When Attention Betrays: Erasing Backdoor Attacks in Robotic Policies by Reconstructing Visual Tokens

作者：Xuetao Li, Pinhan Fu, Wenke Huang, Nengyuan Pan, Songhua Yang, Kaiyan Zhao, Guancheng Wan, Mengde Li, Jifeng Xuan, Miao Li · 会议/期刊：ICRA 2026 · 日期：2026-02-03 · 来源：Low-Level Learning-Based Action Modelling / Input Modelling / 2D Vision Language Action Models with Robustness / Attack