Self driving buses, trains, trucks etc

Twenty Grand

cnocbui wrote: »

The only ones will be test vehicles.

Unbridled optimism is a suburb of ignorance.

Optimism is essential to achievement and it is also the foundation of courage and true progress.

cnocbui

Twenty Grand wrote: »

Optimism is essential to achievement and it is also the foundation of courage and true progress.

Operative word: 'unbridled'

Stephen15

Why do people want autonomous vehicles anyway. Do people hate driving or something like that or do they want to put people who drive for living out of a job.

Twenty Grand

Stephen15 wrote: »

Why do people want autonomous vehicles anyway. Do people hate driving or something like that or do they want to put people who drive for living out of a job.

Lots of reasons: the hope of less road deaths, more efficient transport services, cheaper transport costs, more productive use of time, greater mobility for people in rural areas or those who can't drive, less congestions, more even use of road networks etc etc...

There's loads of benefits.

Most people don't care for driving. Most people hate commuting. I love driving, but the idea of commuting 2 hours each day, 10 hours a week is a waste of time IMO, and judging by recurring threads on Boards, most people would agree.

Regarding job losses, that's an unfortunate consequence, but it will be a slow rollout over 2 or 3 decades.

TheQuietFella

markodaly wrote: »

Autonomous vehicles will be on our roads within the next 5 years, in some shape or form. Denial is not just a river in Egypt.

And Christmas is being moved to the month of June also!

markodaly

Stephen15 wrote: »

Why do people want autonomous vehicles anyway. Do people hate driving or something like that or do they want to put people who drive for living out of a job.

150 people die on the roads every year in Ireland. How about we reduce this down to 15 or less?
People who commute an hour each way, how about if these people could work or do something more productive with their time, rather than just drive?

markodaly

TheQuietFella wrote: »

And Christmas is being moved to the month of June also!

They are already on the roads in the US and being tested in Germany. You honestly don't think that there will be zero autonomous vehicles in Ireland within 5 years?

One can argue that even the Tesla as somewhat autonomous, and they have been on Irish roads for the past few years.

The years 2025-2030 will be the point of inflection in terms of electronic propulsion AND smarter more autonomous cars. From then on in, the death of the dumb ICE cars will only be a matter of time.

markodaly

No one is stating its going to happen tomorrow, or even next year but this freight train is steaming towards us at a quicker pace than some may like or want. I repeat denial is not just a river in Egypt.

TheQuietFella

markodaly wrote: »

No one is stating its going to happen tomorrow, or even next year but this freight train is steaming towards us at a quicker pace than some may like or want. I repeat denial is not just a river in Egypt.

It's a bit like time travel, it just can't be done!

cnocbui

markodaly wrote: »

150 people die on the roads every year in Ireland. How about we reduce this down to 15 or less?
People who commute an hour each way, how about if these people could work or do something more productive with their time, rather than just drive?

There is absolutely no justification for baldly stating that autonomous driving systems will be inherently safer than human drivers.

Far more time, effort and money has gone into computer and mobile phone operating systems security, than has been spent on autonomous driving, and they are all still vulnerable to hacking. Stating autonomous driving will be safer than humans is like claiming they will soon have a general use OS that has perfect security.

The makers and boosters of the F35 fighter, the most complex and expensive weapons system ever, sold it with a vision that it could do a whole host of amazing things, on the faith and optimism that they could easily write the software to make it happens. They were considerably over optimistic about that and have been unable to write the software.

Autonomous driving safety is just like that, an optimistic faith that it is obviously something easily achievable, polluted by the ill-informed idea that humans are bad at the task. They aren't.

Elon Musk thought he could produce the Tesla model 3 in huge numbers, cheaply, because he thought you could just fully automate the whole process with robots and software. He also thought existing manufacturers were stupid for not already doing this and that he was so much smarter than them and would make it happen. He fell flat on on his face. As an explanation for not being able to achieve his vision, he said 'humans are underrated'.

Believing autonomous driving just has to be able to do the job better than humans is the same as Musk thinking you could make cars more efficiently with robots than with humans.

Cordell

TheQuietFella wrote: »

It's a bit like time travel, it just can't be done!

