Summary of Less Is More: a Stealthy and Efficient Adversarial Attack Method For Drl-based Autonomous Driving Policies, by Junchao Fan et al.
Less is More: A Stealthy and Efficient Adversarial Attack Method for DRL-based Autonomous Driving Policies
by Junchao Fan, Xuyang Lei, Xiaolin Chang, Jelena Mišić, Vojislav B. Mišić
First submitted to arxiv on: 4 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed novel stealthy and efficient adversarial attack method for DRL-based autonomous driving policies is designed to trigger safety violations (e.g., collisions) by injecting adversarial samples at critical moments. The attack is modeled as a mixed-integer optimization problem and formulated as a Markov decision process, which the adversary learns through training without domain knowledge. To enhance learning capability, attack-related information and trajectory clipping are introduced. Experimental results show that the method achieves more than 90% collision rate within three attacks in most cases, with over 130% improvement in attack efficiency compared to the unlimited attack method. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A team of researchers has developed a new way to test how well autonomous driving systems can handle unexpected problems, like an attacker trying to make the car crash. They created a special kind of “attack” that can trick the system into making mistakes, and then tested it in different scenarios. The results show that this attack is very good at making the system make mistakes, and could help developers create more secure autonomous driving systems. |
Keywords
» Artificial intelligence » Optimization
