Inconvenient truths

There is a prevailing view that in 20 years’ time no one will own a car and we’ll all be travelling around in autonomous vehicles with our feet up reading the paper. The streets will be safer, cleaner, quieter and less congested and we will be happier, richer and more in control.

While it’s relatively easy to imagine autonomous vehicles operating on motorways, it is a much bigger step to self-driving cars in busy and chaotic inner city environments. And it’s a reasonable assumption that new mobility services will be introduced first in big cities, but what about sparsely populated rural areas?

Advocates tend to talk about shared ownership as an ideal way forward, but there might be situations where private ownership is still the best option and others where it makes sense to share. The trends of electric, self-driving and shared are significant, but they will work for different people in completely different contexts and they won’t all work for everybody. We are not looking to a future world in which one size fits all.

Every IT system fails. Banking systems go down, Twitter crashes. Either through cyber-attack or simply as a result of a design glitch, anybody who thinks autonomous vehicles can be designed to be completely failure proof is naïve. And if you take that as the starting point you need to think through the consequences.

The automotive sector has seen increasingly sophisticated technologies entering the mainstream for the last 10 years; parking aids, lane control and cruise control are to varying degrees standard.

It is hardly a point of conjecture that cars will get more sophisticated to the point where they are virtually self-driving, at least in some situations such as motorway driving. Whether we get to 100% or end up at say 95% is largely a matter of legal responsibility.

Should there be an able and qualified driver in-car as a fail-safe? If so that will undermine the arguments about increasing mobility among the elderly and disabled people. And if we have totally eliminated drivers in day to day use, would someone who might not have been in charge of a car for years or even decades be capable of driving in the event that they need to take control?

If all vehicles are automated and talking to each other and the infrastructure, the technical challenges are more containable, but it takes about 30 years to renew a national vehicle fleet and there will always be exceptions. What happens in the transition and could it ever be politically acceptable to ban an old car from the road?

And the technology isn’t cheap. Development costs for computer chips used in cars are massively higher than those used in a phone, because you can’t reboot your car at 120km/h. Resilience and redundancy have to be much better in the automotive sector and designing chips for autonomous vehicles adds at least another order of magnitude to the development costs.

It is no coincidence that there have been lithium ion battery fires in phones and laptops due to design faults, but never in a car. So far, car companies have always got their battery installations right, because they are terrified of the alternative.

The cost of technology will become more affordable with time, but it’s relevant that manufacturers have not yet come up with an electric powertrain that is cheap enough to put in budget cars. With the advent of autonomous vehicles how long will it take for manufacturers to be able to offer full blown self-driving in an affordable entry level model?

And from a marketing and sales perspective would a car manufacturer want to do that? Could marketing strategies slow the universal roll out of connected and autonomous car technology? Hard to imagine that in the early 1980s the entertainment system on an entry level BMW was a radio, you had to pay more if you wanted a cassette player too.

A lot of the development work and trials of self-driving vehicles has taken place in the US. Significantly there aren’t many roundabouts in the US and it is easier to imagine self-driving vehicles working in US cities where roads are typically laid out on a grid and crossroads are managed by traffic lights.

But autonomous vehicles will have to be capable of safely negotiating roundabouts in other parts of the world. With humans in charge of the controls the safe navigation of a roundabout requires a high level of driver to driver visual communication and currently we don’t know whether technology is capable of replicating the digital equivalent of a nod and a wave.

If that proves too nuanced, a possible consequence might be the need to fundamentally rethink how road systems are laid out and how they function. It could even be argued that the emergence of autonomous vehicles is an opportunity to rationalise road infrastructure by removing redundant and ineffective ‘legacy’ infrastructure elements such as roundabouts and even white lines.

A key point is the possible inability of autonomous vehicles to deal with roundabouts is not in itself a valid reason why autonomous vehicles can’t be successful. Of more concern is how to deal with the transition period when autonomous and traditional vehicles will share the same road space.

Society accepts a road accident caused by driver error much more so than a train crash. And in the event of a bad car accident, investigations are carried out to identify who was at fault which might result in somebody being prosecuted.

But with autonomous vehicles, someone will have coded the algorithms that determine how the car responds. A horrible accident resulting from a split-second decision in the heat of the moment is legally and morally different to someone sitting in an office writing computer code and deciding systematically what the outcome will be in a particular circumstance. Does the car kill the child on the pavement to save its occupants?

Unless we get to complete artificial intelligence there will be a degree of pre-programming and someone will have had to have made that value judgement. It is much easier to regulate around driver error than a predetermined outcome coded into the software.

Private cars are used on average less than 5% of the time and the argument runs that if we moved to some kind of shared ownership model such as mobility as a service, it would reduce the number of vehicles on the road and so congestion. However, while most cars are not used most of the time, they are largely used at similar times.

Unless we fundamentally rethink about how we use cars in our daily lives we are still going to need a large number of vehicles. Mobility as a service and self-driving cars will not necessarily solve congestion and by extending access to personal mobility could actually increase the amount we travel on roads and undermine the commercial basis for mass transit.

You’ve also got to think what happens to those shared, self-driving vehicles in the periods when they are not being used. The main peak periods will be as people come to and leave from work. In between demand will die down and the vehicles will have to go somewhere, some will pick up additional trade but others will need a place to park up until they are needed eight hours later – similarly where are they going to go at night? Nobody wants empty vehicles cruising the streets looking for trade.

There will still be a need for a chunk of land for storage and places where vehicles can be refuelled or recharged, cleaned and maintained. It could certainly be in less economically productive places, perhaps close to an electrical substation.

The point is a move to shared, autonomous vehicles can’t be considered in isolation to land use planning. That could lead to improvements in the urban environment and we certainly want to remove the need to continue to cover our towns and cities in tarmac, but it doesn’t necessarily follow as some are suggesting.

There are subtle things, car seats being one of them, that people just don’t want to think about. People with small children and babies are often very particular about how they secure their offspring within cars. Even if shared cars came with integrated folding child seats (or carried one in the boot), would a parent be happy using something they were unfamiliar with especially when they don’t know its history.

Arguably because connected and autonomous vehicles are expected to be much safer, parental attitudes might shift. Child seats (designed specifically to protect the child) were not invented until the 1960s (the same decade that seat belts for adults became standard equipment) and parents today are happy to let their babies and toddlers sit on their laps in an aeroplane. However the plane analogy is subtle. If a plane decelerates rapidly everyone on board is probably dead, but if a car decelerates rapidly, adults are resilient enough to survive if they are strapped in but children need to be secured in a specific way to survive.

And beyond the contentious issue of child seats, people leave all sorts of things in the boots of their cars from golf clubs to wheel chairs. For those people, choosing to use a shared vehicle might be impractical or simply too much of an ask.

While several car manufacturers have developed vehicles that can pilot themselves with an ability unimaginable even a decade ago, a recent article in Fortune magazine lifted the lid on the industry hype that self-driving cars are poised to become increasingly common on our roads within a few years. It concluded after months of interviews with the people shaping the self-driving car industry it's clear that our autonomous future—the one where you take a nap as your vehicle whisks you to your destination in comfort and safety—is not in any real sense here now, nor around the corner, but likely decades away. All claims to the contrary are either based on misunderstanding or are intentionally misleading.