Actually, we all do travel in time.

magentis

If the automotive industry still cannot produce reliable sensors for driver assist like autonomous braking,there is no way I'm putting my life in the hands of a fully autonomous vehicle.

Twenty Grand

magentis wrote: »

If the automotive industry still cannot produce reliable sensors for driver assist like autonomous braking,there is no way I'm putting my life in the hands of a fully autonomous vehicle.

So just because they can't do something now, means they'll never be able to do it.

Good logic there :rolleyes:

Sam Russell

Twenty Grand wrote: »

So just because they can't do something now, means they'll never be able to do it.

Good logic there :rolleyes:

Steam cars never happened because it cannot be done - thermodynamics was against it. It did look possible at the start, but there you go.

Batteries are just about getting there, so electric cars are not just possible, but are actually coming on stream. Previously electric cars required the cars to weigh next to nothing because it was not possible to push them along otherwise because of the lead acid batteries.

Autonomous vehicles will suffer from cyclists and pedestrians and their unexpected changes in direction. If a pedestrian walks out in front of a autonomous car, it will stop. Hmm, looks like the makings of a new game - chicken.

markodaly

TheQuietFella wrote: »

It's a bit like time travel, it just can't be done!

Please explain how autonomous vehicles cannot be done.

markodaly

cnocbui wrote: »

There is absolutely no justification for baldly stating that autonomous driving systems will be inherently safer than human drivers.

Apart from the evidence that shows us, that indeed that they are.
https://www.zdnet.com/article/how-autonomous-vehicles-could-save-over-350k-lives-in-the-us-and-millions-worldwide/
1.2 million people die a year due to accidents, 94% of which are caused by human error.

Far more time, effort and money has gone into computer and mobile phone operating systems security, than has been spent on autonomous driving, and they are all still vulnerable to hacking. Stating autonomous driving will be safer than humans is like claiming they will soon have a general use OS that has perfect security.

Who says it has to be perfect? No one. It just has to be better than humans, which will not be that difficult.
We do not have perfect technology or OS or hardware or software, yet we run stock exchanges, life support systems, air traffic control, military defense systems, weather predictor systems, without this 'perfect' software.

Autonomous driving safety is just like that, an optimistic faith that it is obviously something easily achievable, polluted by the ill-informed idea that humans are bad at the task. They aren't.

When 1.2 million people are dying each year, 94% of them due to human error, I am not sure I would state that, as humans doing a good job. Computers will be able to do a better job.

Elon Musk thought he could produce the Tesla model 3 in huge numbers, cheaply, because he thought you could just fully automate the whole process with robots and software. He also thought existing manufacturers were stupid for not already doing this and that he was so much smarter than them and would make it happen. He fell flat on on his face. As an explanation for not being able to achieve his vision, he said 'humans are underrated'.

So Elon Musk was wrong about the ramping up of the manufacturing process proves that autonomous vehicles cannot be done. LOL.
You do know that Telsa are making over 320,000 EV's a year now? Not bad for a car company that barely made any cars 10 years ago.

Believing autonomous driving just has to be able to do the job better than humans is the same as Musk thinking you could make cars more efficiently with robots than with humans.

Eh, no it's not.

Your argument is a non sequitur.
Self-driving cars are not here today = they will never be here.
Now that is a faith argument if I ever saw it.

markodaly

There are hundreds of SAE 2 cars on Irish roads today.
Any Tesla or new Nissan Leaf with pro-pilot is classed as SAE 2 along with offerings from Mercedes and BMW.
This is today in 2018.

The new 2019 Audi A8 will be classed as SAE 3.


Google managed to test and achieve SAE 3 back in 2008 but decided against bringing it to the market because in their studies they found that humans relied too much on it. Therefore they took the steps to go straight for SAE 5 and have been testing their Waymo on the streets of Phoneix Arizona for the past few years. 4 million miles have been logged so far.

All the new EV's that will be coming on the market over the next few years will be at least SAE 2, if not SAE 3 by 2020.
By 2025 all new cars will be at least SAE 3 apart from a few budget offerings like a Yaris.
You honestly think the likes of Audi or BMW or Volvo or VW will not be offering SAE 5 self-driving vehicles around this time?

