Bryan Mistele, Co-founder and CEO of INRIX.
The last thing I remember before I became a road safety data point was turning on Lake Washington Boulevard. It was August 2020, and my son and I were on our bikes, heading out for a trail ride in Kirkland, Washington. The driver of the car that hit me never saw me coming.
I woke up once in the ambulance for a few seconds, then again in the hospital. I was lucky. I left the hospital that day with more than two dozen wounds, some scars and a massive concussion—but I left the hospital.
There’s nothing special about my accident. The car pulled into the bike lane to check traffic, a large hedge blocking the driver’s sightlines. It was a spot well-known for poor visibility and high risk of accidents. It sits on Lake Washington Boulevard, one of the most dangerous biking roads in the Seattle area. In short, it was an accident waiting to happen.
Is it any wonder, then, that my accident radically changed how I thought about road safety, particularly the protection of vulnerable road users (VRUs)? These are some of the most at-risk people on the road: pedestrians, cyclists and others who increasingly use methods of micro-mobility to navigate our streets and roadways. Drivers often don’t see them. Street infrastructure isn’t designed for them. Their vulnerability is often an unfortunate outcome of our transportation systems.
As cities, traffic planners and state departments of transportation aim to ensure all of us are safe on our shared streets and roads, AI plays a key role in making this happen.
The Role Of AI In Traffic Safety
Intersections are some of the most dangerous places for drivers and VRUs alike. According to the Federal Highway Administration, roughly one-fourth of traffic fatalities and half of all traffic injuries in the country are attributed to intersections. While pedestrian and cyclist fatalities decreased slightly in 2024, traffic fatalities still are stuck above their pre-pandemic levels. This shows an urgent need to make our intersections much safer.
Traffic safety is a complex problem, though. There’s no singular cause for the thousands of accidents and fatalities on our roads each year. Some are caused by bad signage, while others are the result of speeding, cross traffic or—like mine—poor visibility. A problem like this can’t be approached with reductionist thinking. We must address road safety systematically, looking at the problem as a complex whole.
A key piece of the solution lies in data and our use of AI. There exist dozens of different datasets, from speed data to traffic patterns to data on near misses. AI allows us to use that data to understand the risks specific to individual locations. GenAI can look at thousands of cities across the world, inform us of best practices and then predict the benefits of each potential solution.
How AI Is Working To Protect VRUs Today
We’re already taking steps toward safer roads thanks to cutting-edge technology. The University of Washington has taken on intersection safety with its Cooperative Perception System (CPS) that integrates advanced algorithms, visual and thermal cameras, and 3D object detection to identify potential collision scenarios.
Similarly, UCLA has developed a system called InfraShield (now part of Smart Intersection) that integrates multimodal sensor data with machine learning modules. Using high-definition maps and historical trajectory data, InfraShield can identify conflict points and calculate time-to-collision scenarios.
Both systems show how data and innovation can make intersections safer for everyone, especially VRUs. If we combine cutting-edge technology, including AI, with a wealth of available data, we can create solutions to our traffic safety problems that align with the National Roadway Safety Strategy and USDOT’s goals of making roads safer for all of us.
Piecing Together The Puzzle Of Safer Roads
At INRIX, we’ve learned firsthand how challenging it can be to make our roadways safer for all users. One of the first challenges we ran into while developing an AI-powered mobility intelligence product was the absence of integrated data.
Though we had data on traffic fatalities, we were missing key pieces—data about bike and pedestrian accidents, data around traffic near misses and more. To get a more complete sense of the problem, we worked to get near-miss data from GM and integrate that data to create a fuller picture of the problem.
Traffic safety comes with another core problem. Often, there is no single cause for accidents. Some might stem from speeding, others from bad signage, and still more from poor design or visibility. Without one root cause, finding solutions can take time and resources.
This is where AI can play a vital role. AI helps to make intelligent assessments across multiple datasets. Using AI, we can look at thousands of cities worldwide and see how engineers have solved similar problems. AI can also predict the benefits of possible solutions.
Getting Us Home Safely
AI moves fast. It’s difficult to imagine what possibilities for traffic safety will exist two years from now. This is partly because more data is becoming available from cars. Historically, we had just two data points to work with—speed and location. But now we can pull data from cameras, things like near misses, parking spaces and other info that can tell the causes behind unsafe environments.
To protect VRUs—people like me and my son, out for a ride on our bikes—AI-driven improvements will continue to play a key role.
This means opening access to internal data from cities and states, allowing AI to crawl these data pools so that our understanding of traffic risks and solutions is sharper and will protect our most vulnerable users. More and quicker access to data means better traffic outcomes. A key step is to forge working partnerships between government agencies with this data and organizations that can use this data to foster solutions.
Traffic safety is not just data; it’s people. AI allows us to make intelligent assessments and create solutions to make each of us safer. Let’s all work together to make sure each of us gets home safely.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
