Driverless cars will take longer to get on to roads than advocates anticipate

Irish Times

05-06-2025

The UK government has indicated a delay to at least 2027 for its plans to start rolling out self-driving cars on British roads. Governments, at least those in democratic societies, should stop putting a number on it, particularly as the challenge extends far beyond the technology itself.
Think of Apollo 11, indeed the whole Apollo programme. Getting humans to the moon was quite an achievement. There were issues along the way, most notably the Apollo 1 tragedy. The methods both for testing the technology and execution were, however, relatively easy to control.
The total number of people at risk remained low because of the nature of the endeavour. Everyone involved, including everyone at risk, was an expert to some degree in the field. The risk factor could be mitigated enormously.
Now think about the microwave oven. It was invented in the 1940s, about a decade before the first person went into space. Yet it didn't become a household item before the 1980s, more than a decade after Neil Armstrong took his stroll.
READ MORE
Every fear, founded or unfounded, about the technology in microwave ovens had to be tested substantially. The end user was rarely going to be an expert and fears had to be allayed.
[
Would I use a driverless taxi again? It obeyed traffic signs, stopped for pedestrians and drove cautiously
Opens in new window
]
There were elements that took years to fix, such as evenly cooking an item. Then there were infrastructure issues including the method of power supply to homes coupled with the increased demand. When you factor all that in, no wonder it took so long.
Self-driving cars have managed the moon landing bit. Their problem is the chilled or frozen ready meals part, including telling chilled apart from frozen.
For full disclosure, self-driving cars would be quite beneficial to me. I don't drive and have no desire to ever learn. Having a vehicle that just brought me places while I played with my phone oblivious to the world would be fantastic.
But having them on actual roads in a regular form is still a matter of fantasy for quite a few reasons.
Getting approval for roadworthiness is more than a government, or rather the executive branch of a government, waving a magic wand. In pretty much every functional democracy on earth, there are agencies or departments that focus on rules around transport
The technology is the obvious part. It has come an awfully long way. As a feat of engineering it merits real credit. The improvements made in terms of recognition of pedestrians, animals, other vehicles and even inclement weather conditions are considerable compared with when the hype began.
The problem is that great isn't good enough. Quite frankly, amazing isn't a high enough standard. When interacting with the public at large, anything short of perfect is a problem. No combination of lidar, radar and camera technology has yet hit a standard that could be considered flawless.
The contention in defence of the technology might be that plenty of accidents occur already on our roads due to simple human error. The difference is that there are penalties for making mistakes on the road or being reckless up to and including the option of prison for the offending parties. Putting a robot behind bars isn't really an option – and even less so penalising the person who designed or programmed the robot.
Beyond the functional technology, there's an enormous logistical challenge – really a combination of enormous tasks.
Across Europe, let alone Ireland, the road system varies substantially – even to the extent of driving on different sides of the road. All of that data needs to be manageable for a driverless vehicle.
Solving that includes relying on technology beyond the control of self-driving car makers. GPS is the big one here as gaps in coverage are a real problem that human drivers have to navigate for themselves today.
The biggest issue is administrative, however. Getting approval for roadworthiness is more than a government, or rather the executive branch of a government, waving a magic wand. In pretty much every functional democracy on earth, there are agencies or departments that focus on rules around transport. That's the barrier before a taoiseach or prime minister can think of giving a green light.
Rightly, those agencies have rigorous standards because their first priority is the safety of everyone on the roads including pedestrians – not just the advance of technological convenience. Their second is optimising the flow of traffic. Self-driving cars have to clear onerous hurdles on both of these which, again, involve macro factors beyond the control of the manufacturers.
If this negativity is sounding repetitive, that's rather the point. I've had a selfish interest in this technology succeeding for a long time. In 2017, I was told we'd see regular use by 2020; in 2019 it was promised by 2023. Granted there was a pandemic but all of the makers missed the wider point.
Factors they cannot and never will control are inevitably going to slow the process. To be approved, they have to be safe and convenient for everyone.
Apollo 11 was never going to get into a pile-up near a toll booth on the M50. It had literally nothing in its path. The very purpose of a self-driving car is to work in a crowded space. Fitting in takes time. Advocates of the technology keep underestimating how much time it will take.