thorough analysis of state-of-the-art models on the B1K Challenge and evaluate policies in terms of robustness via reproducibility and consistency of performance, safety aspects of policies operations, task awareness, and key elements leading to the incompletion of tasks. We then propose evaluation protocols to capture safety violations to better measure the true performance of the policies in more complex and interactive scenarios. At the end, we discuss the limitations of the existing VLAs and motivate future research.
format Preprint
id arxiv_https___arxiv_org_abs_2604_21192
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle How VLAs (Really) Work In Open-World Environments
Rasouli, Amir
Wu, Yangzheng
Li, Zhiyuan
Yang, Rui Heng
Zhao, Xuan
Eret, Charles
Pakdamansavoji, Sajjad
Robotics
Artificial Intelligence
Vision-language-action models (VLAs) have been extensively used in robotics applications, achieving great success in various manipulation problems. More recently, VLAs have been used in long-horizon tasks and evaluated on benchmarks, such as BEHAVIOR1K (B1K), for solving complex household chores. The common metric for measuring progress in such benchmarks is success rate or partial score based on satisfaction of progress-agnostic criteria, meaning only the final states of the objects are considered, regardless of the events that lead to such states. In this paper, we argue that using such evaluation protocols say little about safety aspects of operation and can potentially exaggerate reported performance, undermining core challenges for future real-world deployment. To this end, we conduct a thorough analysis of state-of-the-art models on the B1K Challenge and evaluate policies in terms of robustness via reproducibility and consistency of performance, safety aspects of policies operations, task awareness, and key elements leading to the incompletion of tasks. We then propose evaluation protocols to capture safety violations to better measure the true performance of the policies in more complex and interactive scenarios. At the end, we discuss the limitations of the existing VLAs and motivate future research.
title How VLAs (Really) Work In Open-World Environments
topic Robotics
Artificial Intelligence
url https://arxiv.org/abs/2604.21192