Rethinking Passenger Safety in Driverless Taxis

As the UK prepares for 2026 self-driving taxi pilots, researchers at Loughborough University, in a project funded by the Department for Transport, have used virtual reality to place members of the public inside simulated robotaxi emergencies—from collisions to flooding—to understand how passengers respond without a human driver present. The research shows that emergency handling in autonomous taxis isn’t a single scripted action: needs vary widely across visually impaired, hearing impaired, and neurodivergent users, and passengers want clear, calm, accessible guidance along with the ability to contact a real person. The findings are now feeding into emerging government standards aimed at ensuring future autonomous taxi services are inclusive, reliable, and trusted by all.

Transcript

00:00:02 So, here at Lur University, we've been working on a project funded by the Department for Transport, investigating how members of the public respond to different emergency scenarios in self-driving taxis. And we've been using virtual reality to put people into these experiences so they feel that they're really there in those taxis. When our users put on a VR headset, they

00:00:24 are effectively taken into the seat of a robo taxi. So although we don't currently yet have self-driving taxis on our roads, deployment is nearing. We're going to have pilots in spring 2026. And so this raises a really important question because self-driving taxis remove the driver. What happens in an emergency?
>> So for instance, they might have a

00:00:46 pedestrian that's trying to get in. Maybe they're stuck in a flood. Maybe they're being hit from behind by another vehicle and then they're dealing with the aftermath of a collision. So, lots of different scenarios to really understand what it might be like to encounter that and most importantly, what do you think you would do and what the vehicle and the service should

00:01:06 provide. The main finding really is emergencies in self-driving taxis aren't one action. And without a driver, that process becomes really pushed onto the passenger themselves. So, needs really vary. Visually impaired users are going to need strong audio guidance whereas hearing impaired users are going to need visual prompts and captions. We also found that neurode divergent users found

00:01:29 it very difficult in that moment to be able to think clearly. So they needed very calm structured low simulation instructions. And the overriding message that people had is that they wanted the ability to contact a human. Virtual reality is brilliant for doing this type of thing, mainly because it would be totally unsafe to do this in the real world. We can't put people into

00:01:51 emergency scenarios with like a flood or a fire. Also, these vehicles don't exist on our roads yet. They will shortly in a limited way, but we can put people into the future here to understand what it might be like for them to experience this.
>> So, next is to actually turn these insights into practical standards for operators and designers. And that's the

00:02:13 process that's now going through at government stage. So as self-driving taxis scale and as we have pilots in the UK in 2026, this is how the government is going to ensure that any type of self-driving taxi is inclusive and reliable for every passenger.