Physical Backdoor Attack can Jeopardize Driving with Vision-Large-Language Models
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913323122950144 |
|---|---|
| author | Ni, Zhenyang Ye, Rui Wei, Yuxi Xiang, Zhen Wang, Yanfeng Chen, Siheng |
| author_facet | Ni, Zhenyang Ye, Rui Wei, Yuxi Xiang, Zhen Wang, Yanfeng Chen, Siheng |
| contents | Vision-Large-Language-models(VLMs) have great application prospects in autonomous driving. Despite the ability of VLMs to comprehend and make decisions in complex scenarios, their integration into safety-critical autonomous driving systems poses serious security risks. In this paper, we propose BadVLMDriver, the first backdoor attack against VLMs for autonomous driving that can be launched in practice using physical objects. Unlike existing backdoor attacks against VLMs that rely on digital modifications, BadVLMDriver uses common physical items, such as a red balloon, to induce unsafe actions like sudden acceleration, highlighting a significant real-world threat to autonomous vehicle safety. To execute BadVLMDriver, we develop an automated pipeline utilizing natural language instructions to generate backdoor training samples with embedded malicious behaviors. This approach allows for flexible trigger and behavior selection, enhancing the stealth and practicality of the attack in diverse scenarios. We conduct extensive experiments to evaluate BadVLMDriver for two representative VLMs, five different trigger objects, and two types of malicious backdoor behaviors. BadVLMDriver achieves a 92% attack success rate in inducing a sudden acceleration when coming across a pedestrian holding a red balloon. Thus, BadVLMDriver not only demonstrates a critical security risk but also emphasizes the urgent need for developing robust defense mechanisms to protect against such vulnerabilities in autonomous driving technologies. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_12916 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Physical Backdoor Attack can Jeopardize Driving with Vision-Large-Language Models Ni, Zhenyang Ye, Rui Wei, Yuxi Xiang, Zhen Wang, Yanfeng Chen, Siheng Cryptography and Security Vision-Large-Language-models(VLMs) have great application prospects in autonomous driving. Despite the ability of VLMs to comprehend and make decisions in complex scenarios, their integration into safety-critical autonomous driving systems poses serious security risks. In this paper, we propose BadVLMDriver, the first backdoor attack against VLMs for autonomous driving that can be launched in practice using physical objects. Unlike existing backdoor attacks against VLMs that rely on digital modifications, BadVLMDriver uses common physical items, such as a red balloon, to induce unsafe actions like sudden acceleration, highlighting a significant real-world threat to autonomous vehicle safety. To execute BadVLMDriver, we develop an automated pipeline utilizing natural language instructions to generate backdoor training samples with embedded malicious behaviors. This approach allows for flexible trigger and behavior selection, enhancing the stealth and practicality of the attack in diverse scenarios. We conduct extensive experiments to evaluate BadVLMDriver for two representative VLMs, five different trigger objects, and two types of malicious backdoor behaviors. BadVLMDriver achieves a 92% attack success rate in inducing a sudden acceleration when coming across a pedestrian holding a red balloon. Thus, BadVLMDriver not only demonstrates a critical security risk but also emphasizes the urgent need for developing robust defense mechanisms to protect against such vulnerabilities in autonomous driving technologies. |
| title | Physical Backdoor Attack can Jeopardize Driving with Vision-Large-Language Models |
| topic | Cryptography and Security |
| url | https://arxiv.org/abs/2404.12916 |