Our research group uses sensing technologies, models, data, and advanced analytics to understand mobility systems so that we can improve them. Our research projects explore the interplay between autonomous vehicles and phantom traffic jams. We also use large mobility datasets to understand urban traffic congestion at city scales, and freight rail traffic at the regional scale. We build mathematical models and systems theory tools to understand the underlying behavior of traffic flow.
Our research group is featured in Vanderbilt's Quantum Potential Series, hosted by Jad Abumrad (Radiolab, Dolly Parton's America).
See the video above featuring Dr. Jonathan Sprinkle and Dr. Dan Work below. It covers TDOT's I-24 MOTION Testbed, and our collaborative work with the CIRCLES Consortium on running "the world’s largest real-world road test on a stretch of I-24 in Nashville to see how semi-autonomous vehicles can ease traffic for more sustainable and less stressful driving."
Field Deployment of Multi-Agent Reinforcement Learning Based Variable Speed Limit Controllers
We deployed the first ever multi-agent reinforcement learning based variable speed limit controllers in a real-world traffic freeway segment. We trained the agents in a microscopic simulation environment and deployed them on a 17-mile stretch of I-24 in Nashville, TN.
Enabling Mixed Autonomy Traffic Control with 100 Connected Automated Vehicles
As summarized in the video abstract below, we make experimental connectivity and automation in vehicles scalable to implement and deploy a team of 100 Connected Automated Vehicles (CAVs).
This large CAV team enables, for the first time, the capability of freeway traffic control with the CAVs as mobile distributed sensor/actuator network. This is part of collaborative work with the CIRCLES Consortium.
This research, led by PhD candidate Matt Nice, introduces a new V2X application in Connected and Automated Vehicles: SAILing CAVs. We create a vehicle which can automatically match the infrastructure-based variable speed limits as they dynamically change on the interstate.
Our research shows how a small number of automated vehicles can help smooth traffic jams.
Now we're building I-24 MOTION: 6 miles of dense instrumentation on I-24 in Nashville, TN, to study traffic dynamics and the control of traffic using automated vehicles.
The I-24 MOTION project is being conducted along with the Tennessee Department of Transportation. In 2022, the highway will be instrumented with 43 roadside camera poles holding nearly 300 4K-resolution cameras. These video feeds will be processed into anonymized vehicle trajectories that capture the time and position of individual vehicles continuously across the roadway. This anonymous vehicle data will unlock new research and better understandings of how we drive and interract with autonomated vehicles. I-24 MOTION will also be an experimental environment for automated vehicles, traffic control, and congestion mitigation, since it gives us an unparalleled ability to understand the effects on the entire roadway.
In collaboration with the Southwest Research Institute, we are developing an AI-based decision support tool that will be used for corridor management on I-24.
