Donate to Remove ads

Got a credit card? use our Credit Card & Finance Calculators

Thanks to johnstevens77,Bhoddhisatva,scotia,Anonymous,Cornytiv34, for Donating to support the site

Driverless Vehicles

The Big Picture Place
odysseus2000
Lemon Half
Posts: 6366
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1535 times
Been thanked: 959 times

Re: Driverless Vehicles

#476507

Postby odysseus2000 » January 27th, 2022, 11:12 am

CliffEdge wrote:My idea of a driverless car is,

I get in, tell it where to go, anywhere in the UK, sit back and relax, read a book, listen to music, enjoy the views, have a nap, it wakes me up when I get there.

I don't need a licence, or even to be able to see.

Ain't gonna happen sadly. Anything else is driver assistance, not knocking it. Useful to have, but I'll still need a licence, be competent to drive, eyesight etc.

Not self driving private cars except in very restricted situations like the motorway where they might do most of the driving but you certainly won't be allowed to fall asleep! Never gonna happen in the next twenty years.


How do you know?

Sure it hasn’t happened yet, but what fundamental reason is there to support your thesis?

Back when the internet started, what was called the information super highway was stated to need massive government spending & take a long time, but it all happened very quickly.

Folk used to say with certainty that machines would never beat grand masters at chess & that GO would not be beaten for decades if at all. Then neural nets appear & now they defeat all grand masters, often many simultaneously & ditto for GO.

If fsd was a pure academic exercise it might not have much momentum, but there are trillions of $ to be made from
a working fsd, bringing in the best talent in this field, very like how the Manhattan project got all the best physicists in the US, & repeat in Russia, China etc.

If fsd is to happen all the ingredients to make it happen soon are in place.

In my humble opinion based on having worked on several software hardware projects, it is now probable that fsd will become a reality & that we will soon have humanoid robots. Tesla imho is very inexpensive, but of course can get cheaper if the whole market crashes.

Regards,

CliffEdge
Lemon Quarter
Posts: 1554
Joined: July 25th, 2018, 9:56 am
Has thanked: 452 times
Been thanked: 434 times

Re: Driverless Vehicles

#476513

Postby CliffEdge » January 27th, 2022, 11:32 am

odysseus2000 wrote:
CliffEdge wrote:My idea of a driverless car is,

I get in, tell it where to go, anywhere in the UK, sit back and relax, read a book, listen to music, enjoy the views, have a nap, it wakes me up when I get there.

I don't need a licence, or even to be able to see.

Ain't gonna happen sadly. Anything else is driver assistance, not knocking it. Useful to have, but I'll still need a licence, be competent to drive, eyesight etc.

Not self driving private cars except in very restricted situations like the motorway where they might do most of the driving but you certainly won't be allowed to fall asleep! Never gonna happen in the next twenty years.


How do you know?

Sure it hasn’t happened yet, but what fundamental reason is there to support your thesis?

Back when the internet started, what was called the information super highway was stated to need massive government spending & take a long time, but it all happened very quickly.

Folk used to say with certainty that machines would never beat grand masters at chess & that GO would not be beaten for decades if at all. Then neural nets appear & now they defeat all grand masters, often many simultaneously & ditto for GO.

If fsd was a pure academic exercise it might not have much momentum, but there are trillions of $ to be made from
a working fsd, bringing in the best talent in this field, very like how the Manhattan project got all the best physicists in the US, & repeat in Russia, China etc.

If fsd is to happen all the ingredients to make it happen soon are in place.

In my humble opinion based on having worked on several software hardware projects, it is now probable that fsd will become a reality & that we will soon have humanoid robots. Tesla imho is very inexpensive, but of course can get cheaper if the whole market crashes.

