Why waste time and resources building driverless cars?

[Pan Pan Pan]

if any transport system would benefit from fully autonomous operation it is the railways. Trains run on tracks, so there is no requirement for steering input. They also run from the fixed points of stations., not from random variable locations. Trains ideally should leave a location at one fixed/known point in time, and arrive at another location at a known fixed point in time.
People using trains already know that in order to catch a train, leaving at a known/fixed point in time, the must be at a station at a known/fixed point in time.
Go into any fully automated factory, and they operate a series of tracks on which the product arrives at certain stations, at precise times. The only difference is that instead of jars or packs of product, in the case of railways the people are the product which want and would like to arrive at various stations at prescribed times. Since most railways are electrified drivers could be got rid of, possibly achieving an increase in safety, punctuality, and overall efficiency.
Trying to make cars operate autonomously on the other hand is like trying to herd cats. which want to go in all directions to thousands of different locations, at all different times, and at all different speeds. Trains benefit from a rigidity in the system, whereas the strong benefit for cars lies in their flexibility of operation

[Dingu]

Vast simplification.

This assumes humans cause a lot of the lack of punctuality rather than line capacity and general reliability. You would also need back up to the automation for when it goes wrong so would still need a trained person of some sort even if just to manage situations that arise.
The ROI is probably minimal if there is any return whatsoever.

[98elise]

Pan Pan Pan said:

if any transport system would benefit from fully autonomous operation it is the railways. Trains run on tracks, so there is no requirement for steering input. They also run from the fixed points of stations., not from random variable locations. Trains ideally should leave a location at one fixed/known point in time, and arrive at another location at a known fixed point in time.
People using trains already know that in order to catch a train, leaving at a known/fixed point in time, the must be at a station at a known/fixed point in time.
Go into any fully automated factory, and they operate a series of tracks on which the product arrives at certain stations, at precise times. The only difference is that instead of jars or packs of product, in the case of railways the people are the product which want and would like to arrive at various stations at prescribed times. Since most railways are electrified drivers could be got rid of, possibly achieving an increase in safety, punctuality, and overall efficiency.
Trying to make cars operate autonomously on the other hand is like trying to herd cats. which want to go in all directions to thousands of different locations, at all different times, and at all different speeds. Trains benefit from a rigidity in the system, whereas the strong benefit for cars lies in their flexibility of operation

Humans manage it without rigid timetables and rails to run on. Who is herding us?

Most of what it takes to drive a car can already be done better then humans. Knowing the rules, controlling the car, sensing the world around etc.

The harder bit is accurately processing what's being sensed, and what to do when something new happens. It will come though, it's only a matter of time.

The ROI for the manufacturer is selling something people will buy. For the vast majority of people driving is a chore. Even for enthusiasts there are times you would rather not be driving.

I'm driving to Wales next week. Most of it will be on cruise control, so I would love to let the car do the rest of the work.

[Condi]

The OP is correct that driverless trains would be much easier (the new London Underground trains are designed to be operated without drivers at some point in the future - I think even from day 1 the driver is a "watchkeeper" rather than a driver), but that doesn't mean that driverless cars would be a waste of resources. They can both co-exist together.

[Gecko1978]

We are on the cusp of a systemic change. AI and robotics really are likely to be science fact in my lifetime (44 currently). So what does this mean well at a guess:

1. Driverless cars
2. Driverless Buses
3. Driverless delivery vehicles
4. Driverless trains
5. Driverless waste collection along with robotic bin men
6. Call centers replaced by AI
7. Basic medical diagnosis done by AI
8. Complex medical diagnosis supported by AI
9. 1000s of low skilled jobs in finance and IT gone
10. Fully automated warehouses
11. Education provided by AI
12. Fully robotic supermarkets (if they even exist 30 years from now).

Lots of jobs just won't need replacing. I am sure MPs councils the Lords will be fine of course but its likely in 30 years time the next super rich will be those who made AI and robotics work day to day.

A bit like EV v ICE I know EVs are better in 99% of driving but I just don't like them as much even the Rimac etc. Robots are the same and AI but course 99% why will I care if I am speaking to a person in a csll center of clever AI

[valiant]

How are we defining ‘driverless’ here?

