Self drive and snow

Bruthal wrote: »

If a driver makes an error, and gets killed, did we let him die?

In the context of consciously stalling autonomous vehicle development of keep the number of road deaths high, then yes.

mloc123

eeguy wrote: »

That cars worth 100k. I wouldn't be risking it either

250k... but if he is afraid that some snow will damage it, he should not have been out driving

El_Duderino 09 wrote: »

Maybe I did miss the point. What was the point in relation to self driving cars vs current cars?

Current cars the driver is in control. Any modifications, they are still fully in control.
Fully automated cars might be frustratingly restrictive. New alternative unofficial firmware comes out to get around restrictions. It happens in every major official software system. The restrictions that will save lives we would assume. It was a tiny little point. If people are going to discuss subjects, points will be made, whether good or bad.

Again, it is absolutely nothing to do with others hacking into your car, or cutting brake lines with snips.

As far as I’m concerned, both could be tampered. It’s not something I would waste time worrying about. Do you worry about someone tampering with your car?

No, but the point never was anything to do with others tampering with your car. Perhaps my wording was not the best.

eeguy wrote: »

In the context of consciously stalling autonomous vehicle development of keep the number of road deaths high, then yes.

How many will it save in your lifetime?

If I live to 2070 and autonomous cars are released ASAP, then they'll save 1.1 million in the US alone. Considering the US accounts for about 3% of worldwide road deaths, if this was extrapolated out then about 36 million people total, equivalent to the entire population of Canada.

https://www.rand.org/blog/articles/2017/11/why-waiting-for-perfect-autonomous-vehicles-may-cost-lives.html

eeguy wrote: »

If I live to 2070 and autonomous cars are released ASAP, then they'll save 1.1 million in the US alone. Considering the US accounts for about 3% of worldwide road deaths, if this was extrapolated out then about 36 million people total, equivalent to the entire population of Canada.

https://www.rand.org/blog/articles/2017/11/why-waiting-for-perfect-autonomous-vehicles-may-cost-lives.html

Would you prefer travelling in one than driving yourself?

Absolutely. If I could buy one tomorrow I would.

The amount of hours people lose commuting in cars is staggering. If you're driving an hour each way, that's 10 hours a week, more than a whole working day, or a fantastic night's sleep

Ubbquittious

eeguy wrote: »

Absolutely. If I could buy one tomorrow I would.

The amount of hours people lose commuting in cars is staggering. If you're driving an hour each way, that's 10 hours a week, more than a whole working day, or a fantastic night's sleep

It's not so much about cars but the whole commuter lifestyle American dream. Most people commute for absolutely no reason and would be better off staying at home. Unless you're one of the few people still making physical objects or dealing with real people in real life.

The whole thing is just a massive hamster wheel to get money changing hands frequently but it's a huge waste of everyone's time. If they were looking down at this codology of throngs of people getting on the move every sunrise to get into another building and tap buttons for the day they'd be laughing their holes off at us. Communism should be given another chance I reckon

El_Duderino 09 wrote: »

That’s a terrible attitude towards road deaths It seems any old argument will do for those who don’t want the technology to develop.

Does a person who believes resources should be used to end world hunger, automatically not want other areas to progress?

eeguy wrote: »

Absolutely. If I could buy one tomorrow I would.

The amount of hours people lose commuting in cars is staggering. If you're driving an hour each way, that's 10 hours a week, more than a whole working day, or a fantastic night's sleep

It would be handy for that alright.

privateBeavis

Car ownership will be a big change if/when autonomous cars are common place. Most people's cars are parked > 90% of the time. When you need a car you'll just hail one with an app.

Dr_Kolossus

On average self drive cars may cause less accidents, but, this is an average. On average (not all, as anyone can have an accident) most accidents are caused by a few idiots, morons etc.

Self drive cars are no where close to a replacing a good/safe driver.

You've made two claims in that post, neither of which I'd agree with, based on my own limited experience.
One the first point, I'd say that most people who cause fatal crashes never drive, or even breathe again.

