Driverless cars in UK this year

[AlaninWales]

According to NS: https://www.newscientist.com/article/mg23030732-600-london-to-see-fleet-of-driverless-cars-on-public-roads-this-year/.
Anyone in the area planning to test the cycle-friendly nature of these pods?

[mercalia]

I thought we already had cars with no drivers ( to speak of)

[Richard D]

Swap to a car with an electronic brain and a manufacturer who will be rightly terrified of product liability claims if the software isn't *perfect*, or stick to a car driven by a moron who spends a lot of time staring at what passes for his electronic brain hoping for a Facebook status update?

I know which I'd prefer to share the roads with.

[Vantage]

Swap to a car with an electronic brain and a manufacturer who will be rightly terrified of product liability claims if the software isn't *perfect*, or stick to a car driven by a moron who spends a lot of time staring at what passes for his electronic brain hoping for a Facebook status update?

I know which I'd prefer to share the roads with.

+1

[ANTONISH]

A few weeks back there was a report of a driverless car in America colliding with a bus. The report said that the driverless car was coming out of a junction at 2mph and the bus (which had right of way) was travelling at 15mph - obviously the driverless car hadn't detected the bus.
Substitute a cyclist for the bus and you would have a familiar scenario.

[Bmblbzzz]

The New Scientist article is essentially a reannouncement of the pods that were announced last year. Nevertheless, it contains some interesting information.

Firstly, we should note that what's announced for the UK is not "cars" but "pods" which will only operate on certain routes. This is much more limited in scope than what Google and others are doing, which is essentially making a car drive itself everywhere. In some ways though the pods are more advanced -- it seems they will be totally self-driving, with no human in charge -- or maybe (more likely) there will be but it's not mentioned. The pods are a step further away from cars in that they don't look like cars. As the "Interior Design" sub-story expands on, self-driving cars will be more like mobile offices/living rooms/hotels/etc than cars; I've thought this to be likely for some time, it's good to see NS saying the same!

Secondly, the terms "chauffeur" and "guardian angel", while perhaps not quite satisfactory linguistically, are useful labels to hang on the processes. "Guardian angel" technology has been creeping into cars for a long time, of course, some obvious examples being anti-lock brakes, parking sensors and those gadgets which turn your lights and wipers on without you thinking about it. It's not such a long step from there to a car which can brake and steer for you in an emergency.

The safety fears over self-driving cars are groundless IMO. There will still be collisions for various reasons but far fewer. The real disaster of self-driving vehicles will be social. With driving made easier, less stressful, cheaper, more comfortable and quicker (intelligent vehicles will cause fewer jams, by talking to each other, negotiating priority at junctions, and not doing stupid things like stopping on a crossroads), it will be even more attractive to even more people. But self-driving cars will not reduce the effects of isolation, both on vehicle occupants and on those marooned the wrong side of a busy road. Urban planning will be more important than ever. Add to this the effects on a whole sector of employment -- lorry drivers, bus drivers, taxi drivers, pizza delivery scooters -- and think that driving will no longer be an excuse not to be working, and we're in for some upheaval.

[AlaninWales]

A few weeks back there was a report of a driverless car in America colliding with a bus. The report said that the driverless car was coming out of a junction at 2mph and the bus (which had right of way) was travelling at 15mph - obviously the driverless car hadn't detected the bus.
Substitute a cyclist for the bus and you would have a familiar scenario.

No, the car was rejoining a lane after (needlessly) avoiding minor road debris. The software which made the mistake was then updated in all google cars so it won't be repeated. A pity we cannot do that with human drivers, because it would mean that we wouldn't repeatedly have the same mistakes killing cyclists.

[Pete Owens]

And - the reason it was noteworthy is that this was the very first example of a collision in which a driverless car was deemed at fault. After millions of miles of testing ALL the other crashes they have been involved in have been caused by human drivers.

[Postboxer]

It would be easy to bully driverless cars though, just step out in front of them to cross the road, safe in the knowledge that they'll stop for you. This may also cause problems when some cars have a computer in control, some have a human, some drivers may take advantage, assuming the driverless cars will be easily identifiable.

[Pete Owens]

Why on earth would you consider it to be a problem that pedestrians could cross the road without fearing being mown down?

[Postboxer]

I'm talking about people stepping out so the driverless car does an emergency stop, the occupants might spill their cornflakes, or hot drinks, then be rear ended by the driver behind. Kids might play chicken more. A google car got confused by a cyclist doing a trackstand and wouldn't move.

[AlaninWales]

I'm talking about people stepping out so the driverless car does an emergency stop, the occupants might spill their cornflakes, or hot drinks, then be rear ended by the driver behind. Kids might play chicken more. A google car got confused by a cyclist doing a trackstand and wouldn't move.

Excellent: The google car did exactly what I do if a cyclist to does a trackstand in front of me: I have no idea if (s)he is going to fall over or accelerate in an unpredictable direction and will wait whilst (s)he sorts himself out. I have also been known to do an emergency stop when kids decided to jump out in front of me. So a google car does the same = good news, better than the 'cars rule the road' attitude of many human drivers.

[Bmblbzzz]

Computer-driven cars will not (do not! cannot!) defy the laws of physics. They still need time to come to a stop. What they will usually be able to do is detect the presence and movement of people, animals and objects even when they are not visible (because they will use multiple senses -- not just vision) and predict and prepare for their actions. So playing chicken won't be such fun because actually the robocar will be altering its course and speed before you step out.

[Pete Owens]

I'm talking about people stepping out so the driverless car does an emergency stop, the occupants might spill their cornflakes, or hot drinks,

Well I suppose it comes down to whether you consider spilt cornflakes to be more important than a childs life.

[Vantage]

Re playing chicken with driverless vehicles. Technology isn't infallible.
A number of years ago, a bloke invented a tablesaw that detected a human hand and stopped spinning before the users hand was chopped off. This doesn't mean you go sticking your fingers in to see if it works...your fingers won't magically grow back if it doesn't. A bit like playing chicken with a robot car that can't possibly go wrong. If it does, it can carry on. Unlike a dead person.