By Robert L. Reid


At Columbia University, a three-year project — Hybrid Twins for Urban Transportation: From Intersections to Citywide Management — that began in 2021 is working to create a digital twin of key intersections and other locations in New York City.


The project will determine how sensing data and machine learning can help optimize traffic flows, notes Xuan Sharon Di, Ph.D., an associate professor of civil engineering and engineering mechanics at Columbia’s Fu Foundation School of Engineering and Applied Science. The university’s Data Science Institute is also involved the project.


Digital twins can help traffic managers and urban planners simulate traffic and congestion conditions and test mitigation strategies, Di explains. Ultimately, the digital twin should function as a hybrid of machine learning and traffic modeling.


The goal is to predict traffic conditions, accidents, and congestion points to optimize traffic flow to reduce stops at intersections and even reduce emissions. Ultimately, “we plan to implement adaptive traffic signal control, learned from real-time traffic data collected from Internet of Things sensors, to coordinate traffic lights along a corridor and a subregion to move traffic more efficiently with fewer stops,” Di says.


Di is collaborating on the project with Qiang Du, Ph.D., a professor of applied mathematics, and Zoran Kostic, Ph.D., and Gil Zussman, Ph.D., both professors of electrical engineering at Columbia. The project is being funded by a $1.2 million grant from the National Science Foundation and the U.S. Department of Transportation’s Federal Highway Administration.


Columbia is developing the digital twin via open-source traffic simulation software that integrate systems that are known as SUMO, CARLA, and NS-3.


SUMO, which stands for Simulation of Urban Mobility, is used to model intermodal traffic systems, including road vehicles, public transport, and pedestrians. CARLA, which stands for Car Learning to Act, is designed to support development, training, and validation of autonomous driving systems. NS-3 is an open, extensible network simulation platform for networking research and education that provides a simulation engine for users to conduct simulation experiments and models of how packet data networks work and perform.


The researchers are using real-time traffic data collected by the smart cameras and communication nodes of the COSMOS “beyond-5G” test bed in New York City, which was designed as a proving ground for wireless technologies as part of the Federal Communications Commission’s Innovation Zones program.


The COSMOS test bed will eventually cover a 1 sq mi area in West Harlem, bordering Columbia’s Morningside Heights campus.


Columbia is one of several universities involved in the COSMOS test bed, along with New York City and Silicon Harlem, a New York City-based technology incubator.


The COSMOS test bed team announced in an Aug. 8 press release that an additional $100,000 grant from the NSF and the FHWA is extending Columbia’s digital twin project by one year and expanding its scope to include researchers from two Indian universities: the Indian Institute of Technology Roorkee and the Indian Statistical Institute in Kolkata.


The expanded project will work to develop a networked traffic signal control system using IoT sensors and other smart technologies. That project, Networked Adaptive Traffic Signal Control in IoT-Enabled Smart Cities, will develop a digital twin of urban traffic systems from data collected by the COSMOS system, calibrated with traffic data provided by an Indian test bed in Mohali, Punjab.


Digital details


The original New York City portion of the project is focusing on a series of intersections near the Morningside campus covered by the COSMOS cameras. Using the data collected, Columbia’s researchers are working to develop an adaptive model that functions like a neural network, which will collect data at certain intervals to adjust its predictions based on the real-time information, Di says.


Public safety is also a key concern. For example, the system could help researchers develop a smartphone app to address any potential pedestrian blind spots at an intersection. “The app might beep or otherwise remind (the pedestrian) to look around,” Di explains.


Columbia and other universities also recently received a $26 million NSF grant to develop an engineering research center for smart streetscapes. Di hopes that the streetscapes center can build on results obtained in the digital twin project.


“Understanding complex streetscapes in real time requires progress in fundamental engineering knowledge and enables exciting opportunities for deploying public interest technology,” explains an Aug. 10 press release from Columbia about the streetscape center project. “A smart streetscape of the future can instantly sense human behavior and guide disabled pedestrians, collect refuse, control pests, amplify emergency services, and protect people against environmental and health threats. It can address unmet needs in road and public safety, traffic efficiency, assistive technologies, outdoor work, and hyper-local environmental sensing.”


Twinning technologies


Other civil engineering projects involving digital twins include a computer model being created of the Houston water system and a sensor-based, real-time decision-support system used by the city of South Bend, Indiana, to better understand the hydraulic conditions in its sewer system.


Civil engineers and others have also formed a new organization, the Coalition for Smarter Infrastructure Investments, to promote greater use of digital technology in infrastructure projects.