HPC4mfg – Reinforcement Learning-based Traffic Control to Optimize Energy Usage and Throughput

Project Description

We propose to develop and test algorithms to adjust traffic signal control based on traffic flow, vehicle mix, and other variables in order to optimize the flow of traffic for energy efficiency and throughput goals. Due to the scale of the optimization problem, we will be using HPC simulations based on real-world data from equipped intersections, traffic simulations, and other sources. For the real-world intersection data, we propose to partner with GRIDSMART Technologies, whose unique fish-eye “bell” cameras instrument intersections, and gather traffic data using computer-vision based detection algorithms.