The result? A paper entitled: "D2MON: Detecting and Mitigating Real-Time Safety Violations in Autonomous Driving Systems" summarized in this abstract:
This paper proposes D2MON, a data-driven realtime safety monitor, to detect and mitigate safety violations of an autonomous vehicle (AV). The key insight is that traffic situations that lead to AV safety violations fall into patterns and can be identified by learning from existing safety violations. Our approach is to use machine learning techniques to model the traffic behaviors that result in safety violations and detect their symptoms in advance before the actual crashes happen. If D2MON detects surroundings as dangerous, it will take safety actions to mitigate the safety violations so that the AV remains safe in the evolving traffic environment. Our steps are twofold:
- We use software fuzzing and data augmentation techniques to generate efficient safety violation data for training our ML model.
- We deploy the model as a plug-and-play module to the AV software, detecting and mitigating safety violations of the AV in runtime. Our evaluation demonstrates our proposed technique is effective in reducing over 9% of safety violations in an industry-level autonomous driving system, Baidu Apollo.
The paper, co-authored with Li, alumnus Bohan Zhang (BS '19; MCS '22) and PhD advisee Yafan Huang, was just accepted to RSDA 2022 "The 7th IEEE International Workshop on Resiliency, Security, Defenses and Attack."
Chen was also kind-enough to answer a few questions about this project and more:
What – personally or academically – prompted you to join Prof. Li's research team and investigate Autonomous Driving Systems safety ?
During the summer when I was learning how to drive, I used my dad’s car that has a feature that allows some autonomous driving. The mechanics behind this interested me, and I wondered how these features work. During the classroom portion of my driving lessons, we also discussed the issues of what leads to car accidents, and so I thought about how autonomous driving techniques could help humans avoid accidents. I thought I could explore this area through a summer internship. I learned that UIOWA is a popular destination for Illinois students, and I started searching the CS faculty website and found that Dr. Li is active in the research of autonomous driving. I contacted Dr. Li and asked about a possible internship with him.
What was your mathematical and CS background before joining the IOWA-HPC Lab?
I have taken all the math courses up to Calculus BC at my high school. Currently, I am learning Calculus 3 and Statistics. I self-learned a few basics such as the binary number system and computer programming. Math courses, especially statistics, help me better understand a lot of the concepts, data collection, analysis, and interpretation methods used in the research.
How would you describe your experience working with faculty, MCS, and PhD students?
The team has been extremely welcoming, patient, and understanding. As a high school student, I don't understand many things they do. WhenI have any confusion or questions, they would take their time to explain them to me in a way that I can easily understand. For instance, I struggled to understand the sequence to sequence model used in their AV safety monitor. When I asked the team to explain its purpose, they simplified the concept by explaining it as predictions. Along with being accommodating, the team are all passionate about their research work.
Do you see yourself applying lessons learned towards other academic pursuit or research projects (e.g., undergraduate / graduate studies in Computer Science; other internships)?
Definitely; I would like to pursue a career in computer science or engineering where computer science can be applied, and conduct my own research and build safe systems that can help improve people's lives. This opportunity has allowed me to gain exposure to the processes behind research. Further, many of these concepts learned through this internship with Dr. Li can be applied to other fields such as the healthcare industry, as the world is becoming increasingly technology-dependent. In the future, on top of autonomous vehicles, I hope to investigate and apply these techniques in more areas to help increase safety and reliability. For example, we know that computers have been used to diagnose diseases, but I think a much more advanced system can be built to analyze the internal organs of a human body. With more time and research, we could build systems that reliably detect fatal diseases such as cancers in their early stages and even automatically apply treatments.
We hope this was an enriching experience. Any advice for fellow high schoolers / prospective UIowa computer science students?
One piece of advice I have for fellow high schoolers and future UIowa computer science students is to always take initiative and to not shy away from opportunities. For instance, doing a research internship at a higher education place as a high schooler may sound intimidating, but the outcome is rewarding as you get to challenge yourself and experience new things. Hesitation can often lead to regret and not fulfilling your full potential.
"Miss Chen is smart and has great potential to succeed in the area of computer science. She has done a fantastic job in participating the project. Her performance exceeds my initial expectation from a high school student."
"In the future, we will look into:
- deploying and evaluating our proposed technique on real-world commercial autonomous vehicles
- improving our technique by deploying more powerful and advanced machine learning models."
Dr. Guanpeng Li
Undergraduate students interested in signing up for their own Independent Research Study should talk to their faculty of choice (based on past interactions or research interests), submit this Independent Study Contract then head to MyUI to register, when authorized.
Our "REU Computing for Health and Well-being" will offer a final opportunity, in Summer 2023, for undergraduate students studying near and far to join faculty mentor's research group, which consists of other faculty, graduate and sometimes undergraduate students, to work on an interdisciplinary project. Students also participate in a Data Science Bootcamp, career development workshops, social events, and fun excursions. REUs are a full-time commitment that provide a stipend, housing and meals.
While rarer, the department and its faculty researchers have been honored and impressed to work with driven high school students!