Regards,

I certainly hope you're right (about driverless cars) but in my opinion you (like many) are over promising.
Accordingly my initial intense excitement about driverless cars is over and I intend to forget about the unlikely possibility that I'll ever have one in my lifetime. Hope I'm proved wrong but don't expect to be.
What a delight they would be.

murraypaul
Lemon Slice
Posts: 785
Joined: April 9th, 2021, 5:54 pm
Has thanked: 225 times
Been thanked: 265 times

Re: Driverless Vehicles

#476530

Postby murraypaul » January 27th, 2022, 12:07 pm

odysseus2000 wrote:The Telegraph article was about traffic lights. Sure this is a non trivial problem but comparing a public road Tesla to a mining truck makes no allowance for the quality of the Tesla AI neural net training. I have not seen any traffic light mistakes in the US.



https://futurism.com/the-byte/tesla-aut ... ffic-light

murraypaul
Lemon Slice
Posts: 785
Joined: April 9th, 2021, 5:54 pm
Has thanked: 225 times
Been thanked: 265 times

Re: Driverless Vehicles

#476534

Postby murraypaul » January 27th, 2022, 12:11 pm

odysseus2000 wrote:If fsd was a pure academic exercise it might not have much momentum, but there are trillions of $ to be made from
a working fsd, bringing in the best talent in this field, very like how the Manhattan project got all the best physicists in the US, & repeat in Russia, China etc.

If fsd is to happen all the ingredients to make it happen soon are in place.


Eventually, you are probably right. It is the soon word that is the problem.
Because you would have said exactly the same thing last year, and the year before, and the year before that.
And why wouldn't you, because that is what Tesla and Elon were saying.
But it still isn't here, and doesn't really seem that much closer.

I don't think we will have an actual full self-drive system (level 5 autonomy, no driver needed in the vehicle, no remote control) driving freely on the roads in any major European country in the next 10 years. Do you?

I think we may have dedicated highways/highways lanes for long haul autonomous trucking, which is both simpler and of immediate economic advantage.

murraypaul
Lemon Slice
Posts: 785
Joined: April 9th, 2021, 5:54 pm
Has thanked: 225 times
Been thanked: 265 times

Re: Driverless Vehicles

#476567

Postby murraypaul » January 27th, 2022, 1:51 pm

One way of looking at this. The UK law commissions have just recommended that owners of truly self-driving cars shouldn't be liable for driving incidents, the car manufacture should be. (https://www.bbc.com/news/technology-60126014)

On that basis, how soon do you think car manufactures will consider their technology robust enough that they are willing to accept liability for it.

I don't see that happening any time soon.

Edit: Actual report available here: https://www.lawcom.gov.uk/project/automated-vehicles/

odysseus2000
Lemon Half
Posts: 6366
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1535 times
Been thanked: 959 times

Re: Driverless Vehicles

#476582

Postby odysseus2000 » January 27th, 2022, 2:31 pm

murraypaul wrote:One way of looking at this. The UK law commissions have just recommended that owners of truly self-driving cars shouldn't be liable for driving incidents, the car manufacture should be. (https://www.bbc.com/news/technology-60126014)

On that basis, how soon do you think car manufactures will consider their technology robust enough that they are willing to accept liability for it.

I don't see that happening any time soon.

Edit: Actual report available here: https://www.lawcom.gov.uk/project/automated-vehicles/


So the car maker takes out insurance and has definitive data on what events led to the accident allowing pay if it's the cars fault, no pay if it's another's fault etc.

This legislation paves the way for self driving robots.

Regards,

CliffEdge
Lemon Quarter
Posts: 1554
Joined: July 25th, 2018, 9:56 am
Has thanked: 452 times
Been thanked: 434 times

Re: Driverless Vehicles

#476620

Postby CliffEdge » January 27th, 2022, 4:48 pm

odysseus2000 wrote:
murraypaul wrote:One way of looking at this. The UK law commissions have just recommended that owners of truly self-driving cars shouldn't be liable for driving incidents, the car manufacture should be. (https://www.bbc.com/news/technology-60126014)

On that basis, how soon do you think car manufactures will consider their technology robust enough that they are willing to accept liability for it.

