UK roads deemed insufficient for autonomous cars
22 May 2017
Author: Jack Carfrae
The fleet industry and technology experts have dubbed the UK's road network unfit for autonomous cars in its current state, citing road surfaces, highway furniture and the frequency of roadworks as obstacles to wide-scale adoption.
Speaking to BusinessCar, the chief executive of the British Vehicle Rental and Leasing Association, Gerry Keaney, said: "There's still a lot of work to be done for the best and most effective application for an autonomous vehicle. There's also a question mark about whether or not our roads are up to scratch, and we're going to get into this rather perverse situation where the autonomous vehicle is actually more reliable than our roads.
"These things are only as smart as they are allowed to be, and if they are relying on sensors to pick up road markings or speed signs, then it needs to be there and consistent."
When asked if current UK roads could cope with autonomous cars, Matthew Avery, director of research at vehicle safety experts Thatcham, replied: "The answer is no. Very simply, autonomous cars need roads they can read."
Avery, who advises the insurance industry on autonomous car technology, said the vehicles depend on their ability to clearly decipher a lane or road surface, which can be hampered by poor conditions or a lack of maintenance. "They rely on something like the white lines as their principal reference point. If you've got two white lines - a dotted line in the middle and a solid line at the edge of the road - that's the best. These camera systems are black and white, so they need contrast, but as soon as you lose the contrast due to weather conditions - standing water, snow, anything that degrades the white line - then the car has a problem following it.
"If that white line's not there at all, or there's only one white line, the car will try and judge from the road's edge. With potholes, the road surface is integrated with white lines, so suddenly the line disappears [when a pothole forms]; then someone fills it in with tarmac and doesn't replace it. If that line's been tarmaced over, the car hasn't got a chance."
Early automated cars will be confined to motorways, where lane markings are generally less of an issue, but roadworks would require the driver to resume control.
"The initial automated cars you're going to see are likely to be geofenced to motorways, and that's where most of our best white lines are," said Avery. "Generally, the initial phases of automated driving probably won't have too much of a problem - apart from where you've got roadworks. Then, the systems are not going to be able to function, and you're going to have to have a robust automated system that hands back control to the driver. It's going to have to tell the driver 'I can't work in these conditions reliably; I can't understand traffic cones'."
Avery said poor road surfaces could also make it difficult for the vehicles to predict the behaviour of other traffic: "When you've got other road users, such as cyclists, they swerve round potholes, and it's going to be very difficult for an automated vehicle to try to predict path of the other vehicle; it shouldn't just run into them, but it is a difficulty. As a driver, you might predict it - 'there's a pothole coming up, I think he's going to pull out, I'll back off' - but the automated vehicle will suddenly react to a cyclist bizarrely moving out in front of it."
David Wright, director of strategic initiatives at Coventry University, said autonomous cars would likely be able to handle potholes as the technology developed, but mass adoption would require dedicated lanes, particularly in urban areas.
"Larger perturbations in the road surface might be an issue if an autonomous vehicle came along tomorrow, but I don't see it as a massive issue when we get to larger-scale adoption. I think, to make it work, we are likely to have to see dedicated autonomous lanes in cities.
"It's an opportunity, not to increase congestion, but to increase traffic density.
If we've got control over certain restricted areas for autonomous vehicles, then they can travel very closely together, and if we do that, we're actually making more efficient use of the road surface. So there's an argument that if we do autonomy well, we can actually reduce the pace at which we need to invest in new road infrastructure - so we use every square foot of tarmac instead of having big gaps between vehicles."