It won’t have escaped your attention that artificial intelligence is having something of a ‘moment’ and, like every other industry in the world, intelligent transportation is working out how to use it most effectively. So let’s start with a basic question: what are the main problems that AI can help solve in ITS?
“The challenge we’re facing today transcends basic congestion,” says Akif Ekin, CEO & founder at Ekin Smart City Technologies. “Cities are grappling with overwhelmed road networks, rising accident rates, and serious inefficiencies in transport operations. Traffic delays result in economic losses — billions in wasted fuel and lost productivity — and they degrade public safety by delaying emergency response times. This is no longer just a transportation issue. It’s a systemic, cross-sector problem affecting quality of life, economic growth, and urban resilience.”
For Steven Crutchfield, senior vice president, MD of international at Verra Mobility, the first issue is safety, particularly in terms of reducing dangerous driving behaviours. “The second thing would be efficiency,” he says. “AI can sense traffic congestion and then help redirect traffic in those locations. It can also help when emergency management officials need to get somewhere fast – by either redirecting them to a different route or clearing the first one through redirects. These first two assist in helping to solve another problem – pollution and cleanliness. By improving traffic flow, you reduce idling time and gas use, which helps the community for a variety of reasons. All of these items are interrelated and work to improve several problems simultaneously.”
Promise of AI
The promise of AI in transportation has always been alluring: fewer crashes, faster commutes, cleaner air, suggests Jatish Patel, CEO and founder of Flow Labs. “But for many years, we lacked the essential ingredient to make it happen: good data,” he says. “Today, we face the opposite problem. Cities are drowning in data, from GPS pings and traffic cameras to signal timing plans and crash records - but often can’t make sense of it quickly enough to act. This is where AI is already transforming traffic management and where it’s heading next. What cities really need isn’t more data. They need better decisions, made faster. This is where AI, especially spatial-temporal machine learning, has unique power. It can ingest high volumes of variable-quality data, correct for inaccuracies, and provide predictive insights in real time. Rather than just flagging problems after they occur, AI can diagnose root causes, simulate alternatives, and recommend interventions, often within seconds.”
"AI can analyse patterns in real time and anticipate congestion or violations before they escalate” Akif Ekin, Ekin Smart City Technologies
This is welcome, because city and transport authorities are under immense pressure, points out Ekin. “They’re being asked to reduce congestion, improve safety, and cut emissions — all while operating with ageing infrastructure and tight budgets,” he says. “In many cases, cities are still relying on reactive approaches and decade-old systems that lack real-time adaptability. The demand is growing, but the tools haven’t kept up — leading to a widening gap between what’s needed and what’s possible with current resources.”
What AI does is to introduce predictive intelligence and automation into traffic systems. “Rather than simply recording what has already happened, AI can analyse patterns in real time and anticipate congestion or violations before they escalate,” Ekin says. “It enables faster incident detection, dynamic traffic signal optimisation, and automated enforcement without human bias. The result is not just smoother traffic flow, but safer, fairer, and more efficient cities.”
For Crutchfield, AI-powered traffic management is another tool, not a one-size solution. “It makes work streams a lot more efficient and enables us to dive deeper and get more meaningful and actionable insights,” he says.
“One example is addressing obscured licence plates where drivers deliberately break the law. These enforcements are there for a reason – to have compliance on the roadways and to keep people safe. With AI, even when plates are deliberately obscured, systems can look for the signature on the vehicle themselves, including markings or unique damage, to maintain enforcement capabilities. This puts everyone on the same playing field and is held accountable to the law. Another example would be observing and enforcing (where necessary) driver behaviour, such as using mobile phones while driving or correctly using seatbelts. Initially, this information could be used for better driver education and enforcement to ensure drivers are using seatbelts and are not distracted by mobile phones while driving.”
Legacy systems
Some confusion arises on a couple of points: 1) what will AI do which is not done by legacy systems? And 2) don’t we already use some AI in these systems? “The traditional model for managing traffic relied heavily on hardware-centric solutions: sensors in the road, cameras on poles, controllers in cabinets,” says Flow Labs’ Patel. “These systems are expensive to install and maintain, take years to deploy, and often deliver incomplete or low-resolution data. Worse still, they don’t scale well, particularly at a time when transportation departments are being asked to do more with shrinking budgets.”
He sees a fundamental shift from hardware-based infrastructure to software-first intelligence. “Instead of laying more devices in the ground, we use geospatial AI [GeoAI] to harness and fuse existing data streams, regardless of their source or quality, into actionable insights. This not only lowers costs and deployment time but also makes advanced traffic management accessible to cities of all sizes.” Flow Labs views GeoAI as a new kind of infrastructure, rather than an add-on; what once required tens of thousands of dollars in equipment and weeks of field set-up can now be achieved through software that deploys in days, and at a fraction of the cost, Patel argues.
