Awesome Robotics Manipulation · full_paper

Inject Once Survive Later: Backdooring Vision-Language-Action Models to Persist Through Downstream Fine-tuning

作者：Jianyi Zhou, Yujie Wei, Ruichen Zhen, Bo Zhao, Xiaobo Xia, Rui Shao, Xiu Su, Shuo Yang B · 单位：Harbin Institute of Technology, Shenzhen, Meituan Academy of Robotics Shenzhen, Meituan, Shanghai Jiaotong University, National University of Singapore, Central South University · 会议/期刊：arXiv · 日期：2026-01-31 · 来源：Low-Level Learning-Based Action Modelling / Input Modelling / 2D Vision Language Action Models with Robustness / Attack

视觉语言动作安全机器人学习
新标签打开 AlphaXiv新标签打开 Paper新标签打开 Project
Inject Once Survive Later: Backdooring Vision-Language-Action Models to Persist Through Downstream Fine-tuning figure
AlphaXiv 中文论文页面（可滚动查看）