Autonomous service vehicles for mining industries and ports are being developed by the big truck manufacturers and are far more likely to succeed in the near term (less for the AI to handle, private land so fewer regulatory hurdles), but they just don't get as much hype (positive and negative) as the idea of revolutionising our commute.

Rio Tinto, the world's second largest mining company, already has autonomous trucks and trains.

They've been at it for some time now[0]. Even with a massive budget and highly controlled environments (trains don't have to follow lanes, after all) it's proved to be quite difficult.

[0] http://www.riotinto.com/australia/pilbara/mine-of-the-future...

Trucks driving long-haul routes. That would be one of the easiest things to automate. Not even the pickup or delivery at terminals, just going over the highway.

Probably there is lots of work in this area and it will happen first. It just doesn’t get much press because it isn’t consumer facing and it’s boring.

The challenge with going big is that the risks and costs raise substantially, while the price of having a local expert to manage both the drive and any eventualities (ie a human driver), isn't changing much...

Long-haul routes are important, and shaving minutes off deliveries is a big deal. Improved driver logistics and routing are probably wildly more profitable than automation (and risks of highly visible automation failures).

After all some trucking companies can sometimes even get away with not paying their drivers because the drivers didn't drive enough miles to cover the truck lease this week and therefore go home with a bill rather than a paycheck.

1. VC prefer technical risk to market risk. Industrial is riskier.

2. The hype around ML has turned it into magic, so why not “disrupt” a huge market?

Imho a lot of this is the direct result of the deep learning hype cycle.

>Imho a lot of this is the direct result of the deep learning hype cycle.

I'm right there with you. The first round was just 'web connected'. Uber - we're going to do taxis, but with the internet. For example.

Now we've moved to stage 2 - machine learning. Auto-Uber - we're going to do taxis, and uber, but with machines!

It makes for nice headlines.

I can't wait for blockchain ride-sharing.

Never mind https://www.iride.io/

almost.. like trains?

Almost, but on a road and not on a rail. Also a little bit like cars, but these ones drive themselves. A bit like a tram, but with the ability to move freely and not be guided by a rail. Not at all like a bar of soap. Somewhat like a bicycle, but bigger and enclosed.

Like trains with Automatic Train Control. Unfortunately, it's hard enough to get our train operators to install Positive Train Control, never mind Automatic Train Control.

True, but there's also the logistics around getting their goods from a factory to a train. The assumption around automated deliveries would be that the entire chain is automated - from creation to lorry to destination.

Curious have you ever been on one? Not sure how fast or slow they are for cargo but I have taken a train from Orlando to Hollywood (near the Miami area) approx a 3 hour drive by car. On a good day its a 6 hour train ride, I think I have been on there for 10 hours before. Some people for waaay longer from further up north.

I rather be on the road but a train was cheaper than spending money on tolls plus gas. Saving money then was a necessity for me.

You speak about trains as if they are some incredibly exotic things? Maybe they are in US, I don't know. But at least UK trains are almost always faster than driving. I could get to London by train in like 2:50h, or 5 hours of driving(and that's just to outskirts of London, not the city centre). Or to Edinburgh in 1:20h by train, or 3 hours of driving. Not to mention proper European trains, where taking a TGV from Paris to say Brussels, Amsterdam or Berlin is in no way or form comparable to driving - it's just much much faster.

Now the cost is an issue - in the UK driving is almost invariably cheaper than taking the train, even when going solo. When taking any passengers with you driving will win every single time, sometimes it's even worth renting a car for the journey and it will still be cheaper than the train.

I can't believe that "have you ever been on a train" is a real question.

Depending on the route, I could see those numbers being reversed in the UK and Europe; intercity trains are fast but pricey, even allowing for the much higher fossil fuel taxes.

Commuter trains are a similar story; while it's expensive and unpleasant to commute into central London by train, doing it by car will be even slower, incur the congestion charge, and there's nowhere to park.

The price various a lot.

- In Belgium you can ride from anywhere in the country to anywhere else for 7.5€ (if you buy a prepaid ticket for 10 rides), which is hard to beat even with a car with great mileage.

- In Portugal, you can go from Lisbon to Porto for 15€ (buying a couple weeks in advance). By car, that won't even cover the tolls.

I commute into central Edinburgh from Fife - I wouldn't say it expensive (£12 return) and it's pleasant journey (maybe once a month I have to stand for 15 mins on the way home). The equivalent journey by car would be much more expensive due to parking costs, take much longer and, in my view, be rather unpleasant.

I like having a car - but not for commuting into the center of a city!

Why?

In the US, I guarantee you that the vast majority of people have never been on a train. In a few select cities the number will drop if you include subways and other commuter trains, but the general conclusion is the same.

I have taken a train from Berlin to Saarbrücken, approx an eight hour drive. On a good day it's a 6:30h train ride. For Berlin-Munich it's seven hours in a car or 4:30 in a train. For Tokyo-Kyoto it's six hours by car or 2:30 by train.

The US rail system privileges freight over passengers, which is a big part of what makes it (by world standards) a slow system to ride.

Money? Maybe people are more likely to pay.

It doesn't sound sexy, therefore not good for publicity thus making it harder to get funding.

A less cynical idea would be that the problems are less interesting and easier, so the best people may be reluctant to work in that field?

I am absolutely sure companies & large organisations are far more likely to pay AND are willing to pay larger amounts, since there are actual cost-savings when you're able to replace a human with an AI driver