“What cities really need isn’t more data: they need better decisions, made faster” Jatish Patel, Flow Labs
“While some legacy systems use rudimentary algorithms or rules-based logic, true AI goes far beyond that,” insists Ekin. “AI systems can learn from millions of data points, adapt to changing conditions, and optimise themselves over time. For example, traditional enforcement cameras might capture a red-light violation — but an AI system can recognise nuanced behaviours, distinguish between emergency and civilian vehicles, and reduce false positives through contextual awareness. Also, legacy systems typically function in silos — AI breaks these down by integrating enforcement, traffic flow, public safety, and even environmental impact into one continuous loop of insight and action.”
AI can do calculations for you in real time, or in any computer model run that you want, adds Verra’s Crutchfield: “That enables you to make quick, informed, and dynamic decisions. Rather than using historical data, AI can look at a series of its own parameters and quickly determine that a current traffic situation is caused by an x-incident or y-incident. We've been using AI for many years now; I think the big insight is just the sheer amount of different factors AI can take into account when it comes to decision-making. AI helps because you're not just looking at traffic flow, you're looking at the environment around you, how many vehicles are involved, the people impacted, and the infrastructure.”
Speed and accuracy
The speed and accuracy that AI enables is unprecedented, he adds: “Therefore, unsafe or incorrect use of the roadways and transport infrastructure can be easily monitored and identified. Examples could be unauthorised driving in dedicated bus lanes or peak-hour flow lanes, unsafe driving around public transport systems like buses and trams, real-time erratic driving alerts to emergency services, tail-gating/driving too close on major arteries, illegal/unsafe u-turning and even impeding traffic by driving too slow.”
US state agency deployments in places like Florida, North Carolina and Virginia are proving that AI can modernise entire signal systems without the need for costly new infrastructure. “In North Carolina, what once took 30 days of manual labour, assembling controller data, cross-referencing spreadsheets, and writing signal performance reports, now takes just 30 seconds,” Patel says. “With real-time analytics deployed at scale across more than 2,500 intersections, engineers now spend less time wrangling data and more time solving real problems. These gains aren't just operational, they're financial. Analytical tasks that once required dedicated consultants or extensive staff hours can now be completed instantly, helping the state reduce programme costs while expanding its capabilities. And as the next evolution takes shape, AI-powered tools will begin to automate signal retiming itself, transforming a task that once cost thousands per intersection into a fast, software-led process measured in hundreds.”
Ekin has deployed AI-powered mobile and fixed systems in over 40 countries. “Our solutions help cities automate speed enforcement, monitor illegal parking, and analyse traffic flow in real time,” says Ekin. “One standout case is our work in the US East Coast, where our bike-mounted enforcement solution helps cities maintain safety in congested downtowns — all without the need for new infrastructure. In Europe and the Middle East, cities use our unified AI platform to manage traffic and public safety across hundreds of intersections, dramatically improving both compliance and commuter experience.”
“The big insight is just the sheer amount of different factors AI can take into account when it comes to decision-making” Steven Crutchfield, Verra Mobility
Verra’s Crutchfield says: “AI can help with road maintenance and help identify when there are problems like roadway debris or potholes. Cities can preemptively get ahead of maintaining their roads or infrastructure, which could cause not just inefficiencies but potential danger safety hazards. AI has a lot of potential to assist us, but it’s important to know that it’s one tool in the tool belt. Using it as part of the road safety toolkit helps to create a more impactful, efficient, and safer community.”
Also, AI is only as effective as the data it’s built on, cautions Patel. “In Florida, challenges like device malfunctions, power outages, and network instability made reliable traffic data hard to come by, especially in regions where physical infrastructure was incomplete or ageing.” To tackle this, Florida Department of Transportation partnered with Flow Labs to launch an integrated platform - called Heidi – where data from existing ITS devices and controllers is combined with high-resolution probe data and processed through machine learning models that identify and correct for detection failures and gaps in coverage.
“The result is not just cleaner data, it’s a far more accurate picture of traffic operations in real time,” says Patel. “Whether calculating turning movement counts or measuring the impact of red-light running policies, FDoT is now using this enhanced dataset to guide decisions across hundreds of signals, with plans to expand the system to more than 2,000.”
Integrated solution
And unlike traditional methods that rely on costly field studies or fixed-site sensors that can run upwards of $50,000, this integrated solution delivers actionable data at a scale and cost that was previously out of reach.
In major cities like Seattle and Washington, DC, AI is helping integrate multimodal data, prioritise safety interventions, and better understand travel patterns, Patel continues. “Meanwhile, in smaller municipalities like Brownsville, Texas and Grand Forks, North Dakota, AI is levelling the playing field, providing access to the same advanced capabilities once reserved only for the largest and best-funded cities.”
All this means that AI is not just a buzzword, it’s a practical tool for solving real-world challenges at scale, with real budget impact, says Patel. “Whether reducing per-intersection study costs from four figures to double digits, or compressing project timelines from months to weeks, the financial efficiency of AI is proving as transformative as its technical capability,” he concludes. “The future of transportation won’t be defined solely by the vehicles on the road or the sensors in the pavement. It will be defined by how smartly, how quickly, and how fairly we manage the systems behind them.”