It’s one thing that can keep up with traffic on a motorway without intervention and it’s completely another to summon your car whilst pissed to take you to your front door. One is practically here already, the other? Probably many decades away.

As for trains? Well, it’s possible to an extent but how much are you willing to pay as a taxpayer to fund it? The cost would make HS2 look like purchasing a Hornby set in comparison and NR are already scaling back projects and warning of increasing delays as infrastructure of the existing railway is falling apart.

Either true driverless option is so far away it’s practically meaningless.

[durbster]

Yeah I think I agree to an extent.

I have reluctantly been convinced that public transport is the future.

Looking at it objectively, having everyone moving about in cars all the time makes absolutely no sense, especially when it comes to getting people to and from work. There are simply too many cars and building and maintaining a road network to support them is massively expensive, consumes loads of resources and the whole system takes up an enormous amount of space compared to trains, trams and buses.

I was suckered in by Elon Musk essentially lying about fully autonomous cars a few years ago but now I think they're a long way off. The technology simply isn't there.

However, driverless public transport that follows a consistent route does seem significantly more plausible, and is perhaps where the investment should be going.

I do love cars but the evidence is pretty clear, annoyingly.

[otolith]

Because eventually they will be able to sell them for a profit to people who want them. That’s how capitalism works.

[IJWS15]

With a driverless car I don’t have to get to the station and I don’t have to share the space with the great unwashed.

[alangla]

Self driving buses are already here in a limited way https://www.stagecoachbus.com/promos-and-offers/ea...
Though they still carry someone in the driver’s seat who can take over control.

[Gecko1978]

durbster said:

Yeah I think I agree to an extent.

I have reluctantly been convinced that public transport is the future.

Looking at it objectively, having everyone moving about in cars all the time makes absolutely no sense, especially when it comes to getting people to and from work. There are simply too many cars and building and maintaining a road network to support them is massively expensive, consumes loads of resources and the whole system takes up an enormous amount of space compared to trains, trams and buses.

I was suckered in by Elon Musk essentially lying about fully autonomous cars a few years ago but now I think they're a long way off. The technology simply isn't there.

However, driverless public transport that follows a consistent route does seem significantly more plausible, and is perhaps where the investment should be going.

I do love cars but the evidence is pretty clear, annoyingly.

Year 2000 few cars had parking sensors or cameras touch screens were still a thing of the future the internet was 256k and ordering off a website exciting and cool. 23 years later we all carry a phone way more powerful than our old pc. A phone is now a music device a camera a TV a banking app and health app all in one. Most cars have parking sensors and cameras many can park themselves many can provide basic lane keep assist you can talk to your car you touch a multi functional screen etc. 20 years from now things will look much different again and AI will play a part

[Condi]

otolith said:

Because eventually they will be able to sell them for a profit to people who want them. That’s how capitalism works.

Save the cost of all the drivers on all the trains and how much do you make? London Underground drivers are paid £60k a year on average, which is probably £100k for the employer after tax and NI and pension etc. There must be thousands of them to run the Underground, and that's only one transport system.

The benefits are £10's if not £100's of millions per year. It also allows trains to run closer together, which is more beneficial on overground networks, but if every train knows where every other train is and what it's doing then all you need is a minimum braking distance, if one stops the next one knows instantly and stops too.

[otolith]

Condi said:

otolith said:

Because eventually they will be able to sell them for a profit to people who want them. That’s how capitalism works.

Save the cost of all the drivers on all the trains and how much do you make? London Underground drivers are paid £60k a year on average, which is probably £100k for the employer after tax and NI and pension etc. There must be thousands of them to run the Underground, and that's only one transport system.

The benefits are £10's if not £100's of millions per year. It also allows trains to run closer together, which is more beneficial on overground networks, but if every train knows where every other train is and what it's doing then all you need is a minimum braking distance, if one stops the next one knows instantly and stops too.

Sounds like you have the start of a business plan there, have you found any investors yet?

[Condi]

otolith said:

Sounds like you have the start of a business plan there, have you found any investors yet?

