New research has suggested that Autonomous vehicles (AVs) should adopt more cautious driving styles in riskier road scenarios and signal safely earlier through deceleration. 

The new International study led by Coventry University’s Research Centre for Future Transport and Cities, in collaboration with Deakin University, Australia, comes as the government projects AVs could be on our roads by 2026 with up to 40% of new cars expected to be self-driving by 2035.

To ensure safety as a top priority this study of 281 participants has revealed how pedestrians make decisions when crossing roads in front of AVs.

It found that people assess factors like road gradient, weather, vehicle type and distance before crossing and that pedestrians were more likely to cross when AVs were travelling uphill, in sunny weather, and when smaller vehicles were further away.

Study lead Sachita Shahi, a postgraduate researcher at Coventry University and Deakin University, said: “By identifying the conditions that make pedestrians feel safer, we can inform AV design to better support safe and predictable road interactions.” 

Dr William Payre, Assistant Professor at the Research Centre for Future Transport and Cities, who contributed to the study and is overseeing Sachita’s research, added: “Regardless of how advanced AV technology becomes, people still rely on their lived experiences as pedestrians. It’s not just about the technology - it’s about how people interpret risk based on what they know.” 

“When the road was steep, they were more cautious. It wasn’t about the AV, it was about their own experience. The sensors and the vehicle try to interpret how a pedestrian is behaving to adapt how the vehicle drives. We’re trying to give a human flavour to the data, not just numbers, to help AVs think like humans and understand human behaviour.” 

The team now plans to take their findings into a virtual reality setting, using eye-tracking and behavioural data to deepen understanding of pedestrian decision-making.