CAR Researcher Creates Traffic Model as a Digital Twin of Woody Hayes Drive

As part of an ongoing project to develop a digital model of The Ohio State University, known as a “digital twin,” Center for Automotive Research (CAR) researcher Punit Tulpule is creating a digital replica of one specific street: Woody Hayes Drive.  This digital twin project could have a wide variety of uses, from calculating the power savings of electric-powered CABS buses to determining the best locations for recycling cans on campus.

The term “digital twin” is ultimately just a word for a very fancy computer model of an object or system, with some add-ons.  Where a model is a system of equations designed to simulate a particular system, a digital twin also contains data on parts of the system that can’t be modeled, and can sometimes be considered to include whatever infrastructure is needed to run the model.  While the term isn’t particularly well-defined, it’s generally understood to be a digital replica of an entire system, from a model of what the system is and what it’s made of, to models of how other things will interact with the system.

This is useful because it allows complicated or costly systems to be tested digitally at a fraction of the cost it would take to actually build and implement a physical prototype.  It’s also a way to test something that’s potentially dangerous without any real risk of harm.

“The idea is that as things have gotten more complicated…* we can’t afford to do experimentation in the way we’ve done in the past,” explained Shawn Midlam-Mohler, the director of Ohio State’s Simulation Innovation and Modeling Center.  “You’re doing that (creating digital twins) to avoid what could be expensive, could be time consuming, could be a safety risk.”

A digital twin of the entirety of Ohio State could be used to test all sorts of projects and ideas, from the energy consumption of buildings to the effectiveness of autonomous shuttles on campus to the psychology of whether students would even want to use such shuttles.  However, building such a large model would be expensive and extremely time-consuming, even if it would save a large amount of money and time in the future.  This is why Tulpule is testing the idea of a digital twin by creating a digital twin of the traffic on a single street on Ohio State’s campus.

The street being modeled is Woody Hayes Drive, which Tulpule explained was chosen because it has a large amount of CABS traffic (an important part of building the model), it contains many intersections, and it connects West campus and East campus, allowing different parts of campus to be modeled.

To start building the model, Tulpule had to model the physical roads.  This was a very easy step, accomplished by downloading the maps from Google Maps and adding lane splits and the like.  The next step, adding the traffic, was accomplished using the GPS tracking data from the CABS buses, which is why the high volume of CABS traffic on Woody Hayes Drive was so important.  The other vehicles, which don’t have GPS data, can be filled in using either data on the ratio of CABS buses to other cars, or by using the CABS GPS data to measure how long it takes for a bus to travel from one intersection to the next, and using that information to infer the number of other vehicles.

While Ohio State does have cameras at intersections which could be used to better model the traffic at the intersections, the cameras produce so much data that even transferring it all would be a monumental task, much less storing it all, making using it for the digital twin impossible.

Once the model is complete, it will be time to start using it to simulate different “what-if” scenarios, which is a stage Tulpule hopes will be reached very soon.

“That’s the main target, we want to get to the what-if scenarios quickly,” Tulpule said.

The first scenario to be tested will be what the energy savings would be if all CABS buses were replaced with electric vehicles.  A basic model for this has already been built and is currently being calibrated.

And this model is only the first step in building a campus-wide digital twin.  This larger digital twin would simulate all sorts of parts of Ohio State, from vehicle traffic, to foot traffic, to energy use, to even student psychology and even more.  As such, it requires collaboration with researchers from a wide variety of fields.

“It’s going to be a collaboration with the psychology department, civil engineering, mechanical engineering, and obviously electrical engineering, and maybe we’ll incorporate the computer science people,” Tulpule said.

These other departments all have various models already built but putting them all together and adding any missing ones will take a lot of time and effort.

Another important part of the digital twin project would be being able to select models to a specific level of fidelity, Tulpule explained.  It’s impossible to build a model that will predict everything 100% accurately, so being able to pick a model that gives results to a desired level of specificity.

For example, when modeling traffic, such as the current Woody-Hayes model, it isn’t necessary to model individual cars.  Instead, it’s better to use something like a fluid flow model since the vehicles follow a similar pattern of movement.  However, when modeling a car crash, being able to simulate not just individual vehicles, but specific parts of vehicles becomes important.  In this way, being able to select a model of desired fidelity from a database of models would be an extremely useful part of the digital twin project.

A complete digital twin would be an invaluable resource for planning new projects for the university, and Tulpule’s research on traffic modeling makes for a very promising start to it.

Written by Laura Smith, CAR writing intern