Deploying a Model Predictive Traffic Signal Control Algorithm - A Field Deployment Experiment Case Study

Oct 8, 2022·
Qichao Wang
,
Joseph Severino
,
Harry Sorensen
,
Jibonananda Sanyal
,
Juliette Ugirumurera
,
Chieh (Ross) Wang
Andreas Berres
Andreas Berres
,
Wesley Jones
,
Airton Kohls
,
Rajesh Paleti Ravi VenkataDurga
· 1 min read
Abstract
This paper presents a field deployment experiment of a real-time traffic signal control algorithm. We implemented the model predictive control (MPC) algorithm based on the virtual phase-link (VPL) model. We selected the deployment locations and times based on an energy saving potential concept. We developed a set of experiment systems, which included sensing, processing, and actuating components, to enable field deployment. We tested the systems rigorously before the experiment days. We reported the key procedures on the experiment days, including the steps taken, the real-time control procedure, and the monitoring of the experiment. We evaluated the impact of the deployment by looking at the changes in delay and energy consumption.
Type
Publication
In 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC)

A vertical flow diagram showing the progression from sensing to processing to actuation.
Architecture diagram illustrating the workflow for real-time traffic signal control for our field experiments. GridSmart traffic data and TomTom speeds were preprocessed to feed into the predictive model. The model produced optimal splits (time spent for each traffic direction) for all traffic signals, and our system automatically actuated the signal controllers to implement the new split.