AI road asset assessment pilot aids Henrico County Public Works

A recent artificial intelligence pilot project helped the Henrico County Department of Public Works assess and manage its infrastructure.

Henrico worked with Infra.ai, an infrastructure intelligence company that helps public-sector asset owners create a comprehensive digital baseline of their transportation, utility, and municipal assets through the use of a 360-degree camera mounted to the vehicle.

The pilot project created a digital baseline of municipal assets. It evaluated streetlights, signage and striping on pavement, identifying and geolocating assets within a designated area, and then gave them a condition score and rated the health of the entire portfolio.

“We wanted to pick an area in western Henrico around Deep Run High School that had a number of roundabouts because we wanted an area that had sidewalk, curbs, gutters, ditches, bike lanes and four-lane roads,” said Terrell Hughes, Henrico County director of Public Works.

Infra.ai uses high-resolution imagery, structured data capture, and integration with geographic information systems and computerized maintenance management systems to enable agencies to inventory assets, assess condition, identify risk, and prioritize investment — without adding administrative burden to existing teams, according to the company’s website. The assessment was meant to improve operational visibility, strengthen capital planning and stretch public dollars further.

Henrico and Arlington are the only two Virginia counties responsible for maintaining their own secondary road systems. Henrico’s road network consists of about 3,600 lane miles, with about 1,400 center line miles, 170 traffic signals and an increasing number of pedestrian enhancements.

“We're very much like a city in that sense, where we take care of our road assets,” said Hughes. “With that comes a lot of signs, drainage infrastructure, curbs, gutters, ditches, vegetation – you name it.”

Hughes said that using AI is an efficient way to confront the rising costs of road management.

“We've seen increased infrastructure costs in terms of paving,” Hughes said. “It puts a strain on your budget. You're constantly trying to find new ways to be more efficient with your funding.”

Hughes said the AI pilot could improve customer service by helping Henrico identify problems such as potholes and damaged street signs. He acknowledged that there are areas that are more likely to get service because residents are more likely to request service.

“One of the things that we're trying to tackle on a customer service or an equity front [is] being more proactive, identifying where our assets need improvement,” Hughes said.

Emerging concepts in transportation

Hughes discussed the pilot project during a panel on public-sector technology innovations at the Virginia Governor's Transportation Conference, held Oct. 13–15 in Richmond, alongside Infra.ai CEO Brandon Kelley.

Other panelists at the session included staffers from the Virginia Department of Motor Vehicles and a Virginia Department of Transportation Bicycle and Pedestrian Planner with VDOT’s Transportation and Mobility Division.

The Virginia DMV has been testing an Automated Road Test System (ARTS), developed in partnership with Sapios, that uses smartphone technology, artificial intelligence, computer vision and eye tracking to evaluate driving performance. The system is designed to provide more objective assessments without requiring an examiner inside the vehicle.

The VDOT staffer presented a new statewide dataset created using a Geo AI tool that mapped approximately 27,000 miles of sidewalks and nearly 69,000 crosswalks across Virginia to support multimodal planning and safety initiatives.

The yearly conference brought together about 500 people in government, along with people in the transportation industry to discuss pressing issues with subject matter experts focusing on transportation in Virginia.

“VDOT controls over 100,000 [lane] miles of roadway; The ability to do these surveys and understand all of our assets and where they go missing, or what has been impacted, using AI gives us the ability to do it on a more frequent scale, to then start to understand how we can better deploy our resources to take care of our system and assets,” Virginia Secretary of Transportation Nicholas Donohue said at the conference. “When we must rely on full human field surveys, it’s very difficult to do, or cost prohibitive.”

Donohue praised Henrico Public Works for dipping a toe into AI to address road and asset maintenance in the growing county.

Other sessions at the conference focused on data, airports, public transit and autonomous vehicles. Virginia has not established a comprehensive regulatory framework for fully autonomous commercial vehicles, an issue lawmakers are expected to revisit during the upcoming General Assembly session.

Donohue said the purpose of the conference was to get feedback from stakeholders and push some of those emerging concepts. His overall goal is to decrease road fatalities.

“We had a downward trend for a long time, and for the last 10 years, they've been increasing. In 2014, there were 703 highway fatalities. But for the last five years, Virginia has averaged 925 road fatalities,” Donohue said.

“Virginia is moving in the wrong direction there. And I would like, by the time that this administration concludes, that we have reestablished a downward trend on that. That's going to require partnership, collaboration and engagement.”

Funding is another challenge.

VDOT and the Commonwealth Transportation Board are facing a maintenance-funding gap driven primarily by construction-cost inflation and a multi-year decline in planned transfers from construction funding to VDOT’s Highway Maintenance and Operating Fund.

Geospatial technology

Kelley said Infra.ai's geospatial technology, M4AI, creates a digital baseline of every asset that a municipality or a county may own, from streetlights to signage to striping on pavement. The technology identifies and geolocates it, and then gives municipalities a condition score along with a report on what health its entire portfolio is in.

“We believe that access to better data can improve how public dollars are spent,” Kelley said in a company video focusing on the Henrico County project and a project for Richmond’s GRTC bus system he showed at the conference panel. “There is a credible amount of efficiency that comes with a digital baseline. They get situational awareness as to where a problem may be without having to go out into the field, they can coordinate with other departments. They can determine how big of a team do I have the right players on my team.

“Now we move from guessing to planning, and we believe that the benefits of those decisions will be felt for decades.”

GRTC also partnered with Infra.ai to conduct an inventory of hundreds of bus stops and rider amenities throughout the region to assess needed repairs and plan for transit infrastructure.

Hughes said there was still a human component to Infra.ai pilot. Henrico County staffers performed quality control to verify asset locations, ensure items were identified correctly and evaluate their conditions.

With the assessment, Henrico Public Works then could quickly deploy staff to fix the issue, often addressing issues and assets that tend to go unnoticed because they're not necessarily urgent in nature.

“It could be a clogged driveway pipe,” Hughes said. “Those are some things that we were able to identify like ditches, vegetation or a cracked curb. It was nice getting some visuals on some things that typically might not be something that is an emergency.

“In the mind of stewardship, this is something that we've been looking for, getting more from an asset standpoint, being able to leverage technology, and get a better visual on what we have in terms of an asset, and then also be able to plan, budget, and then do more essentially with less. The efficient collection process was one of the other pleasant surprises of the pilot.”

Going forward, some assets on roads are still going to always need a human evaluator, like guardrails, where measurements and a visual inspection is needed by someone who's certified.

“But with AI we could keep an eye on faded crosswalks or bike lanes that need new symbols. The goal here is knowing more granularly what we'll be working on,” Hughes said.

“Instead of having those numbers that have to adjust and shift, we're able to have a better idea of what those numbers need to be. So we'll have a little bit more awareness on how we need to slice the pie in a given year.”

Dina Weinstein is the Citizen’s community vitality reporter and a Report for America corps member, covering housing, health and transportation. Support her work and articles like this one by making a contribution to the Citizen.