The recent introduction of self-driving buses in England was hailed as a world first, but self-driving vehicles are nothing new. A similar scheme was tested years ago in the centre of Helsinki and the first self-driving vehicles were cruising the German autobahns as early as the 1980s as part of the Prometheus system. Each new iteration makes a small step forward. But the truth is, that when it comes to self-driving technology, we’ve already picked all the low-hanging fruit. The last stretch will be most difficult to conquer. The question is, what price are we ready to pay for self-driving vehicles to overcome the remaining issues?
Self-driving vehicles and the complexities of terrain
Waymo self-driving cabs might be driving to acceptable standards on the simple grid of wide asphalt roads in the sunny cities of the American south, but even this simple environment proves to be too challenging for them on regular basis. Would you trust them to drive over Bealach na Bà in the middle of a snowy Scottish winter? If placing a traffic cone in front of their sensors can render them completely helpless, how do you think they would navigate through a typical Polish roadwork site?
As a truck driver, I have the dubious pleasure of driving modern trucks with the newest driver assistance systems. And those systems are nowhere near ready to tackle challenges of a typical driving day. My opinion seems to be shared by virtually every truck driver I’ve discussed this with. Yes, the system might save many lives by preventing a tired driver from rearending another truck that stopped in traffic on the motorway jam, but it goes completely off the rails when faced with challenges such as single-track roads in the Scottish Highlands or a typical Finnish winter day.
A driver, whose job is a bit more than just plodding up and down the motorway all day, will be by now all too familiar with their truck slamming its brakes for no reason. Sometimes several times per day. The most recent Mercedes Actros I drove last year in Finland had a dangerous habit of accelerating suddenly on its own over a bend. My suspicion is that its speed limit information was out of date. And remember, that’s a vehicle manufactured by the same company that claimed to have designed a fully self-driving vehicle 40 years ago!
Artificial intelligence will solve the problem
Proponents of AI will argue that those problems will be solved in no time if we allow the companies to gather more ‘real life’ data. ‘Let those cars drive around and they will learn it quickly’ – they say. There is a problem with this though. By agreeing, we’re basically allowing tech companies to make all of us their unwilling guinea pigs without sufficient government supervision. And we’re already dying because of it. Moreover, most of the companies never even asked our permission to take part in their tests.
And they’re not teaching their vehicles how to drive according to the rules. Tesla has, or at least had, an option where you could set your autopilot to drive a certain percentage over a speed limit. It has also emerged that Waymo cabs do not yield to pedestrians even when required by law – and it’s been suggested that their algorithms are set that way so they remain competitive on travel time with human drivers, who also often don’t do it. The company spokesperson denied it, but said it’s an “opportunity for continued improvement in these highly dynamic social interactions between man and machine”. And it’s great that the machines can learn lessons, but if they hit someone in the process would that be consolation for the person injured, or family bereaved?
Legalities and ethical conundrums
Tech companies claim that self-driving cars cause fewer crashes than human drivers. But independent research proves this claim to be untrue. There is evidence that the car companies underreport incidents or attempt to hid them from statistics.
Tesla, for example, switches off its self-driving system seconds before inevitable crash. Consequently, the computer log shows that the incident actually occurred when vehicle was driven under human control. They might be doing it to avoid the legal kerfuffle. Because it’s still unclear who’s responsible for an autonomous vehicle that breaks the highway code, or ends up causing the crash. Is it the vehicle owner, the vehicle manufacturer or the software provider?
And how do we solve the infamous trolley problem? Should a self-driving car decide who to crash into when faced with a group of people on the road? Prioritise kids over elderly, or vice versa? Or should it put the safety of the occupants – because they paid for it – over the safety of the third party? And if not, would you buy – or board – a vehicle knowing that its algorithms will chose to sacrifice your life over the lives of others?
Congestion and self-driving cars
But let’s address the elephant in the room. Contrary to what proponents of this new technology are trying to tell us, self-driving vehicles will not solve the problem of our cities being oversaturated with cars. Because in terms of congestion, if it gets stuck in traffic, it’s irrelevant whether your vehicle’s a two-stroke Trabant car or a self-driving electric pod.
In my opinion with self-driving technology the congestion will be even worse. Because the idea is that after dropping you off at your destination, the self-driving vehicle will drive away to pick up the next passenger.
Currently, for those who commute to work by car, you then leave the vehicle at work for the day before heading home. If our transport is to be based on a system of self-driving taxis or minibuses, they will continue to drive, heading for another pickup. And since the number of them will need to match the demand during peak commuting hours, for most of the time, many will have to be parked somewhere – most likely on the outskirts of town, where the land is cheaper – or they’ll just drive aimlessly in circles waiting for another fare. Either option will add miles to their journeys, causing noise and pollution from tires and brakes.
You might think I am a Luddite, but don’t take it from me. Listen to Elon Musk, a leading proponent of the idea that self-driving vehicles are the future of urban transport. During a recent event that premiered Tesla’s new CyberCab, he proudly pointed to a bit of road jammed with 20 vehicles driving by with no people in them. I rest my case.
Build a tram
Self-driving vehicles might find some niche use of course. This could be an interesting solution as a form of on-demand transport in rural areas. That is, if and when AI becomes sophisticated enough to safely tackle the issues of driving on narrow, winding rural roads in all weathers.
But apart from that specific case of low-density environment, why should we focus on automating driving buses with cutting edge unreliable technology? We have already a much simpler solution for keeping public transport vehicles on pre-determined route: two metal bars placed along the street.
Whatever the issue you’re trying to address – be it lack of drivers, accessibility, environment, safety or efficiency – tracked vehicles beat most of them. And automation of tracked vehicles is much, much less problematic. As any passenger of London’s DLR or model train enthusiast can attest.
So, if you want to move vast quantities of people, swiftly, in a busy town, just build a tram!
More in transport.
Like this article? Why not share it on Bluesky?