It's not going to happen tomorrow, of course, but those thinking that we are decades and decades away from this are deluded. It's going to be on our streets within 10 years. Maybe not in huge numbers at first but once we reach that inflection point like mobile phones, take up of it will be exponential.

mfceiling

TheQuietFella wrote: »

We're just not designed to follow the rules and this will be the reason that this concept will ultimately fail!

Ha!! Car companies investing billions in this technology but because Ireland "don't follow rules" then it will ultimately fail. Would you go away out of that.
Autonomous vehicles are coming and faster than you all think...30 to 50 years...try 10 - 15. Look at a car that was released in 1968 compared to a Tesla today.

cnocbui

markodaly wrote: »

Apart from the evidence that shows us, that indeed that they are.
https://www.zdnet.com/article/how-autonomous-vehicles-could-save-over-350k-lives-in-the-us-and-millions-worldwide/
1.2 million people die a year due to accidents, 94% of which are caused by human error.

That article isn't worth the paper it's printed on, it's complete and utter nonsense - pure fantasy. It is entirely based on the idea of an infallible autonomous driving system that DOESN"T F'ing EXIST! This is your proof!?

The fatality rate for Ireland is 3.9 deaths per billion km driven.

The distance driven fully autonomously by WAYMO is 9 million km. But that isn't actually autonomous because a human has to intervene every 13 to 5,600 miles and take over, so humans are still doing the heavy lifting, but that's good, because no one has been killed yet.

Tesla on the other hand, is a company of out and out cowboys and have no such safety backup, and it shows, as 3 people have so far been killed by Teslas being driven by autopilot. The estimates of the total distance driven under autopilot are very suspect. There is also no record of disengagements so the data would be meaningless.

Uber have one fatality so far, but that's not very surprising given their brilliant system supposedly had a disengagement rate of once every single mile.

No autonomous car has managed even 100,000 km without a disengagement, yet you are proclaiming that autonomously driven cars will be at least 10 times safer than humans and will be near infallible. What a steaming load of BS.

Who says it has to be perfect? No one. It just has to be better than humans, which will not be that difficult.

Oh really? Based on what? Nothing but your wishful thinking, is what.

We do not have perfect technology or OS or hardware or software, yet we run stock exchanges, life support systems, air traffic control, military defense systems, weather predictor systems, without this 'perfect' software.

What a load of... ATC is done by humans. Weather prediction is highly innacurate beyond a short time frame. And all of that is overseen by humans who monitor and override continuously.

When 1.2 million people are dying each year, 94% of them due to human error, I am not sure I would state that, as humans doing a good job. Computers will be able to do a better job.

Once again, there is absolutely no evidence for your claim. If any of the current autonomous driving systems had been in control for all the distance driven that resulted in those 1.2 million deaths, it is beyond certain that the death toll would likely have been at least 100 times higher - presuming no humans intervened and saved the day

Eh, no it's not.

Your argument is a non sequitur.
Self-driving cars are not here today = they will never be here.
Now that is a faith argument if I ever saw it.

Don't accuse me of saying things I haven't. I do not discount the possibility that autonomous driving might one day prove to be better than humans. What I am disputing are people who state that it is absolutely certain that they will be better and that such an outcome is unquestionably inevitable. It's not.

I don't think fully autonomous driving systems that are as good as humans are achievable in the next ten years. I think they may not be possible at all, as 40 years of driving experience leads me to believe deep learning probably isn't going to be good enough and that true general intelligence might be required to get bet better than human performance. If that turns out to be the case, then you can kiss the prospect of 'better than human' goodbye.

Stephen15

mfceiling wrote: »

Ha!! Car companies investing billions in this technology but because Ireland "don't follow rules" then it will ultimately fail. Would you go away out of that.
Autonomous vehicles are coming and faster than you all think...30 to 50 years...try 10 - 15. Look at a car that was released in 1968 compared to a Tesla today.

And what do they both have in common pedals and a steering wheel

mfceiling

Germany launches world's first autonomous tram in Potsdam

https://www.theguardian.com/world/2018/sep/23/potsdam-inside-the-worlds-first-autonomous-tram?CMP=Share_AndroidApp_Copy_to_clipboard

It's coming like I said.

