New Study Reveals Runners Take Greater Risks Around Self Driving Cars Prompting Rethink of Road Safety Systems

New Study Reveals Runners Take Greater Risks Around Self Driving Cars Prompting Rethink of Road Safety Systems

(IN BRIEF) A study by the University of Glasgow and KAIST has revealed that runners behave significantly differently from walkers when interacting with self-driving cars, often taking greater risks to maintain their pace. Using augmented reality simulations, researchers found that runners process less information, rely more heavily on visual signals, and are more likely to make unsafe crossing decisions, including instances of simulated collisions. The findings suggest that current automated vehicle systems may not fully account for this behaviour, highlighting the need for improved communication tools such as external Human-Machine Interfaces. The study proposes new solutions, including the DualBeam lighting system and wearable alerts, to help vehicles communicate more effectively with runners and improve road safety as automated transport becomes more widespread.

(PRESS RELEASE) GLASGOW, 31-Mar-2026 — /EuropaWire/ — University of Glasgow, in collaboration with KAIST, has conducted a new study exploring how runners interact with self-driving vehicles, revealing important behavioural differences that could reshape how automated vehicles are designed to ensure road safety.

The research challenges prevailing assumptions that pedestrians behave cautiously around traffic, showing instead that runners often take greater risks when crossing roads. Using augmented reality technology, the team simulated real-world scenarios to compare how runners and walkers respond to approaching automated vehicles at junctions.

The findings indicate that runners are more inclined to prioritise maintaining their pace over assessing road conditions. As a result, they often spend less time evaluating traffic and are more likely to attempt crossings in front of oncoming vehicles. In several simulated cases, participants running through the test environment were virtually “struck” by vehicles, highlighting the potential risks associated with this behaviour.

The study suggests that current automated vehicle systems may not adequately account for these differences, as they are typically designed with more cautious pedestrian behaviour in mind. To address this, researchers are advocating for improved communication systems between vehicles and road users.

A key focus of the research is the development of external Human-Machine Interfaces (eHMIs), which use visual signals on the exterior of vehicles to communicate intent. These systems could replace traditional cues used by human drivers, such as eye contact, gestures, or subtle changes in speed, by providing clearer and more immediate signals to nearby pedestrians.

The research team tested two eHMI concepts: a LightRing system, which uses simple red and green signals, and a CyanBand system, which employs animated lighting to indicate acceleration or deceleration. While both systems improved participants’ ability to interpret vehicle behaviour, the results varied significantly between walkers and runners.

Walkers were generally able to interpret both systems effectively, often slowing down to confirm the vehicle’s intent before crossing. This allowed them to make safer, more informed decisions. Runners, however, were less likely to adjust their pace, which reduced their ability to process more complex visual signals. The study found that runners struggled particularly with the animated CyanBand system, whereas the simpler LightRing design proved more intuitive for both groups.

The research also highlighted behavioural differences in decision-making. Runners tended to rely heavily on visual signals without cross-checking them against the vehicle’s movement, sometimes leading to unsafe decisions. In contrast, walkers combined visual cues with observed vehicle behaviour, resulting in higher levels of trust and safer outcomes.

Professor Stephen Brewster noted that despite the widespread popularity of running globally and the rapid growth of automated vehicle usage, little prior research has examined how runners interact with driverless cars. He emphasised the importance of designing systems that can communicate effectively with a diverse range of road users.

The study also draws on insights from researcher Ammar Al-Taie, who highlighted that runners often face both physical and cognitive pressures when approaching crossings, making them more likely to accept risk in order to maintain momentum.

Building on these findings, the team has proposed a new eHMI concept called DualBeam. This design uses colour-coded light signals—amber to indicate that the vehicle will not yield and purple to signal that it will allow safe passage—aimed at improving clarity without relying on conventional traffic light colours. The concept also includes the possibility of integrating alerts via wearable devices such as smartwatches or earbuds, providing runners with additional cues about approaching vehicles.

The research, titled Running into Traffic: Investigating External Human-Machine Interfaces for Automated Vehicle-Runner Interaction, will be presented at the CHI 2026 in Barcelona in April 2026. The findings are expected to contribute to the development of safer and more responsive automated vehicle systems capable of accommodating a wider range of human behaviours.

Media Contact:

email: media@gla.ac.uk

SOURCE: University of Glasgow