• Our booking engine at tickets.railforums.co.uk (powered by TrainSplit) helps support the running of the forum with every ticket purchase! Find out more and ask any questions/give us feedback in this thread!

Driverless Cars - the future?

Status
Not open for further replies.

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
Evidence please.
The fact is that computers are able to react quicker, that can't be denied. At the Olympics they will not be having people standing at the finish line with stopwatches as this would rely on a human reacting to the person crossing the line which is not going to react as fast as fast as an automated system. Also at the start of a race if an athlete moves of the start line less than a tenth of a second after the start then it is declared a false start and this is based on the best possible reaction time which isn't going to be this fast for everyone.
 
Last edited:
Sponsor Post - registered members do not see these adverts; click here to register, or click here to log in
R

RailUK Forums

gswindale

Member
Joined
1 Jun 2010
Messages
788
That is a completely different situation though.

Is there any actual evidence that 2 cars being controlled by computers (only) can successfully run at 70mph only 1.5 seconds apart and both come to an emergency stop without hitting each other?
 

S19

Member
Joined
5 Apr 2010
Messages
275
The fact is that computers are able to react quicker, that can't be denied. At the Olympics they will not be having people standing at the finish line with stopwatches as this would rely on a human reacting to the person crossing the line which is not going to react as fast as fast as an automated system. Also at the start of a race if an athlete moves of the start line less than a tenth of a second after the start then it is declared a false start.

That isn't evidence that cars will be able to run faster and closer with computers.

So have you got any evidence that they'll be able to?
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
That isn't evidence that cars will be able to run faster and closer with computers.

So have you got any evidence that they'll be able to?
It's quite clear that a faster reaction time with the computers will allow it to react quicker than a human can and so allow the cars to run closer without affecting safety. Also the reaction times of some humans are quite poor compared to others without taking computers into consideration. With driverless cars though the fact that the computer would be able to react faster would allow it to be closer to the vehicle in front as it would be able respond to a change in speed faster than a human could. The possibility of running cars closer together at higher speeds has also been reported in the media http://www.foxbusiness.com/on-air/stossel/blog/2011/06/25/nevada-legalizes-self-driving-cars-0 following the decision by Nevada to legalize driverless cars. Here's an article on the subject by Randal O'Toole: http://online.wsj.com/article/SB10001424052748703580904575131511589391150.html
--- old post above --- --- new post below ---
Is there any actual evidence that 2 cars being controlled by computers (only) can successfully run at 70mph only 1.5 seconds apart and both come to an emergency stop without hitting each other?
It should be quite obvious based on the reaction times of computers compared to humans that a computer controlled car would be able to do this better than a human can.
 
Last edited:

starrymarkb

Established Member
Joined
4 Aug 2009
Messages
5,985
Location
Exeter
That is a completely different situation though.

Is there any actual evidence that 2 cars being controlled by computers (only) can successfully run at 70mph only 1.5 seconds apart and both come to an emergency stop without hitting each other?

Only ways I can see is if they are wireless networked, but then how much bandwidth would you need and how do you prioritise and secure the data...

The other option would be radar, but might that be better as a driver aid as it might be subject to interference from other cars on the same frequency
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
Only ways I can see is if they are wireless networked, but then how much bandwidth would you need and how do you prioritise and secure the data...
Yes, this would allow the car to know what other vehicles are in the immediate area even if they are not visible and what they are going to be doing. It would also allow vehicles to quickly be informed of any disruption so they can take an alternative route.
The other option would be radar, but might that be better as a driver aid as it might be subject to interference from other cars on the same frequency
Radar cruise control is starting to appear now.
 

gswindale

Member
Joined
1 Jun 2010
Messages
788
It's quite clear that a faster reaction time with the computers will allow it to react quicker than a human can and so allow the cars to run closer without affecting safety. Also the reaction times of some humans are quite poor compared to others without taking computers into consideration. With driverless cars though the fact that the computer would be able to react faster would allow it to be closer to the vehicle in front as it would be able respond to a change in speed faster than a human could.
--- old post above --- --- new post below ---

It should be quite obvious based on the reaction times of computers compared to humans that a computer controlled car would be able to do this better than a human can.

In theory yes, but is there actually any evidence - can you give me a link from, for instance, the TRL proving this?

Currently; the best way to fit as much traffic onto the M25 as proven through science is to pretty much be stationary.

