What self-driving cars should be used for now - and it's not the rich, declares Edinburgh University AI expert

Subramanian Ramamoorthy, professor of robot learning at the University of Edinburgh, warned fully autonomous vehicles could still be some way off, and interim advances to assist drivers would be equally valuable.

He said: “One of the things I'm interested in in driverless systems is not just how you make cars autonomous, but how you make assistance more useful.

“Right now, there's no assistance feature that takes into account the loss of attention and reaction time that comes with ageing or other forms of reduced driving skill.

“It would be nice if we could design assistive features that go towards that.”

Prof Ramamoorthy said that with an ageing population, that could be crucial in enabling people to drive safely for longer, since having to hang up the car keys could lead to a major loss of mobility and independence.

By contrast, he said driver assistance features such as adaptive cruise control, which adjusts a car’s speed to maintain a safe distance from the vehicle in front, were confined to high-end models.

He said he would like to focus future research in this area rather than the goal of full automation, which is known as “level five”, on a scale of zero to five.

Prof Ramamoorthy said: “You can get very far with very clever assistance [rather than full automation] in terms of road safety gains. 

“People have to stop driving at some point and that changes their mobility and their life, but there is no assistive feature in a car yet that could help with that.

“All the assistance features seem to be aimed at healthy, but very wealthy, people who say ‘I want to take my meeting in the car’."

Prof Ramamoorthy, chair of robot learning and autonomy at the university’s school of informatics, said older drivers had different needs.

He said: "You might not be watching your blind spots when navigating junctions, or you might be slower to react.

“You could have a lower-end car with very good assistive features that could be used by an ageing population.

“The most common thing that people might not pay attention to is when they’re driving beside a cycle lane, so they may be at greater risk of hitting a cyclist.”

Alastair Dalton’s newsletter will bring you the best Transport stories and analysis from across Scotland

Prof Ramamoorthy said future devices could also detect that a driver had not checked their blind spots before manoeuvring, or they were not following the best path through a junction, and could slowly adjust the vehicle’s course.

He said: “These are things on my research plan. That's the kind of thing that would be unique about the way we are approaching the problem, because we are not a company trying to do level five autonomy.”

Prof Ramamoorthy said increasing automation was unlikely to lead to incapacitated drivers being driven home, but it would help on the margins - although he stressed he was not encouraging drink driving.

He said: “The idea that somebody who's completely knocked out is going to drive an autonomous car is less realistic.

“What's much more realistic, which is more typical, is if you've had one drink too many, it would be nice if your car could have a safety blanket around you and protect others on the road who will be impacted as well.”

Prof Ramamoorthy also urged caution over expectations that fully self-driving vehicles were just round the corner, despite optimistic projections from some in the sector.

He said: “When I first started working in this area ten years ago, I was told 'you're already a bit late because it's all solved and going to happen now'.

“I would point out - what about all the safety issues? I was told: 'You don't understand, that stuff just gets done'. And of course, nothing got done.

“They are still discussing the same things. We still need a lot of basic science to be done. A few diehards like Elon Musk have been saying for years that it's only a few years away.

“Most level-headed people have accepted that people were over optimistic.”

Although Waymo - formerly Google’s self-driving car project - is operating taxis with no drivers on board in San Francisco, it remains very much the exception rather than the rule.

How soon does Prof Ramamoorthy expect such “robo-taxis” to appear in the UK? His response reflects the uncertainty among experts, as he said: “Good question. I do not know the answer.”

He said Waymo’s initial operating conditions in Phoenix, Arizona - little rain, wide roads, low traffic and few pedestrians - were a world apart from Scotland.

He said: “They have expanded since then to urban environments, and they are rightly being very careful in how they are rolling it out.

“Still, I don't think it would cope well with European traffic out of the blocks. They will need to do more.”

That “more” is the basic science around how everything else on the road behaves.

Prof Ramamoorthy said: “What would it take for a car like this to meaningfully interact with people and dense traffic.

“The ‘long tail’ is all about people and cyclists and cats and dogs - things that are not in the datasets.

“We don't have a very systematic understanding of what people do in crowded urban environments.

“For instance, in Britain, people think it's perfectly normal, especially in such areas, to walk on the road.

“Everybody breaks the rules to some extent. You have a cycle lane and a cyclist comes out of it, and we all accept that.”

When MPs passed the Automated Vehicles Act in May, the-then Conservative UK transport secretary Mark Harper said it cleared the way for self-driving cars by 2026, declaring: “Britain stands at the threshold of an automotive revolution”.

But Prof Ramamoorthy contends: “I take a more measured view of the prospects. There's a lot of cause for optimism and there is plenty happening in the UK.

“But that said, getting these developments into useful products needs more careful thought.”