I don't see that happening any time soon.

Edit: Actual report available here: https://www.lawcom.gov.uk/project/automated-vehicles/


So the car maker takes out insurance and has definitive data on what events led to the accident allowing pay if it's the cars fault, no pay if it's another's fault etc.

This legislation paves the way for self driving robots.

Regards,

You believe in it. I hope you're right. I hope you're a patient customer.

gryffron
Lemon Quarter
Posts: 3607
Joined: November 4th, 2016, 10:00 am
Has thanked: 551 times
Been thanked: 1587 times

Re: Driverless Vehicles

#476628

Postby gryffron » January 27th, 2022, 5:08 pm

The problem I see with driverless vehicles is, even if the technology could be made perfect, humans will abuse it.

In a hurry to cross a busy city street? No problem. Just stride boldly out in front of a driverless car, confident in the knowledge it will stop. Bringing all the other traffic to a halt, so all the other pedestrians can now cross too.

I don't think it will ever be possible to mix fully automatic and humans in the same space.

:(

murraypaul
Lemon Slice
Posts: 785
Joined: April 9th, 2021, 5:54 pm
Has thanked: 225 times
Been thanked: 265 times

Re: Driverless Vehicles

#476634

Postby murraypaul » January 27th, 2022, 5:33 pm

odysseus2000 wrote:
murraypaul wrote:One way of looking at this. The UK law commissions have just recommended that owners of truly self-driving cars shouldn't be liable for driving incidents, the car manufacture should be. (https://www.bbc.com/news/technology-60126014)

On that basis, how soon do you think car manufactures will consider their technology robust enough that they are willing to accept liability for it.

I don't see that happening any time soon.


So the car maker takes out insurance and has definitive data on what events led to the accident allowing pay if it's the cars fault, no pay if it's another's fault etc.

This legislation paves the way for self driving robots.


It does, but when?

How much do you think it would currently cost to buy that self-driving car insurance, if anyone would even write it?

We aren't arguing about if, but when.

odysseus2000
Lemon Half
Posts: 6366
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1535 times
Been thanked: 959 times

Re: Driverless Vehicles

#476652

Postby odysseus2000 » January 27th, 2022, 7:20 pm

murraypaul wrote:
odysseus2000 wrote:
murraypaul wrote:One way of looking at this. The UK law commissions have just recommended that owners of truly self-driving cars shouldn't be liable for driving incidents, the car manufacture should be. (https://www.bbc.com/news/technology-60126014)

On that basis, how soon do you think car manufactures will consider their technology robust enough that they are willing to accept liability for it.

I don't see that happening any time soon.


So the car maker takes out insurance and has definitive data on what events led to the accident allowing pay if it's the cars fault, no pay if it's another's fault etc.

This legislation paves the way for self driving robots.


It does, but when?

How much do you think it would currently cost to buy that self-driving car insurance, if anyone would even write it?

We aren't arguing about if, but when.


Insurance companies write insurance for 17 year old males who have appalling odds of an accident.

Also to the earlier point of someone stopping traffic by deliberately jay walking, he or she would be caught on camera & could be traced via their mobile phone if legislations are changed to allow this in case of illegal activity.

Regards,

murraypaul
Lemon Slice
Posts: 785
Joined: April 9th, 2021, 5:54 pm
Has thanked: 225 times
Been thanked: 265 times

Re: Driverless Vehicles

#476658

Postby murraypaul » January 27th, 2022, 7:35 pm

odysseus2000 wrote:Insurance companies write insurance for 17 year old males who have appalling odds of an accident.


Any single 17 year old male will have a massive variance in expected payout, because they are risky, but all different.

They are all individual, unconnected risks.

In aggregate, the expected costs of insuring 17 year old males can be calculated quite precisely.