Stephen15

mfceiling wrote: »

Germany launches world's first autonomous tram in Potsdam

https://www.theguardian.com/world/2018/sep/23/potsdam-inside-the-worlds-first-autonomous-tram?CMP=Share_AndroidApp_Copy_to_clipboard

It's coming like I said.

But there was still a driver in the cab who could intervene so not exactly driverless is it. It will be a very long time before there are vehicles without driver controls such as pedals and a steering wheel that's not gonna happen anytime soon. Look at the DLR in London which still has an on-board member of staff who can take over if needs be.

AndrewJRenko

TheQuietFella wrote: »

If anyone one seriously thinks that Irish Roads, and the Irish road mentality will cater for this concept ye're out of ye're minds! No programming can predict or adequately perceive the 'what's next factor' for the Irish road user be it car driver, bus, HGV, road runner or pedestrian.

We're just not designed to follow the rules and this will be the reason that this concept will ultimately fail!

That could actually be an argument FOR technology/AI/autonomous cars - that the technology is more capable of reacting to unexpected events and actions than human drivers, who operate based on a whole pile of assumptions.


Did anyone else get to ride the bus on Friday? It worked fine for me, with the exception of one unexpected and unexplained sudden stop. The French guy said 'it could have been a leaf'. [I don't think he was referring to the Nissans].

Stephen15

AndrewJRenko wrote: »

Did anyone else get to ride the bus on Friday? It worked fine for me, with the exception of one unexpected and unexplained sudden stop. The French guy said 'it could have been a leaf'. [I don't think he was referring to the Nissans].

I wouldn't that yoke was in the Docklands a bus. A capsule would be a better definition. You could probably walk at a quicker pace than it too.

Twenty Grand

Stephen15 wrote: »

But there was still a driver in the cab who could intervene so not exactly driverless is it. It will be a very long time before there are vehicles without driver controls such as pedals and a steering wheel that's not gonna happen anytime soon. Look at the DLR in London which still has an on-board member of staff who can take over if needs be.

That's a great thing though! The DLR has a "train captain" who's role is more customer service and assistance than driving. Imagine if every DART had the same thing.
Much better use of someones time than sitting in a cab.

markodaly

cnocbui wrote: »

That article isn't worth the paper it's printed on.

Clearly, you didn't even read it, as the data from the article comes from a US Dept. of Transporation study into road fatalities.

The fatality rate for Ireland is 3.9 deaths per billion km driven.

Yes, we have come a long way in the past few decades as the rate in 1990 was 19.2. How did we get there? Are Irish drivers, 80% better than what they were in 1990. No, not by a long shot.

The plain truth of the matter is, is that cars have gotten infinitely safer, along with infrastructure improvement. Things like drink driving campaigns have also helped mind you.

However, it is the technology in the car itself that is the biggest contributor to falling deaths on Irish roads. The fact the cyclists' deaths have been climbing steadily in the past few years shows us that.

Here is a short 2-minute video showing the difference between a 1997 hatch back and its 2017 equivalent.

The cars of today are packed with safety features. The next dawn of safety features are things like Automatic Emergency Braking, Lane assists, things like that, things that are already on Irish roads and any SAE level 2 car.
SAE 2 cars and the features they provide will be standard in a few years.

The distance driven fully autonomously by WAYMO is 9 million km. But that isn't actually autonomous because a human has to intervene every 13 to 5,600 miles and take over, so humans are still doing the heavy lifting, but that's good, because no one has been killed yet.

Despite the fact that for the past year a human has been in the back seat of the Waymo car and that these cars are now being tested in 25 separate cities in the US? This is 2018, imagine what they will be up to in 2025?

Tesla on the other hand, is a company of out and out cowboys and have no such safety backup, and it shows, as 3 people have so far been killed by Teslas being driven by autopilot. The estimates of the total distance driven under autopilot are very suspect. There is also no record of disengagements so the data would be meaningless.

It seems you have a hard-on for Musk. I am not really a fan of him personally but I do admire the way he pushed EV's into the mainstream, so much so that the other car manufacturers are now rushing to get on board.

Those 3 deaths you say, are not due to autonomous driving, it is due to human error. Teslas does not have fully autonomous vehicles, so how one can blame self-driving cars on a car that is not self-driving is beyond me.