From memory; I'm supposed to keep 2 seconds behind the car in front. In the case I posted earlier of a car stalling at 30mph coming off a roundabout I managed to just avoid hitting them despite being more than that behind due to the lack of advance warning.

Are you saying that each car will also know exactly what is happening in each & every other car in the vicinity?

It sounds like a good idea in theory, but the practical implementation doesn't really seem to have been looked at yet by anybody!
 

starrymarkb

Established Member
Joined
4 Aug 2009
Messages
5,985
Location
Exeter
Are you saying that each car will also know exactly what is happening in each & every other car in the vicinity?

It sounds like a good idea in theory, but the practical implementation doesn't really seem to have been looked at yet by anybody!

That's what I'm meaning. What communication method would you use? WiFi, GSM, UMTS or something proprietary. What sort of range would you be looking at. How would you ensure data got through in all conditions
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
In theory yes, but is there actually any evidence - can you give me a link from, for instance, the TRL proving this?

Currently; the best way to fit as much traffic onto the M25 as proven through science is to pretty much be stationary.

From memory; I'm supposed to keep 2 seconds behind the car in front. In the case I posted earlier of a car stalling at 30mph coming off a roundabout I managed to just avoid hitting them despite being more than that behind due to the lack of advance warning.

Are you saying that each car will also know exactly what is happening in each & every other car in the vicinity?

It sounds like a good idea in theory, but the practical implementation doesn't really seem to have been looked at yet by anybody!
Well I think it's clear that computers can react faster than humans can and this is one key benefit to introducing driverless cars. The technology is still under development and testing so far has been quite limited and there certainly haven't been any trials yet of running many of these cars close together at the same time but this doesn't change the fact that it would be possible but due to the technology needed it will take some time before it is fully introduced, there are still quite a few obstacles to be overcome and that's why the cars are not already on the road now. Now if you parked a lot of stationary cars on the M25 that would give the highest capacity possible but that would be pointless. The two second gap between yourself and the vehicle in front is due to safety reasons but due to the ability of computers to react faster a smaller gap could be left and this would allow for capacity to be increased without slowing down the traffic, something that couldn't be done as safely with manually driven cars.
 
Last edited:

jon0844

Veteran Member
Joined
1 Feb 2009
Messages
28,013
Location
UK
Timing at a sports event is not even comparable. I was in China for the Olympics and visited most of the key sponsors, including Atos Origin (yes the very same!) and saw how things are done. Clever technology yes, but not quite enough to have all cars driving around on their own.

Did you know that Atos provides scoring, timing and other key data and in order to give commentators a chance to keep up, the broadcast footage is on a slight delay to allow information such as who is first, second or third to be shown on screens for broadcasters to read without making mistakes? So, a little trickery and not super awesome computers (clever as the set up is).
 

SS4

Established Member
Joined
30 Jan 2011
Messages
8,589
Location
Birmingham
If computers can't parse the routeing guide what hope is there for driverless cars?
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
Timing at a sports event is not even comparable. I was in China for the Olympics and visited most of the key sponsors, including Atos Origin (yes the very same!) and saw how things are done. Clever technology yes, but not quite enough to have all cars driving around on their own.
The fact is though that a computer system is going to react much faster than a human can, regardless of if it is monitoring time when athletes cross the line or the speed of the vehicle in front. Driverless cars are clearly much more complicated than timing but technology is only going to improve in the future.
--- old post above --- --- new post below ---
If computers can't parse the routeing guide
It's certainly not impossible that a computer would be able to do this.
 
Last edited:

jon0844

Veteran Member
Joined
1 Feb 2009
Messages
28,013
Location
UK
Besides driverless cars, what else do you think computers will let us do in the next ten years or twenty years?
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
Besides driverless cars, what else do you think computers will let us do in the next ten years or twenty years?
Technology is only going to improve and in the future it's likely there will be much more automation. It may well be that when you go to a store there will be no cashiers and you will pay using an automated system if indeed people still actually go to stores in the future rather than ordering online. Other forms of transport are likely to see more automation, driverless trains are a possibility although their introduction is likely to be significantly delayed due to the need for new infrastructure.
 
Last edited:

LE Greys

Established Member
Joined
6 Mar 2010
Messages
5,389
Location
Hitchin
Technology is only going to improve and in the future it's likely there will be much more automation. It may well be that when you go to a store there will be no cashiers and you will pay using an automated system if indeed people still actually go to stores in the future rather than ordering online. Other forms of transport are likely to see more automation, driverless trains are a possibility although their introduction is likely to be significantly delayed due to the need for new infrastructure.

