Two of those cities, Columbus and Warner Robins, Georgia, received the awards for projects that involve digital twins. But what, exactly, is a digital twin? And how can the technology be used to solve community problems?
We talked with John Taylor, the Frederick Law Olmsted Professor and associate chair for graduate programs and research innovation in the School of Civil and Environmental Engineering, and Neda Mohammadi, city infrastructure analytics director in Georgia Tech’s Network Dynamics Lab to get some answers. These are edited highlights from an interview.
Q: What is a digital twin?
Taylor: A digital twin is an intelligent, adaptive system that pairs virtual and physical worlds. In community development work, a Smart City Digital Twin (SCDT), like those used in Warner Robins and Columbus, pairs a real city to its digital counterpart to generate data-driven feedback loops of interactions between cities’ three main components: (1) human systems, which includes government, industry, and residents; (2) infrastructure systems, which are physical systems and the services they provide; and (3) technology systems, such as devices, sensors, and data analytics infrastructure.
Q: They’ve been used in manufacturing for some time. How is that different from a SCDT?
Taylor: They're somewhat easier to implement in a manufacturing context, because everything's under control, under a roof. They model all the different manufacturing machinery and they use that to see when a part might need to be changed, and when they need to do maintenance. And they can play with the system, using real-time running data to see what happens if this piece does wear out. How bad would it be? They could either adjust that piece or adjust that machine or maintain it, whatever it might be, based on the scenario analysis.
Q: How does that translate to the less controlled environment of a city?
Taylor: It involves replicating multiple systems. For example, if a tall building is on fire, there will be multiple systems brought into play. First, you can see what's happening in the city at a basic level. You can see that there's traffic building up, for example. The next level is, why is it happening? And that's where it gets a little bit more interesting. Most of the digital twin work that we've seen — that anyone's doing out in the world — is to understand why things are happening the way they're happening. But really, the value starts to unlock the third and fourth levels.
The third level is the “what if” scenario. In the context of a city, for example, in Midtown they've just installed new traffic signals. Hopefully, someone tested that out in advance. But one “what if” analysis could be: We've got bad traffic in Midtown. What if we put these traffic signals in the Tech Square area? What effect will that have on the flows in the city? With a digital twin, you can know that before you install the lights. That is one of the big opportunities.
The fourth level is the idea that the infrastructure could start to intervene on behalf of the citizens. And so in the example of the tall building fire, the traffic signals might preemptively allow the fire trucks through. But they could also do other things like make all of the signals around the building red, so no traffic is moving and there's more space for people to evacuate the building. That would be something we might allow the systems to do for us.
Q: How is that different from, for example, a project in Valdosta that allows first responder vehicles to change the traffic lights so they can get to an emergency more quickly?
Mohammadi: A digital twin will update itself based on data that keeps coming in. If you think about the interaction with the traffic signal, it doesn't care about what happened five minutes ago, 10 minutes ago. At that moment, they know that the driver probably has a better situational awareness than the automated system. So they let the driver interfere and put useful inputs into the systems to make a better decision.
The digital twin is accumulating data as it comes because it is based on prediction. The definition of prediction is looking at past data and, based on past experience, predicting what's likely to happen in the future. We know that time is a moving target. As we move on, things that happened in the past accumulate. There are more things that we know. A digital twin is really at the edge of this moving target.
Q: Tell us about the river safety project in Columbus, which uses a digital twin to create an alert system to prevent drownings in the Chattahoochee River. The city was recently named a Smart 20 award winner by Smart Cities Connect for the Citizen Safety Digital Twin project.
Taylor: A good project from our perspective involves a complicated enough scenario where multiple sensors are involved. With the river safety project, we had to understand and predict water levels with a water level sensor. We use visual sensing to understand, if people were in the environment when hazardous conditions might begin to occur, whether we could get them out of harm's way before they get swept away into the water.
We had to build a digital twin of the entire river basin, so we would know just what the danger is if the water level rises this much. Are the islands that people are standing on before the water level rises going to vanish?
That one was particularly interesting to us. If you look at the smart city digital twin work we did first, it was related to energy consumption. We're increasingly excited about having a more direct effect on people's lives. This one is stopping people from drowning.
Q: Tell us about the digital twin you developed for the Warner Robins’ Citizen Safety Digital Twin for Community Resilience project, which deploys dynamic license plate reader cameras to help deter crime. It received the Intelligent Community Forum’s Smart21 Community Award at the 2024 Taipei Smart City Summit and Expo.
Taylor: This project is pretty complicated from our perspective, because we had to build a geographic information systems (GIS) map of the city. We also have to know where crimes have been occurring. We've got more than 10 years of crime data, including very recent crime data. We're deploying sensors in part to deter crimes, but also to detect and collect more information about crime patterns. It comes down to taking the information about where crimes are occurring and coupling that with predictions about routes people would take if they did commit a crime, so that the car would come into view of one of the cameras. We don't hide the camera; we put it on a very visible structure, where we predict most likely the crimes are going to occur this week. We put this very visible thing to discourage people from doing anything once they realize they're being watched. And we found that it did in fact, reduce crimes in those high-crime spots by 20%.
Q: What are some other ways communities can use digital twins?
Taylor: We published something this spring, and we're working on a funding proposal now, about how ambulances move around during a period of inundation — coastal flooding, coastal inundation, or heavy rains. We’ve met with Charleston, South Carolina, and Savannah about this. We looked at data in Virginia Beach to see if, in real time as the flooding is changing, we could deploy ambulances in different parts of the city ahead of where they're needed. It’s ambulance routing during a natural disaster event.
Q: Are there limitations to smart city digital twin technology?
Taylor: When we travel around and we present this, some clever student or faculty member will say, “Wouldn't a great research project be to figure out how to build a central platform for the collection of this data or a standard format for the way this data should be sent so that all the systems can talk to each other?” And they’re right. It's difficult to get the value across a whole city if you're only looking at one system at a time. A future research topic is figuring out those data flows and the centralization of that data.
