敵対的蒸留による言語条件付きロボットモデルへの攻撃:中間特徴の再考
Rethinking the Intermediate Features in Adversarial Attacks: Misleading Robotic Models via Adversarial Distillation
言語条件付きロボットモデルに対し、任意のプロンプトに付加する普遍的な敵対的プレフィックスを生成し、意図しない動作を引き起こす攻撃手法を提案。連続行動表現と中間自己注意特徴の負勾配を活用して攻撃効果を高めた。
著者: Ke Zhao, Huayang Huang, Miao Li, Yu Wu
分類: cs.LG, cs.AI, cs.RO
原文アブストラクト
Language-conditioned robotic learning has significantly enhanced robot adaptability by enabling a single model to execute diverse tasks in response to verbal commands. Despite these advancements, security vulnerabilities within this domain remain largely unexplored. This paper addresses this gap by proposing a novel adversarial prompt attack tailored to language-conditioned robotic models. Our approach involves crafting a universal adversarial prefix that induces the model to perform unintended actions when added to any original prompt. We demonstrate that existing adversarial techniques exhibit limited effectiveness when directly transferred to the robotic domain due to the inherent robustness of discretized robotic action spaces. To overcome this challenge, we propose to optimize adversarial prefixes based on continuous action representations, circumventing the discretization process. Additionally, we identify the beneficial impact of intermediate features on adversarial attacks and leverage the negative gradient of intermediate self-attention features to further enhance attack efficacy. Extensive experiments on VIMA models across 13 robot manipulation tasks validate the superiority of our method over existing approaches and demonstrate its transferability across different model variants.