If your self driving tech has a bug that causes it to always make the same mistake in the same situation, your variance is much higher. Either the bug doesn't exist, and no cars are affected, or it does, and all of them are.

The variance is much higher, and so harder to price.

Insurance is based on the idea that when you aggregate a huge number of very high variance events, the overall variance is very low. That relies on each event being independent.

It is an interesting problem, I would be intrigued to see what the calculations would be.

But would you be comfortable unwriting Tesla's first version of actual full self drive?

Howard
Lemon Quarter
Posts: 2178
Joined: November 4th, 2016, 8:26 pm
Has thanked: 885 times
Been thanked: 1017 times

Re: Driverless Vehicles

#476705

Postby Howard » January 27th, 2022, 9:44 pm

If Tesla were that confident in FSD they'd write the insurance themselves. ;)

regards

Howard

odysseus2000
Lemon Half
Posts: 6366
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1535 times
Been thanked: 959 times

Re: Driverless Vehicles

#476734

Postby odysseus2000 » January 28th, 2022, 12:10 am

murraypaul wrote:
odysseus2000 wrote:Insurance companies write insurance for 17 year old males who have appalling odds of an accident.


Any single 17 year old male will have a massive variance in expected payout, because they are risky, but all different.

They are all individual, unconnected risks.

In aggregate, the expected costs of insuring 17 year old males can be calculated quite precisely.

If your self driving tech has a bug that causes it to always make the same mistake in the same situation, your variance is much higher. Either the bug doesn't exist, and no cars are affected, or it does, and all of them are.

The variance is much higher, and so harder to price.

Insurance is based on the idea that when you aggregate a huge number of very high variance events, the overall variance is very low. That relies on each event being independent.

It is an interesting problem, I would be intrigued to see what the calculations would be.

But would you be comfortable unwriting Tesla's first version of actual full self drive?


If the testing routine is good it should catch any serious errors before the neural net is sent to the wild.

Regards,

TUK020
Lemon Quarter
Posts: 2039
Joined: November 5th, 2016, 7:41 am
Has thanked: 762 times
Been thanked: 1175 times

Re: Driverless Vehicles

#476770

Postby TUK020 » January 28th, 2022, 8:51 am

Seen headline for DT:
Calling cars 'self-driving' should be criminal offence if human input needed, says Law Commission
Do not have access to the full article

CliffEdge
Lemon Quarter
Posts: 1554
Joined: July 25th, 2018, 9:56 am
Has thanked: 452 times
Been thanked: 434 times

Re: Driverless Vehicles

#476790

Postby CliffEdge » January 28th, 2022, 9:32 am

odysseus2000 wrote:
murraypaul wrote:
odysseus2000 wrote:Insurance companies write insurance for 17 year old males who have appalling odds of an accident.


Any single 17 year old male will have a massive variance in expected payout, because they are risky, but all different.

They are all individual, unconnected risks.

In aggregate, the expected costs of insuring 17 year old males can be calculated quite precisely.

If your self driving tech has a bug that causes it to always make the same mistake in the same situation, your variance is much higher. Either the bug doesn't exist, and no cars are affected, or it does, and all of them are.

The variance is much higher, and so harder to price.

Insurance is based on the idea that when you aggregate a huge number of very high variance events, the overall variance is very low. That relies on each event being independent.

It is an interesting problem, I would be intrigued to see what the calculations would be.

But would you be comfortable unwriting Tesla's first version of actual full self drive?


If the testing routine is good it should catch any serious errors before the neural net is sent to the wild.

Regards,

It's very hard to burst your bubble.

servodude
Lemon Half
Posts: 8272
Joined: November 8th, 2016, 5:56 am
Has thanked: 4435 times
Been thanked: 3565 times

Re: Driverless Vehicles

#476793

Postby servodude » January 28th, 2022, 9:47 am

CliffEdge wrote:
odysseus2000 wrote:
murraypaul wrote:
odysseus2000 wrote:Insurance companies write insurance for 17 year old males who have appalling odds of an accident.