In the 49 mins between my post and yours, on a sunny Saturday afternoon?

No, not yet. You're welcome to be the first, just ask for my bank details. Minimum investment £5k. Capital at risk. Past performance is no indicator of future returns.

[Mark V GTD]

I’m game! ☺️

[otolith]

If you ask anyone involved in the rail industry, automating their job is far more difficult than any outsider could possibly imagine, practically impossible. I suspect that the biggest hurdle in the UK would be the staff and unions. But in terms of why car makers are wasting their money automating cars when it would be easier to automate railways, I think it’s probably because having an automated car to sell looks like a better investment than selling something to make cars less competitive in the transport market.

[Ashfordian]

It is simply a ROI decision.

There are 24k trains in the UK.

For comparison there are:

33m cars
4.6m vans
540k lorries
150k buses and coaches

Now where are you going to focus your R&D based on those numbers?

Automating road transport(lorries alone) provides way more riches than automating trains. Train automation will happen, as a spin off of road automation, if trains survive road automation, which I doubt they will.

[mikey_b]

Ashfordian said:

It is simply a ROI decision.

There are 24k trains in the UK.

For comparison there are:

33m cars
4.6m vans
540k lorries
150k buses and coaches

Now where are you going to focus your R&D based on those numbers?

Automating road transport(lorries alone) provides way more riches than automating trains. Train automation will happen, as a spin off of road automation, if trains survive road automation, which I doubt they will.

24k trains? That seems an awful lot. What are you calling 'a train'?

[2xChevrons]

mikey_b said:

Ashfordian said:

It is simply a ROI decision.

There are 24k trains in the UK.

For comparison there are:

33m cars
4.6m vans
540k lorries
150k buses and coaches

Now where are you going to focus your R&D based on those numbers?

Automating road transport(lorries alone) provides way more riches than automating trains. Train automation will happen, as a spin off of road automation, if trains survive road automation, which I doubt they will.

24k trains? That seems an awful lot. What are you calling 'a train'?

Indeed - the ORR data I can find says that there are 4726 multiple units and locomotives (whereby, say, a DMU formed of three cars counts as a single unit, and HST power cars count as one 'locomotive' each) in service as of the end of 2022 and 567 carriages. plus freight wagons. Perhaps if every MU car and every carriage and wagon was counted you'd get to 24k items of rolling stock, but I can't see why every single one would need adapting for driverless operation.

Whether driverless operation is remotely practical in the present or foreseeable future on a national heavy-rail network (rather than isolated systems or specialised mass transit networks) is quite another matter.

[Pan Pan Pan]

Dingu said:

Vast simplification.

This assumes humans cause a lot of the lack of punctuality rather than line capacity and general reliability. You would also need back up to the automation for when it goes wrong so would still need a trained person of some sort even if just to manage situations that arise.
The ROI is probably minimal if there is any return whatsoever.

You seem to be (Deliberately?) missing the point.
Trains dont go ` all over the place', they run on rails.
They dont go to variable locations, they start at stations, and go to specific `other' stations.
They dont generally have to contend with other trains, other vehicles, people, or animals wandering onto their track, involving swerving or even continuously stopping and starting.
They are (allegedly) supposed to leave a station at a specific time, and arrive at their destination at a specific time.
It is the responsibility of those wishing to use a train, to arrive at a station at a specific time in order to catch a train leaving at a given time. Trains dont wait for a passenger who is late, If a passenger is late, they miss the train, it is that simple.
There are already fully automated trains in operation successfully all over the world.
So I say again, if ANY means of transport, lends itself to fully autonomous operation, more than any other, It is trains and railways.
Cars leave their destination when the driver wants it to, not in accordance with a printed schedule. the driver may stop off at the local shop for a paper and a coffee, or they may not. They might want to take the scenic route, or take a detour to pick someone up. They might want to take the slower back roads, or get onto a faster motorway as soon as possible.
It just seems odd, that people are trying to ,make `cars' which are controlled by so many variables, run autonomously, but trains which dont have nearly as many variables to deal with are left to a driver ,to try to make it run `like' an autonomous system.