paper-with-me

홈 › Papers

Predictive Red Teaming: Breaking Policies Without Breaking Robots

2025-02-10 · Anirudha Majumdar, Mohit Sharma, Dmitry Kalashnikov, Sumeet Singh, Pierre Sermanet, Vikas Sindhwani

Visuomotor policies trained via imitation learning are capable of performing challenging manipulation tasks, but are often extremely brittle to lighting, visual distractors, and object locations. These vulnerabilities can depend unpredictably on the specifics of training, and are challenging to expose without time-consuming and expensive hardware evaluations. We propose the problem of predictive red teaming: discovering vulnerabilities of a policy with respect to environmental factors, and predicting the corresponding performance degradation without hardware evaluations in off-nominal scenarios. In order to achieve this, we develop RoboART: an automated red teaming (ART) pipeline that (1) modifies nominal observations using generative image editing to vary different environmental factors, and (2) predicts performance under each variation using a policy-specific anomaly detector executed on edited observations. Experiments across 500+ hardware trials in twelve off-nominal conditions for visuomotor diffusion policies demonstrate that RoboART predicts performance degradation with high accuracy (less than 0.19 average difference between predicted and real success rates). We also demonstrate how predictive red teaming enables targeted data collection: fine-tuning with data collected under conditions predicted to be adverse boosts baseline performance by 2-7x.

📄 PDF Abstract BibTeX arXiv:2502.06575

Code (0)

등록된 구현이 없습니다.

Tasks

Imitation LearningRed Teaming

Methods 이 논문이 사용한 방법론

Diffusion Diffusion models generate samples by gradually removing noise from a signal, and their training objective can be expressed as a reweighted variational lower-bound…

Similar Papers 제목 키워드 기반

JBFuzz: Jailbreaking LLMs Efficiently and Effectively Using Fuzzing

2025-03-12 · Vasudev Gohil

Large language models (LLMs) have shown great promise as language understanding and decision making tools, and they have permeated various aspects of our everyday life. However, their widespread availability also comes w…

Red TeamingSafety Alignment

Red Teaming Visual Language Models

2024-01-23 · Mukai Li, Lei LI, Yuwei Yin, Masood Ahmed 외

VLMs (Vision-Language Models) extend the capabilities of LLMs (Large Language Models) to accept multimodal inputs. Since it has been verified that LLMs can be induced to generate harmful or inaccurate content through spe…

FairnessRed Teaming

Improved Techniques for Optimization-Based Jailbreaking on Large Language Models

2024-05-31 · Xiaojun Jia, Tianyu Pang, Chao Du, Yihao Huang 외

Large language models (LLMs) are being rapidly developed, and a key component of their widespread deployment is their safety-related alignment. Many red-teaming efforts aim to jailbreak LLMs, where among these efforts, t…

Red Teaming

Jailbreaking LLMs via Calibration

2026-01-31 · Yuxuan Lu, Yongkang Guo, Yuqing Kong arxiv

Safety alignment in Large Language Models (LLMs) often creates a systematic discrepancy between a model's aligned output and the underlying pre-aligned data distribution. We propose a framework in which the effect of saf…

AdvPrompter: Fast Adaptive Adversarial Prompting for LLMs

2024-04-21 · Anselm Paulus, Arman Zharmagambetov, Chuan Guo, Brandon Amos 외

Large Language Models (LLMs) are vulnerable to jailbreaking attacks that lead to generation of inappropriate or harmful content. Manual red-teaming requires a time-consuming search for adversarial prompts, whereas automa…

MMLURed Teaming