As long as there's some Human oversight, that would probably be a lot safer because of the old concept of block signalling and the fact that the computer could not lock itself in a decision loop. That's always been my biggest concern, that the guidance computer could not decide whether to go left or right, and ends up ramming a bridge or something. Computers cannot make judgement calls that require intuition or guesswork, whereas a good driver can.
 

WestCoast

Established Member
Joined
19 Jun 2010
Messages
5,574
Location
Glasgow
It may well be that when you go to a store there will be no cashiers and you will pay using an automated system

Erm, that technology already exists in multiple forms, and IIRC Tesco are trialling an Express store that doesn't have manned tills.

I wouldn't underestimate the human factor when it comes to service functions, machines are unlikely to be able to emulate that in the near future.
 

LE Greys

Established Member
Joined
6 Mar 2010
Messages
5,389
Location
Hitchin
A computer can be programmed though for these circumstances.

Is it possible to programme it to be flexible? Or if it (say) finds itself pinned in a narrow gap with a lorry one one side closing in and a pedestrian on the other, what is to stop it deciding that the pedestrian is a 'soft' target and running them over as the most 'logical' course of action? Obviously, it would be best if it did not get into those circumstances in the first place, but it could happen.
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
that technology already exists in multiple forms, and IIRC Tesco are trialling an Express store that doesn't have manned tills.
Yes it does but it hasn't fully taken over yet, in 20 years time it may well have done.
--- old post above --- --- new post below ---
Is it possible to programme it to be flexible? Or if it (say) finds itself pinned in a narrow gap with a lorry one one side closing in and a pedestrian on the other, what is to stop it deciding that the pedestrian is a 'soft' target and running them over as the most 'logical' course of action?
What's to stop a person doing that if it's going to save their own life?
 
Last edited:

LE Greys

Established Member
Joined
6 Mar 2010
Messages
5,389
Location
Hitchin
Erm, that technology already exists in multiple forms, and IIRC Tesco are trialling an Express store that doesn't have manned tills.

I wouldn't underestimate the human factor when it comes to service functions, machines are unlikely to be able to emulate that in the near future.

I agree. It's not just that a cheerful shopkeeper can brighten your day (where do you find them these days). Computers have a hard time handling syntax, so the onus is on the operator to give clear instructions. Developing a way that one can understand language in the same way that a Human can should be a major priority. Trouble is, neuroscience is still working that one out.
--- old post above --- --- new post below ---
What's to stop a person doing that if it's going to save their own life?

I would hope, a certain level of nobility.
 

DaveNewcastle

Established Member
Joined
21 Dec 2007
Messages
7,387
Location
Newcastle (unless I'm out)
. . . . . what else do you think computers will let us do in the next ten years or twenty years?
Spend more, perhaps without actually consuming any more.

Access facts more readily, but perhaps facts without much context.

Communicate more, but perhaps with less to say.

Alter our patterns of consumption, but to maximise supplier's access to customers and not to maximise our access to materials.

Electronic voting.

Electronic medical diagnostics and therapies.
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
I would hope, a certain level of nobility.
The issue here is that a human could react to the situation to give themselves the best chance of survival regardless of other people involved whereas a computer can be programmed to take the course of action that has the best chance of survival for everyone involved.
 

LE Greys

Established Member
Joined
6 Mar 2010
Messages
5,389
Location
Hitchin
The issue here is that a human could react to the situation to give themselves the best chance of survival regardless of other people involved whereas a computer can be programmed to take the course of action that has the best chance of survival for everyone involved.

Yes, but I'd rather put my faith in a Human conscience than a soulless machine. That's probably foolish, but that's me.
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
Yes, but I'd rather put my faith in a Human conscience than a soulless machine. That's probably foolish, but that's me.
It wouldn't be too difficult to add a rule though to prevent the car from swerving into an area where there are pedestrians.
 

GB

Established Member
Joined
16 Nov 2008
Messages
6,457
Location
Somewhere
Do you actually know anything about AI and what it would take to make a conscious decision on par with a human like that?
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
Do you actually know anything about AI and what it would take to make a conscious decision on par with a human like that?
If the car was conscious then there is the possibility it would want to do everything it can to save itself over everything else regardless of the consequences of this decision.
 
