Online Archive of University of Virginia Scholarship
Responsibility and Liability in Autonomous Vehicle Accidents11 views
Author
Yoo, Justin, School of Engineering and Applied Science, University of Virginia
Advisors
Murray, Sean, EN-Engineering and Society, University of Virginia
Kuo, Yen-Ling, EN-Comp Science Dept, University of Virginia
Abstract
As algorithms get more advanced, they can take over and optimize human activities such as driving and sports betting. My capstone focuses on how algorithms and different data modalities can affect outcomes and accuracy in sports betting, while my STS project investigates how humans and an algorithmic system can share responsibility in case of an accident.
In my technical capstone, I researched the impact that different datasets have on an algorithm’s accuracy. With the topic of NBA sports betting, I found three data modalities to train my algorithms on. The first was statistical data from box scores of games. The second combined the basic statistical data with spatial movement data from Sport VU. SportVU was a tracking system that the NBA implemented, where cameras all over the courts captured all the positions of the players and the ball at 25 frames per second. The third data was statistical data combined with textual data via a play-by-play commentary of the game. By training the same model on these three different types of data, I was able to analyze the importance of data and how it impacted the results of the model.
While machine learning algorithms and models may produce better results on paper, it can be hard to trust them, as you can’t tell what factors the model considered and what it viewed as important.
This leads to my STS project, where I analyzed how responsibility and liability should be distributed in cases of accidents occurring with autonomous vehicles (AV). As AVs get more prevalent in today’s society, it becomes important to have a fair framework to delegate responsibility when accidents will inevitably occur. To thoroughly examine this issue, I used current law regulations, a sociological framework via Actor Network Theory, and the consequentialist ethical framework. By using these three types of analysis, I was able to get to the root of the problem and not only determine all actors within this complex system of humans and computers, but also who/what is in control of the vehicle. This led to a more deterministic approach in deciding responsibility in AV accidents.
Both of these topics have led me to realize how much computers and algorithms play a role in our lives. While autonomous driving and sports betting seem unrelated, they both heavily depend on computer analysis. This shows how much we currently rely on computers/algorithms and whether we should continue to rely this heavily on them.
Yoo, Justin. Responsibility and Liability in Autonomous Vehicle Accidents. University of Virginia, School of Engineering and Applied Science, BS (Bachelor of Science), 2026-05-11, https://doi.org/10.18130/r49z-yw42.