Any single 17 year old male will have a massive variance in expected payout, because they are risky, but all different.


If the testing routine is good it should catch any serious errors before the neural net is sent to the wild.

Regards,

It's very hard to burst your bubble.


I was about to bring up the difficulties of choosing a training set to explain my worries... but I think your post sums up stuff better than I could

ANNs are great fun and we've come a long way from the classic "1 hidden layer abstraction can model anything" (if you get your model order right ;) ) and there is a lot that can be done with machine learning (just look at AWS offering AMIs with graphics cards targetted at the industry for proof of it being commercially viable as a tool - we spun a couple up yesterday for a deep learning data analysis job)

Where using them falls down in practice... possibly unfairly.. is the assumption of risk.

Even if your system is 200x better than humans driving... every mess up falls at your feet. That PR will be hard to wear.

-sd

Lootman
The full Lemon
Posts: 18681
Joined: November 4th, 2016, 3:58 pm
Has thanked: 628 times
Been thanked: 6564 times

Re: Driverless Vehicles

#476846

Postby Lootman » January 28th, 2022, 1:17 pm

servodude wrote:Even if your system is 200x better than humans driving... every mess up falls at your feet. That PR will be hard to wear.

Yes, and in fact it doesn't really matter if driverless cars are 200 times or even 2,000 times safer. The first time a driverless car mows down a schoolchild on a zebra crossing, it will set back driverless cars by years.

The people who don't die because of driverless cars won't know it. It is the families of the ones who did die that will kick off.

Midsmartin
Lemon Slice
Posts: 778
Joined: November 4th, 2016, 7:18 am
Has thanked: 211 times
Been thanked: 491 times

Re: Driverless Vehicles

#476850

Postby Midsmartin » January 28th, 2022, 1:23 pm

This is interesting. Someone on Twitter has been counting how many self driving disengagements they get with each version of their Tesla fsd. It's getting better quite quickly.

https://twitter.com/eliasmrtnz1/status/ ... 2Yj3g&s=19

odysseus2000
Lemon Half
Posts: 6366
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1535 times
Been thanked: 959 times

Re: Driverless Vehicles

#476881

Postby odysseus2000 » January 28th, 2022, 2:43 pm

Lootman wrote:
servodude wrote:Even if your system is 200x better than humans driving... every mess up falls at your feet. That PR will be hard to wear.

Yes, and in fact it doesn't really matter if driverless cars are 200 times or even 2,000 times safer. The first time a driverless car mows down a schoolchild on a zebra crossing, it will set back driverless cars by years.

The people who don't die because of driverless cars won't know it. It is the families of the ones who did die that will kick off.


I don’t think it will work this way at least in the US.

There have been an horrendous number of gun mass murders. Most could be stopped if guns were banned, but despite all the rhetoric it never happens.

Once fsd launches the number of car accidents plummets & the odd disaster becomes like guns, too popular with too many to ban it.

Regards,

odysseus2000
Lemon Half
Posts: 6366
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1535 times
Been thanked: 959 times

Re: Driverless Vehicles

#476886

Postby odysseus2000 » January 28th, 2022, 2:47 pm

Midsmartin wrote:This is interesting. Someone on Twitter has been counting how many self driving disengagements they get with each version of their Tesla fsd. It's getting better quite quickly.

https://twitter.com/eliasmrtnz1/status/ ... 2Yj3g&s=19


Yes this is the way all the Monte Carlo codes I have written have behaved. All kinds of failures at the start, then slowly the number of failures reduces & then plummets.

The more I watch of US Tesla fsd videos the more I swing to thinking it is going to work.

If fsd works I can easily see stock 5x from where ever it was the day before.

Regards,


Return to “Macro and Global Topics”

Who is online

Users browsing this forum: No registered users and 9 guests