With driverless taxi services poised to expand across European cities this year, a new study is urging operators to equip vehicles with IoT enabled on-board traffic lights to signal to runners and pedestrians when it is safe to cross.
Researchers from the University of Glasgow and KAIST in South Korea warn that runners, in particular, take more risks at crossings than walkers, potentially creating new hazards for autonomous vehicles unless safety systems are adapted.
The study used augmented reality (AR) to simulate urban streets and investigate how runners and walkers interact differently with self-driving cars (AVs). While much AV research assumes that pedestrians behave cautiously, the team found that runners are a distinct—and riskier—group of road users.
“In our research group, we’ve focused on developing systems to help autonomous vehicles communicate safely with cyclists and pedestrians,” said Stephen Brewster, Professor of Human Computer Interaction at the University of Glasgow. “But runners have largely been overlooked, even though running is the most popular physical activity worldwide, with over 600 million recreational runners. And with millions of AV trips now happening monthly, it’s clear that runners and self-driving cars will increasingly share the road.”
What are eHMIs?
To study this, 24 participants wore AR headsets outdoors, overlaying virtual AVs on real streets. They were instructed to either walk or run toward a junction while a simulated vehicle approached. The cars either displayed no signals or used one of two “external Human-Machine Interfaces,” or eHMIs—a term for vehicle-mounted lights that communicate the car’s intentions, essentially acting as on-board traffic lights.
The first eHMI, LightRing, used a simple red-and-green light ring around the car. Green indicated it was safe to cross, red signalled the vehicle would not yield. The second, CyanBand, displayed animated cyan lights that swept inward to indicate braking and outward for acceleration.
Results showed that walkers had the time to process the signals and adjust their crossing safely. Runners, however, were focused on maintaining pace and often ignored the signals. “We observed runners taking risks to keep moving, sometimes even running through a red signal in the simulation,” said Ammar Al-Taie, Corresponding Author and Researcher at KAIST. “In two cases, participants collided with the virtual car despite seeing red lights. Runners are less patient than walkers, and their split-second decision-making makes them a challenging class of road users for autonomous systems.”
The researchers argue that traditional AV programming, which expects pedestrians to stop or yield predictably, may not be sufficient for mixed-use streets. The team’s proposed solution is a new eHMI design called DualBeam, which uses two rows of lights on each side of the vehicle with colours optimised for quick comprehension. Instead of the standard red and green, amber indicates the car will not yield, and purple signals the vehicle intends to stop, giving runners more time to process the message without breaking stride.
The role of DualBeam
DualBeam could also integrate early-warning alerts to runners’ smartwatches or earbuds, helping them gauge approaching AVs while continuing their exercise. “Crossing roads while running feels very different from walking,” Al-Taie said. “Slowing down to let a car pass takes physical effort, and your mind is also focused on maintaining pace. This makes runners more tolerant of risk, which AVs need to account for if we want to prevent accidents.”
The study highlights a wider lesson for the IoT and AV industries: autonomous vehicles will need more nuanced communication systems for diverse road users. Simple lights or signals may work for pedestrians, but more dynamic, legible cues are necessary for people in motion, from runners to cyclists.
Interestingly, the type of eHMI makes a notable difference. Walkers found the animated CyanBand lights useful for judging speed and intent, but runners struggled to interpret them in time. The simpler LightRing signals were immediately understandable to both groups, suggesting that clarity and speed of communication are key in eHMI design.
Professor Brewster added: “As AVs proliferate, thinking about human behaviour isn’t just about safety—it’s also about trust. Pedestrians and runners need to understand what a car is doing instantly. Miscommunication could lead to hesitation, near-misses, or worse. eHMIs are essentially IoT devices for the street: they allow machines and humans to share a language and make decisions in real time.”
IoT implications
The implications for fleet operators are clear. DualBeam or similar systems could be integrated into existing AV platforms through software updates and LED hardware.
For IoT device manufacturers, this research suggests opportunities to integrate eHMIs with wearables. Early-warning alerts to smartwatches or earbuds could become a standard safety protocol, connecting AVs directly to the devices people already carry, without interfering with the runner’s experience.
And by leveraging platforms like Strava, which millions of runners use to track pace and routes, AVs could even anticipate when and where runners are likely to appear, enabling smarter, context-aware notifications and enhancing urban road safety.
There’s plenty of other editorial on our sister site, Electronic Specifier! Or you can always join in the conversation by commenting below or visiting our LinkedIn page.
