IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Little Vermont Uses Big Data to Predict Bridge Repairs and Traffic Jams

The state is exploring how AI and neural networks can help them forecast when vital infrastructure repairs are needed, years in advance.

Vermont may be a land of bucolic country roads. It is also a land of big data. State transportation engineers are using artificial intelligence, predictive analytics and even wonkier-sounding neural networks to better understand how roads and bridges might be deteriorating and in need of maintenance.
“It’s going to be key that we use that data to understand, and predict, everything from traffic congestion, to road conditions, to where we need signals,” said John Quinn III, chief information officer for Vermont. “Those types of things will all come from data, I believe.”
Quinn is also secretary for the state Agency of Digital Services, formed in 2017, and brings IT operations from various departments under one banner with the goal of modernizing the state’s IT presence. About a year ago the agency began several pilot projects that more effectively harness the many data sets collected by traffic engineers for inspecting the condition of roads and bridges.
“The engineering group really loves it because there’s no way they could look through this kind of data to come up with this kind of analysis,” said Quinn.
Using predictive analytics and neural networks, which are mathematical models and are common in AI, as a way to organize and analyze vast quantities of data, the state has been able to take a more nuanced look into a bridge’s future one year, five years or even 10 years from now.
“We’ve been taking data from systems and then trying to make predictions, in a manual way, up until the last year or so. And then in the last year, we’ve really harnessed the big data to work for us,” said Quinn.
The use of big data by the Center for Digital Services will prepare the state for the kinds of next-generation transportation technologies such as autonomous driving.
“We’re really trying to let the employees drive the innovation here, and how we use these things,” said Quinn.
One project, known as the Computer Vision Asset Inventory, uses data to review state-owned roadway signs to identify the best locations for signage to be caught in the camera stream of autonomous vehicles that will be on the roads in the near future.

Vermont’s expanded use of road, bridge and traffic data is not unlike the data harvesting by transportation officials in other states, including Florida and Iowa. All of these projects rely on big data, which can be fed into computers, analyzed and provide state transportation officials with answers to an array of challenges as well as the means to predict trouble, such as maintenance issues and traffic congestion, before it happens.
Vermont’s big data transportation AI initiative is an in-house project, while Iowa is collaborating with engineers from Iowa State University. But the private sector is also stepping up to provide transportation agencies with the analytics they require. StreetLight Data is a transportation analytics firm that uses data collected from mobile phones and connected vehicles to analyze traffic and transit patterns.
The company recently introduced a new tool that tells traffic planners, “not just what is happening, but why it’s happening,” said Laura Schewel, CEO of StreetLight Data. “A lot of tools can tell you where a traffic jam happens. Our tool can tell you why the traffic jam is happening, and what the best solution is,” she explained.
The data collected and analyzed by StreetLight is intended to be used at all levels of traffic and transit planning.
“I think of the tool as telling a story. And the story has three parts. In the first part of the story, you find where congestion is,” said Schewel. “The next step allows you to analyze the traffic of every road segment in your community, so then you understand the causes of congestion. You’ve diagnosed the problem. And then in the third step, you look at the solutions, and rank which solutions go for which places.
“Data is expensive, maybe,” Schewel offered. “But infrastructure is really expensive. So a little investment in data analysis at the beginning will allow you to deploy those infrastructure dollars more wisely.”
Skip Descant writes about smart cities, the Internet of Things, transportation and other areas. He spent more than 12 years reporting for daily newspapers in Mississippi, Arkansas, Louisiana and California. He lives in downtown Yreka, Calif.