Revealing Interpretable Failure Modes of VLMs

TL;DR

REVELIO is a framework that systematically uncovers interpretable failure modes in vision-language models, revealing vulnerabilities in autonomous driving and indoor robotics.

Contribution

It introduces a novel search-based approach combining diversity-aware beam search and Gaussian-process Thompson Sampling to identify failure modes.

Findings

Uncovered previously unreported vulnerabilities in state-of-the-art VLMs.

Identified failure modes such as weak spatial grounding and safety hazard misses.

Provided actionable insights for targeted safety improvements.

Abstract

Vision-Language Models (VLMs) are increasingly used in safety-critical applications because of their broad reasoning capabilities and ability to generalize with minimal task-specific engineering. Despite these advantages, they can exhibit catastrophic failures in specific real-world situations, constituting failure modes. We introduce REVELIO, a framework for systematically uncovering interpretable failure modes in VLMs. We define a failure mode as a composition of interpretable, domain-relevant concepts-such as pedestrian proximity or adverse weather conditions-under which a target VLM consistently behaves incorrectly. Identifying such failures requires searching over an exponentially large discrete combinatorial space. To address this challenge, REVELIO combines two search procedures: a diversity-aware beam search that efficiently maps the failure landscape, and a Gaussian-process Thompson Sampling strategy that enables broader exploration of complex failure modes. We apply REVELIO to autonomous driving and indoor robotics domains, uncovering previously unreported vulnerabilities in state-of-the-art VLMs. In driving environments, the models often demonstrate weak spatial grounding and fail to account for major obstructions, leading to recommendations that would result in simulated crashes. In indoor robotics tasks, VLMs either miss safety hazards or behave excessively conservatively, producing false alarms and reducing operational efficiency. By identifying structured and interpretable failure modes, REVELIO offers actionable insights that can support targeted VLM safety improvements.