Under the hood of self-driving cars
The recent start of the pilot programme to test self-driving cars on UK roads is excellent news for road safety, convenience, social inclusion and ultimately quality of life. Rarely does a new technology offer such a large set of benefits with so few potential drawbacks. The pilot is expected to last 18-36 months.
Self-driving cars or pods will bring some new dangers and some new legal and insurance problems that need to be solved. The key thing is that they can be solved, and a pilot is a good time to check the issues through to find the best solutions. A few esoteric issues are thrown up by the AI aspects, such as what the car’s priorities should be in the event of an unavoidable accident – e.g should it swerve off the road to avoid a toddler chasing a ball even if that means killing an old man on the pavement? Who gets to decide the programming?
Every new technology throw up a few issues, and being able to imagine a once-in-a-lifetime potential problem can’t be a good excuse to avoid progress, especially since that same problem affects people even worse because people can’t get the facts and analyse them as quickly.
In any case, self-driving cars could talk to each other at computer speeds. In the microsecond it should take for two well-designed car management systems to exchange messages to discuss their actions and coordinate braking, a car travelling even at 70mph would only have travelled about 30 microns.
That means that self-driving cars can safely travel millimetres apart at 70mph without ever colliding with the car in front, and with a gap that small, it is hard to imagine a toddler getting in between. Even if they do collide, the speed differential that can build up in a microsecond is tiny, so passengers wouldn’t even feel a bump, and certainly no damage would be done.
A child in front of the first car in a road train could be in danger, but cars would be watching full time in every direction, so they would have seen the ball’s trajectory and could all have been braking for seconds before a child even approaches the edge of the pavement. An accident is possible, but far less likely and less severe than if people were driving. In all likelihood they would be stationary by the time the notional toddler is in front of them.
Driving so close together could almost eliminate drag if the cars were a standard size and box-shaped, which also maximises their capacity. That means we could get up to 15 times more cars on the same stretch of road, and they could be going faster, reducing both congestion and travel time. We won’t get such gains while they share the roads with human drivers, but that is an excellent reason to roll such vehicles out in towns and exclude human-driven vehicles from the urban area. The faster the roll-out, the faster we’d see the benefits.
Self-driving cars could find their own way to charging or batteries replacement points, but with developing technology, there is every reason to think we could simply power them as they go, like Scalextric. Super-capacitor banks could be charged by inductive mats embedded periodically in the road surface.
Small batteries could hold enough charge to get to and from the main roads. This sort of solution would be ideal for the many cars that are only used in urban areas. It would make lithium shortages far less of an issue too. We’d simply need far less battery capacity if cars are directly powered by the road surface.
Another major advantage of self-driving cars is social inclusion. A frail elderly person could be picked up at home and dropped at their destination instead of having to walk a kilometre to the nearest bus stop and wait twenty minutes for a bus in the wind and rain. They would be able to take part in society again, with ongoing benefits from loneliness reduction, improved mental health, community care, political inclusivity etc.
It is very hard to see how buses and taxis could compete on cost, convenience, speed or safety with future public transport based on self-driving cars. Fleets of cheap box-like cars would offer the best possible public transport solution.
If such a system were available, few people would bother buying their own car. Paying a tiny fraction of the costs of a publicly owned car and getting end to end convenience, no parking worries, and the freedom to party without worrying about drink-driving would present the new problem of how to spend the thousands of pounds a year saved and how to spend the extra time freed up. People would soon find new ways of showing off their status.
As for losing the joys of driving, those have already been evaporating for many years.
It isn’t good news for everyone of course. We would need far fewer taxis driven by people, just as we already need far fewer people carrying flags in front of cars. But the rollout won’t happen overnight and many of those taxi drivers would be retiring in the same timeframe anyway, so only some would have to retrain.
Car manufacturing would also change. In place of the many varieties of cars, a few models could meet pretty much all needs so a few manufacturers would likely win almost all of the market. The cars would likely be owned by large fleet management companies, which might well be affiliated with energy suppliers, using car batteries as storage for renewable energy.
With vastly improved safety, self-driving cars could be made very lightweight, with cheaper materials. There would be environmental benefits both in resources and energy use. That would come at the cost of the lower sales for producers of those materials, but as for the workers in production lines of the cars made obsolete, they would likely have been automated in the same timeframe anyway. Manufacturing is changing rapidly with or without self-driving cars and no worker in any manufacturing sector should expect to have a job for life.
Many of the uses of self-driving pods revolve around delivery and distribution, where there is no passenger. This could have a powerful stimulus effect on cloud-based manufacturing, cottage industries and small business. It also raises the negative prospect of criminal and terrorist abuse, transporting bombs, dangerous packages or drugs.
We will therefore need to develop sensors for pods used for distribution to detect explosive devices or drugs and to raise alarms (and remain where they are) if any are detected. In other mischief, relatively harmless demonstrators could cause road blocks by all asking for cars at the same time, the road equivalent of denial of service attacks.
Security systems should be developed to ensure that the identity of bookers and passengers are known so that anyone requesting or using a car can be tracked down, a strong deterrent. However, this would cause privacy concerns, so a balance needs to be struck. Finally, hackers could hijack and misdirect cars, possibly even cause accidents, if systems are not designed properly.
This is not a new concern and affects some existing cars. Pilots won’t encounter all of these issues, but they should be obvious to any engineering company, so we should expect self-driving car manufacturers to develop appropriate solutions and counter-measures for any such abuses.
The safety advantages far outweigh these addressable problems. Pods would not drink or take drugs or be careless or reckless. They would look in all directions all of the time and react in microseconds. They could drive millimetres apart, reducing congestion and wind resistance at the same time as preventing children running between them. Less obvious advantages arise from lower journey times to hospital, and greater social inclusiveness leading to increased fitness and mental health.
I read a lot of articles on the drawbacks of self-driving cars. Having read them and thinking them through as an engineer, I don’t see any engineering issues that can’t be solved, and redundancy issues arise with any technology change. Most of them will happen anyway.
What is left is a win-win-win situation. If we do our best to get self-driving cars working well quickly, and roll them out as fast as possible, the benefits will very far outweigh the problems. I’ve never seen such a good benefit to cost ratio for any class of new technology. Done properly, it could be one of the greatest improvements to quality of life in the century.