Digital Optical Theremin; Ethics of Autonomous Vehicles: An Analysis of the “Unavoidable Crash” Scenario

When interning at Uber ATG, I was astonished at how much progress had been made in the autonomous vehicle space. Because of this experience, I wanted to do my computer science capstone in that field along with my STS research paper. My technological capstone focused on collecting data and analyzing auto-agents in simulation while my STS research focused on ethical issues regarding self-driving vehicles.
The technical portion of my project produced a pipeline for simulating randomized scenarios and routes using the CARLA simulator. The goal of the capstone project was to create a system that allows a user to run randomized tests in a simulator on any agent they provide. This agent can be an AI or an intelligent model trained with machine learning. CARLA was used as the simulation engine which ran off Unreal Engine. This system takes in a route and randomly selects points along that path and places certain key traffic scenarios. These scenarios include bystanders, bad road conditions, and red lights. By using a variety of scenarios, thousands of tests can be generated and test on just one route. This system sets the foundation for simulating random scenarios and will allow future researchers an easier way to test out features of agents.
In my STS research, I focused on analyzing the “unavoidable crash” scenario that is common in ethical discussions involving self-driving vehicles. Through my research I explored several topics on the ethical side including the applicability of the trolley problem and effects of personality on decision making during critical moments in driving. To bridge the gap between theoretical discussions and real-life corporations, I analyzed the safety and risk standards of AVs to see how companies and organizations alike tackle these ethical issues. Overall, the research bridged the gap between these two views and show the transition of ethical ideas into real life implementations of self-driving vehicles. This study will pose as an incentive for further studies in this field, with an emphasis on conducting studies with up-to-date surveys and with a sample that is exposed to new development in AVs.
Overall, through my STS research, I realized that there is a heavy need for ethical considerations with developing safety frameworks and risk assessments for self-driving vehicles. Especially in times where nothing is set in stone, external organizations will try and create a set standard and if such organizations are built by private companies instead of a governmental regulating body many issues can arise. Through my technical work, I learned that simulation of data is important and is needed to properly train and improve self-driving agents. Both of these research projects prove that there is much work to be explored in the field.

School of Engineering and Applied Science Bachelor of Science in Electrical Engineering.
Technical Advisor: Harry Powell
STS Advisor: Richard Jacques
Technical Team Members:
Joseph Chen
Woohyeong Cho
Alexander Kim