The people riding in those cars are meant to have their hands on the steering wheel and pay attention to the road conditions. In one of the accidents, the person in the driving seat was watching Netflix on his laptop. Again, human error at fault here. This is actually the reason as to why Google ditched their plans for a SAE level 3 car, and have gone instead for SAE 4/5. The human idiot factor in level 3 was just too much of a risk.

Uber have one fatality so far, but that's not very surprising given their brilliant system supposedly had a disengagement rate of once every single mile.

Is this the fatality where the person jumped out of the darkness and state investigators found out afterward that a human driver would never have spotted this person either, in fact, the Uber car itself responded much quicker than a human ever could. Unfortunately, Uber made some errors here I will admit, however, with each accident, the system can learn and evolve so that the next time such a situation is avoided. Meanwhile, no comment on the tens of thousands of people who die on the roads each day.

No autonomous car has managed even 100,000 km without a disengagement, yet you are proclaiming that autonomously driven cars will be at least 10 times safer than humans and will be near infallible. What a steaming load of BS.

No, not me. The US Department of Transportation.

Oh really? Based on what? Nothing but your wishful thinking, is what.

You incorrectly stated that the software has to be 'perfect'. My contention is that it does not have to be perfect, it just has to be better than humans.

Once again, there is absolutely no evidence for your claim. If any of the current autonomous driving systems had been in control for all the distance driven that resulted in those 1.2 million deaths, it is beyond certain that the death toll would likely have been at least 100 times higher - presuming no humans intervened and saved the day

LOL, care to give me some evidence of your claim

Here is some of my evidence.
https://www.wired.com/2017/01/probing-teslas-deadly-crash-feds-say-yay-self-driving/

Instead, the agency exculpated Tesla, and then some. It crunched the numbers to find that among Tesla cars on the road, those carrying its Autosteer technology, which can keep the car within clear lane markings, crashed 40 percent less frequently than those without.

“It’s very positive for Tesla,” says Bart Selman, who studies AI safety systems at Cornell University. “It puts the whole issue of the Florida accident in the right context.” Meaning, the system isn't flawed. It just isn't fully advanced yet

I don't think fully autonomous driving systems that are as good as humans are achievable in the next ten years.

Ironic seeing that semi-autonomous cars are already proving safer as per NHTSA. SAE level 4/5 cars are already on the road in the US, you honestly think that they will stay there for the next 10-20 years?

To repeat, Tesla, Mercedes, Nissan, Land Rover just to name a few all have SAE level 2 tech right now. Audi is releasing the A8 with SAE level 3.

Do you think level 3 is the highest we can go for the next 10 years? Will we get stuck in the mud with just semi-autonomous driving for the next 40 years? I very much doubt it.

I think they may not be possible at all, as 40 years of driving experience leads me to believe deep learning probably isn't going to be good enough and that true general intelligence might be required to get bet better than human performance. If that turns out to be the case, then you can kiss the prospect of 'better than human' goodbye.

Ah, this is the nub of it. Purely anecdotal evidence on your behalf. 'I am a good driver' there, a computer cannot be better than me.

They said the same about a computer beating a chess grand master, it could not be done until it was done.

You video link didn't work but I found an alternate

https://www.youtube.com/watch?v=xidhx_f-ouU

magicbastarder

Would be curious to hear what the balance of 'cognitive load' is on autonomous systems, of dealing with well behaved traffic vs. dealing with badly behaved traffic. Is it possible that the biggest problem is the factor they're intended to replace? Humans who aren't nearly as engaged with the demands of driving as they should be?

markodaly

Youtube link fixed.

cnocbui

markodaly wrote: »

Clearly, you didn't even read it, as the data from the article comes from a US Dept. of Transporation study into road fatalities.

Yes, we have come a long way in the past few decades as the rate in 1990 was 19.2. How did we get there? Are Irish drivers, 80% better than what they were in 1990. No, not by a long shot.

The plain truth of the matter is, is that cars have gotten infinitely safer, along with infrastructure improvement. Things like drink driving campaigns have also helped mind you.

However, it is the technology in the car itself that is the biggest contributor to falling deaths on Irish roads. The fact the cyclists' deaths have been climbing steadily in the past few years shows us that.

