From sim-to-real: learning and deploying traffic improving cruise controllers

Date: 2022/11/04 - 2022/11/04

Academic Seminar: From sim-to-real: learning and deploying traffic improving cruise controllers

Speaker: Eugene Vinitsky, Assistant Professor, New York University Tandon in Civil and Environment Engineering  (USA)

Time: 10:00 - 11:30, November 4, 2022 (Beijing Time)


While progress in multi-agent reinforcement learning (MARL) has been rapid, many MARL benchmarks are missing key challenges of transportation systems: huge agent numbers, decentralization, communication latency, and partial observability. We investigate the challenges of applying RL to generate cooperative, energy-improving controllers for mixed-autonomy traffic: traffic where partially automated vehicles need to operate amidst human drivers. We discuss our work on using policy-gradient algorithms to design state of the art energy-smoothing controllers in both mixed-autonomy settings and conventional MARL benchmarks. We then present the results of a first deployment of our controllers on 11 vehicles run in summer 2021 in Nashville, TN on the I-24, leading to a follow-on first field operational test in which we will be running our policies simultaneously on 100 vehicles controlling the entire freeway traffic flow. Finally, we will discuss future benchmarks and algorithms for enabling the design of the next-generation of mixed-autonomy controllers.


Eugene Vinitsky is an incoming Professor at NYU Tandon in Civil and Urban Engineering and a current research scientist at Apple in their special projects group. He received his PhD in controls and optimization at UC Berkeley in Mechanical Engineering. Prior to that, he received his MS in physics from UC Santa Barbara and a BS in physics from Caltech. At UC Berkeley, he focused on scaling multi-agent reinforcement learning to tackle the challenges associated with transportation system optimization. As a member of the CIRCLES consortium, he is responsible for the reinforcement learning algorithms and simulators used to train and deploy energy-smoothing cruise controllers onto Tennessee highways. In the past he has spent time at Tesla Autopilot, DeepMind and was a visiting researcher at Facebook AI. His research has been published at ML venues such as CORL, neurIPS, and ICRA and at transportation venues such as ITSC. He is the recipient of an NSF Graduate Student Research Award, a two time recipient of the Dwight David Eisenhower Transportation Fellowship, and received an ITS Outstanding Graduate Student award.