Last edited:

jon0844

Veteran Member
Joined
1 Feb 2009
Messages
28,013
Location
UK
Other forms of transport are likely to see more automation, driverless trains are a possibility although their introduction is likely to be significantly delayed due to the need for new infrastructure.

So, we will have driverless cars in 8-10 years, complete with all the work needed to basically rebuild the entire road network, fence of roads, add in new crossings (a road crossing today can cost well over £100k) and convince the population to give up motoring.

And yet trains, in a pretty controlled area right now (give or take open crossings and of course stations) may take significantly longer because of the need for new infrastructure.

Hmmm.

Why is it that I'd say driverless trains and a whole new rail network would be easier than the road version? And why is it that HS2 won't be complete until long after you've managed to design, build, test and install driverless cars? How many people will you expect to build the network? And I assume it's not just the UK going it alone, so how many people to do the whole of the EU, USA, Japan etc?

Spend more, perhaps without actually consuming any more.

Access facts more readily, but perhaps facts without much context.

Communicate more, but perhaps with less to say.

Alter our patterns of consumption, but to maximise supplier's access to customers and not to maximise our access to materials.

Electronic voting.

Electronic medical diagnostics and therapies.

I'd go along with all of that. Technology will make all of our lives easier (I've been saying for long enough that smartcards will transform travel, and this IS going to happen!) and I doubt many people will want to design anything to actually stop humans being able to continue enjoying things as they do today.

However much technology advances, it is all built around people who aren't going to change much. We'll still be social creatures that want to talk to other real people, travel to see people and so on.

Sure, some people might have a vision of the future that mimics Demolition Man, Total Recall or I, Robot - but that's not a world many people would really want to live in... and the transformation to such a world isn't going to be in our lifetime, or a few after that.

If the car was conscious then there is the possibility it would want to do everything it can to save itself over everything else regardless of the consequences of this decision.

The car isn't going to be conscious. It's going to the running a computer program, which is probably going to need to be updated regularly with patches and updates. How often does your computer need system updates? Now imagine these having to be rolled out to your driverless cars - and the problems if some aren't updated. And what about sabotage? How do you prevent someone installing rogue code?

Don't say it's not a problem as it will be - especially for terrorists of the future. How will you solve that?
 

Zoe

Established Member
Joined
22 Aug 2008
Messages
5,905
So, we will have driverless cars in 8-10 years, complete with all the work needed to basically rebuild the entire road network, fence of roads, add in new crossings (a road crossing today can cost well over £100k) and convince the population to give up motoring.
No, the first driverless cars will likely be available in 8 - 10 years but these will use existing infastructure. It will take longer for them to completely take over and for new infrastructure to be installed.
--- old post above --- --- new post below ---
The car isn't going to be conscious. It's going to the running a computer program, which is probably going to need to be updated regularly with patches and updates. How often does your computer need system updates? Now imagine these having to be rolled out to your driverless cars - and the problems if some aren't updated. And what about sabotage? How do you prevent someone installing rogue code?

Don't say it's not a problem as it will be - especially for terrorists of the future. How will you solve that?
This will all have to be addressed before driverless cars take over but is not impossible. The key here is that there are very clear safety benefits with driverless cars so that should encourage more work in finding solutions to the issues that are delaying the introduction. There will of course need to be extensive testing before any new system can be introduced so that the technology can be proven to work.
 
Last edited:

DaveNewcastle

Established Member
Joined
21 Dec 2007
Messages
7,387
Location
Newcastle (unless I'm out)
. . .
The key here is that there are very clear safety benefits with driverless cars so that should encourage more work in finding solutions to the issues that are delaying the introduction. There will of course need to be extensive testing before any new system can be introduced so that the technology can be proven to work.
This point gets at the heart of one challenge which I cannot see clearly how we might overcome. Its not just the safety but the relative safety.

The roads are a hazardous environment with a much lower safety threshold than many other personal or occupational risks. When testing and certifying an automated transport system, do we simply accept a standard of safety which is equal to or better than current road safety, but which is still more hazardous than any other occupational exposure to risk?
If so, then how do manufactures, regulators, transport service providers and the employers of passengers defend themselves against claims of irresponsible recklessness and inadequate risk mitigation measures?
If not, then how can an automated road service ever reach the standards of safety which exist in other regulated industries and avoid the claim-culture which we find in business and the risk-aversion which has characterised so many public services?
 
Status
Not open for further replies.

Top