Here is a short 2-minute video showing the difference between a 1997 hatch batch and its 2017 equivalent.

The cars of today are packed with safety features. The next dawn of safety features are things like Automatic Emergency Braking, Lane assists, things like that, things that are already on Irish roads and any SAE level 2 car.
SAE 2 cars and the features they provide will be standard in a few years.

Despite the fact that for the past year a human has been in the back seat of the Waymo car and that these cars are now being tested in 25 separate cities in the US? This is 2018, imagine what they will be up to in 2025?

It seems you have a hard-on for Musk. I am not really a fan of him personally but I do admire the way he pushed EV's into the mainstream, so much so that the other car manufacturers are now rushing to get on board.

Those 3 deaths you say, are not due to autonomous driving, it is due to human error. Teslas does not have fully autonomous vehicles, so how one can blame self-driving cars on a car that is not self-driving is beyond me.

The people riding in those cars are meant to have their hands on the steering wheel and pay attention to the road conditions. In one of the accidents, the person in the driving seat was watching Netflix on his laptop. Again, human error at fault here. This is actually the reason as to why Google ditched their plans for a SAE level 3 car, and have gone instead for SAE 4/5. The human idiot factor in level 3 was just too much of a risk.

Is this the fatality where the person jumped out of the darkness and state investigators found out afterward that a human driver would never have spotted this person either, in fact, the Uber car itself responded much quicker than a human ever could. Unfortunately, Uber made some errors here I will admit, however, with each accident, the system can learn and evolve so that the next time such a situation is avoided. Meanwhile, no comment on the tens of thousands of people who die on the roads each day.

No, not me. The US Department of Transportation.

You incorrectly stated that the software has to be 'perfect'. My contention is that it does not have to be perfect, it just has to be better than humans.

LOL, care to give me some evidence of your claim

Here is some of my evidence.
https://www.wired.com/2017/01/probing-teslas-deadly-crash-feds-say-yay-self-driving/

Ironic seeing that semi-autonomous cars are already proving safer as per NHTSA. SAE level 4/5 cars are already on the road in the US, you honestly think that they will stay there for the next 10-20 years?

To repeat, Tesla, Mercedes, Nissan, Land Rover just to name a few all have SAE level 2 tech right now. Audi is releasing the A8 with SAE level 3.

Do you think level 3 is the highest we can go for the next 10 years? Will we get stuck in the mud with just semi-autonomous driving for the next 40 years? I very much doubt it.

Ah, this is the nub of it. Purely anecdotal evidence on your behalf. 'I am a good driver' there, a computer cannot be better than me.

They said the same about a computer beating a chess grand master, it could not be done until it was done.

Here is the one line from the NHTSA report that you cite as proof that autonomous driving systems, will, without doubt, reduce road deaths by over 90%:

So ADSs have the potential to significantly reduce highway fatalities by addressing the root cause of these tragic crashes.

That is no more 'proof' than if the Department of Energy released a similarly styled report stating that Nuclear Fusion has the potential to reduce greenhouse emissions and provide for near limitless amounts of cheap energy.

Both statements would be true, but they are not evidential and are merely aspirational. There is no 'will happen' in the statement you cite as evidence.

You are completely wrong about the Uber death being unavoidable. What actually happened is that the Tempe Police chief jumped the gun and blabbed within two days of the accident that it was unavoidable, even if a human had been driving or paying attention. She probably didn't realise that the Uber video that was released was probably doctored by Uber to make it appear that the environment was near pitch-black, when actually it is quite well lit.


Normal dash cam vs Uber's dash cam

The police chief, Sylvia Moir, later regretted being so hasty to pre-judge, because the 'official' police report actually concluded that the accident was entirely avoidable by a human driver.

On June 21, Tempe investigators released the report that contradicted Moir following a three-month probe conducted with the National Transportation Safety Board.

Besides declaring the crash avoidable, police are seeking a manslaughter charge against Vasquez, who was hired by Uber to watch the road as the experimental car rolled along in autonomous mode.

https://www.phoenixnewtimes.com/news/tempe-police-chief-moir-ryff-uber-autonomous-driving-10556991

You really need to work on being able to distinguish between fact and fiction.