RoboJailBench

RoboJailBench is a benchmark for evaluating adversarial attacks and defenses in embodied robotic agents.

• Project website: https://purseclab.github.io/benchmark-for-robotics-security/
• arXiv preprint: https://arxiv.org/abs/2605.19328
• GitHub repository: https://github.com/purseclab/benchmark-for-robotics-security
• Croissant metadata: https://purseclab.github.io/benchmark-for-robotics-security/croissant.json

Authors

Doguhan Yeke, Yanming Zhou, Leo Y. Lin, Hongyu Cai, Antonio Bianchi, and Z. Berkay Celik. All authors are affiliated with Purdue University.

Citation

Current preprint citation:

@misc{yeke2026robojailbenchbenchmarkingadversarialattacks,
      title={RoboJailBench: Benchmarking Adversarial Attacks and Defenses in Embodied Robotic Agents}, 
      author={Doguhuan Yeke and Yanming Zhou and Leo Y. Lin and Hongyu Cai and Antonio Bianchi and Z. Berkay Celik},
      year={2026},
      eprint={2605.19328},
      archivePrefix={arXiv},
      primaryClass={cs.CR},
      url={https://arxiv.org/abs/2605.19328}, 
}