Mixed-Autonomy Traffic at Scale: The MegaVanderTest Self-Driving Vehicle Study

Learn about the groundbreaking MegaVanderTest experiment in this lecture from UC Berkeley professor Alexandre Bayen, who details the largest-ever deployment of collaborative self-driving vehicles on a public freeway. Explore the sophisticated architecture and algorithms developed by the CIRCLES team for this unprecedented test involving 100 autonomous vehicles on Nashville's I24 highway. Discover the implementation of optimal control, imitation learning, deep reinforcement learning, and Model Predictive Control (MPC) techniques used to manage traffic flow at scale. Gain insights into the practical challenges of deploying autonomous vehicle algorithms in real-world conditions, particularly regarding access to onboard sensor data in modern Adaptive Cruise Control systems. The presentation draws from Bayen's extensive experience in aeronautics, astronautics, and autonomous systems, including his work with NASA Ames Research Center and the French Ministry of Defense.