Tesla’s model doesn’t work

Testing the Tesla autopilot (self driving mode)
Testing the Tesla autopilot (self driving mode).
It’s been revealed that the first person ever has been killed in a crash by a self-driving car:

The 7 May accident occurred in Williston, Florida, after the driver, Joshua Brown, 40, of Ohio put his Model S into Tesla’s autopilot mode, which is able to control the car during highway driving.

Against a bright spring sky, the car’s sensors system failed to distinguish a large white 18-wheel truck and trailer crossing the highway, Tesla said. The car attempted to drive full speed under the trailer, “with the bottom of the trailer impacting the windshield of the Model S”, Tesla said in a blog post.

In spite of this, I still believe self-driving cars will take over. However, it does highlight one fallacy, namely the idea that a human driver can be expected to supervise a near-perfect self-driving car.

Just think about it: If your car has been driving perfectly for a whole year, would you find it easy to keep your eyes glued to road and your hands to the steering wheel, just in case the car’s computer has a nervous breakdown? Wouldn’t you start playing with your smartphone, eat a sandwich or even doze off for ten minutes?

What this accident shows is that Google’s model (where the car is fully autonomous and the passengers don’t have access to a steering wheel) is correct, and Tesla’s is doomed. If a car is driving on its own, nobody should pretend that a human is ultimately in charge.

No normal person will ever own a self-driving car

Google prototype self-driving car
Google prototype self-driving car.
A few things I’ve read recently have convinced me that the average punter will never own a self-driving car.

The main reason for this is that they’re going to be significantly more expensive than an old-fashioned car, mainly because of all the sensors. As pointed out recently in the New York Times, “[a]dding self-driving technology — at least as it stands now — into regular passenger cars could make them absurdly expensive for anyone without the cash of a Silicon Valley mogul. Until recently, the laser sensor used on the Google car project cost $75,000 [£50,000].” Even though that price is clearly going to come down, it’ll always be more expensive to produce a self-driving car than an old-fashioned one.

The additional costs mean that they need to be used much more than normal cars in order to recoup the cost. HGV lorries might (as mentioned in the article linked to above) adopt the technology first, because it means a lorry can then be on the road 24/7 with only one driver, which mean that the additional cost will be recouped quickly.

Normal, old-fashioned cars are just not used enough to make it worthwhile to make them so expensive. According to the RAC, the “average car is parked at home for 80% of the time, parked elsewhere for 16% of the time and is only on the move for 4% of the time.”

Because of this, a self-driving car only really makes economic sense if it’s being used as a taxi, so it’s no surprise that Uber are very interested in this area — they already have put self-driving cars on the road in Pittsburgh.

Even if some crazy individual were to buy a self-driving car, it would be a bit silly to park it when they’re not using it rather than letting it make some money on its own working for Uber or similar. Only multi-millionaires will buy a self-driving car and then leave it in the driveway.

So we aren’t going to replace our old car with a self-driving model. Instead, we’ll simply start using self-driving taxis more and more until we don’t see the point in owning a car any more. It probably won’t be long before young people can’t see any point in getting a driving licence, but I imagine companies like Uber might need to introduce subscription services covering all you transportation needs for a fixed monthly fee in order to tempt current drivers to give up their car.

The move to self-driving cars is of course going to be bad news for the majority of car manufacturers. If normal people don’t buy cars, there is absolutely no reason to have so many brands and models to choose from. It’ll probably be more like the situation in the aviation industry, where companies buy hundreds of planes and then decide how they want them to look.

Self-driving cars

I just discovered that Google have realised some videos that really show how revolutionary their self-driving cars will be. Have a look at the blind gentleman in this one, for instance:

It just demonstrates that the current discussions in many countries (where politicians are still in favour of having a human driver in each car who is legally responsible) are awfully silly. What kind of court would hold a blind man responsible for an injury caused by his car? Also, what if there’s nobody in the car (it might for instance be parking itself)?

It’s interesting that reducing the number of traffic injuries seems to be one of the things that are really motivating Google:

Personally I’m really looking forward to getting a self-driving car. It’ll be great to be able to do useful or fun things instead of watching motorway traffic, and it’ll be wonderful to be able to take the car home from a pub or sending the car up to school to get the kids.

I do wonder how self-driving cars will be furnished after a few years. Will they look like old-fashioned train compartments, where the passengers face each other across a table? Or will they look more like a living room, with comfy couches and a telly in the corner? Or perhaps like a caravan, with convertible beds and other useful things?

Also, will people actually want to own self-driving cars or will they just use them as cheap taxis? That would have the advantage that you could send for very different models depending on your journey, e.g., a tiny office-like car for going to work, a bedroom-like car for going to the 8am meeting in London, or a car full of toys for collecting kids from nursery.

The more I think about it, the more I believe that once self-driving cars become ubiquitous, they will dramatically change the way we live.

Another way to build cities



A House on Stilts
Originally uploaded by Steve Dinn

Fitting cars into cities at the same time as houses and people seems to be a really hard task.

Has it ever been tried, I wonder, to create a new town or city by designing the roads and parking spaces on an empty field, and then put in pillars and build the houses and green spaces on top of these?

In this way, cities would appear to be entirely traffic-free — there would just be gardens and lawns with pretty paths filling up the space between the houses — while the roads would be very straight and efficient, and quite safe too because of the lack of pedestrians on the roads.

It’s possible that it’s more expensive than I imagine to build houses and gardens on stilts, but apart from that I cannot see any problems with my idea.

Has it ever been tried?

Hand luggage



easyJet Titles
Originally uploaded by WexDub

A few years ago, the low-cost airlines started charging for checking in suitcases in the hope that people would start travelling with hand luggage only.

They seem to have been too successful: When I travelled to Paris yesterday with EasyJet, they asked for volunteers to have their hand luggage checked in for free, and the terms & conditions threatened that they had the right to check in your hand luggage if the plane is too full.

However, given you’re only allowed one piece of hand luggage, it’s likely to contain valuables, such as passports, tickets and laptops, and I wouldn’t be a happy bunny at all if they chucked my computer into the hold.

Surely the whole system needs to be redesigned. As I’ve suggested before, one solution could be to abolish checked-in luggage completely, let passengers carry all their luggage out to the plane, and then let them put the big items into the hold themselves, just as people do when they’re travelling by coach.

Alternatively, redesign the planes, abolishing the hold and increasing the luggage space available in the cabin.

Surely even EasyJet and Ryanair can come up with something better than the status quo!

The way forward for Scotland’s airports and railways

There’s only about 60 km between Scotland’s two largest cities, Glasgow and Edinburgh (counting from city boundary to city boundary along the motorway), yet they have separate airports and it takes 50 minutes to take a train from city centre to city centre.

Looking further afar, Inverness is only about 180 km north of Glasgow as the crow flies, or about 270 km by road, yet the fastest train is under way for almost 3½ hours.

I could give similar figures for travel to the other Scottish cities, such as Aberdeen and Dundee.

This is ridiculous! It’s like all efforts go into providing good connexions to London, instead of tying Scotland closer together.

In an ideal world, I’d shut down Glasgow and Edinburgh airports and build a new one south of Falkirk (the exact location would of course have to depend to the geography). I’d then build some very straight rail tracks from Glasgow via the new airport to Edinburgh, so that the trains could achieve a decent speed (I’m imagining something like 15 minutes from either city centre to the airport, or about 30 minutes from Glasgow to Edinburgh).

Furthermore, I’d straighten out the tracks to at least Inverness and Aberdeen, add parallel tracks and electrify the whole lot, so that decent speeds could be achieved there, too. I’m not sure exactly what would be possible, but I reckon it should be possible to get the travel time from Inverness to Glasgow or Edinburgh down to under two hours, and hopefully close to one hour.

The effect would be that all Scottish cities would be within easy reach of each other, which would no doubt do wonders for the Scottish economy. It would also mean only one airport was needed for mainland Scotland, which would result in a big airport with lots of direct connexions, instead of just having small airports mainly sending passengers on to the larger hub airports such as Heathrow.

Besides, I’m sure a big infrastructure project such as this would be just what the doctor ordered against the recession…

Let the EU manage the railway networks

Europe rail electrification

Europe is a real mess when it comes to railway networks. The gauges aren’t the same, the ways they’ve been electrified vary (see the map on the right), and the signal systems aren’t the same.

To quote from Wikipedia:

While most railways use the standard gauge of 1435 mm, some countries, especially Spain and the former member states of the Soviet Union have widespread broad gauge tracks (1,520 mm). Likewise, electrification of lines varies between countries. 15 kV AC has been used in Germany, Austria, Switzerland, Norway and Sweden since 1912, while the Netherlands uses 1500 V DC, France uses 1500 V DC and 25 kV AC, and so on. All this makes the construction of truly pan-European vehicles a challenging task […]

It probably would be prohibitively expensive to unify everything, but rational decisions need to be taken, and that can’t happen on a national basis. For instance, Danish politicians are currently discussing whether to electrify anything other than the route from Copenhagen over Funen to the German border; this is obviously a no-brainer on a European scale, but to a small country the cost can still be prohibitive.

At the moment, efforts are concentrated on the high-speed lines, but wouldn’t it be more efficient to hand over the entire railway networks to the EU? I’m not saying EU politicians should necessarily unify everything, but they could at least change situations where one country is the odd man out.

Actually running trains could still be handled/outsourced by the individual countries, but it would be so much easier to create attractive franchises if a train in most cases could run directly from any point in the EU to another other point, and if trains purchased for use in other country could just be used in another.