On the second point, there's no real level playing field to compare what state the tech is at. Google and Tesla both say their cars are statistically safer in autonomous mode, so where did you get your information there?

El_Duderino 09

Bruthal wrote: »

El_Duderino 09 wrote: »

Maybe I did miss the point. What was the point in relation to self driving cars vs current cars?

Current cars the driver is in control. Any modifications, they are still fully in control.
Fully automated cars might be frustratingly restrictive. New alternative unofficial firmware comes out to get around restrictions. It happens in every major official software system. The restrictions that will save lives we would assume. It was a tiny little point. If people are going to discuss subjects, points will be made, whether good or bad.

Again, it is absolutely nothing to do with others hacking into your car, or cutting brake lines with snips.

As far as I’m concerned, both could be tampered. It’s not something I would waste time worrying about. Do you worry about someone tampering with your car?

No, but the point never was anything to do with others tampering with your car. Perhaps my wording was not the best.

Ah I get you. You mean people might tamper with their own cars like now people tamper with their cars when they get them 'chipped'. I suppose it's exactly like everything else in life. If you tamper with safety features, you increase the risk or harm. That goes for self drive cars as well as anything else.

El_Duderino 09 wrote: »

Ah I get you. You mean people might tamper with their own cars like now people tamper with their cars when they get them 'chipped'. I suppose it's exactly like everything else in life. If you tamper with safety features, you increase the risk or harm. That goes for self drive cars as well as anything else.

Fully autonomous cars will likely have a lot more potential for modifications, jailbreaking etc than a car ECU. Likely without having to have any physical contact with it.

Just for a matter of interest on the subject...

Its all early days anyway.

I seen somewhere that kangaroos confuse the volvo sensing system as well. Doesnt mean much at this stage, but there is a lot more to it than
just navigating along a road. In cities near skyscrapers, gps reflections occur, which give apparent positions due to the reflected gps signal of mirror like glass sides of the buildings.

That can probably be corrected by putting ground based beacons in known exact positions though, and relaying corrections to cars.

Bruthal wrote: »

Fully autonomous cars will likely have a lot more potential for modifications, jailbreaking etc than a car ECU. Likely without having to have any physical contact with it.

Just for a matter of interest on the subject...

Its all early days anyway.

I seen somewhere that kangaroos confuse the volvo sensing system as well.

You'd wonder why a car needs a sign. My satnv knows the speed limit on every road I'm on. It's definitely importnat for temporary signs, like roadworks, but are these less likely to be defaced?

eeguy wrote: »

You'd wonder why a car needs a sign. My satnv knows the speed limit on every road I'm on. It's definitely importnat for temporary signs, like roadworks, but are these less likely to be defaced?

I was thinking that myself alright. I suppose the thing about it is that the sensors can be confused. Whether it be signs or obstacles.

There are a lot of things to consider, and that have not been thought of probably.

The technology probably all exists already. Its implementation is the thing. If you have ever messed around with microcontrollers, its easy enough to program one to avoid objects on a robot car yoke going around on the floor. Or follow a fixed path etc. A car out on the road is tricky though. There is a possibility it could end up being more than one system at first. But eventually will likely have to be one universal system adopted, assuming it happens.

Lantus

billbond4 wrote:

Taking a guy who works on self driving cars, he says he huge amount of computing power needed to work out all the possible scenrios etc on a public road is years away probably 30 to 40. Motorway driving in self driving is relatively easy as theres less variables

A symptom of many roads based on medieval tracks from 500 plus years ago. You cannot expect to apply modern technology to ancient cities in such a manner. Every component needs to be redesigned as fit for purpose.

It's time we built modern cities in more appropriate locations with intelligent layouts and recycle the existing ones for any useful materials or keep bits to show our kids how people used to live.

bladespin

Lantus wrote:

It's time we built modern cities in more appropriate locations with intelligent layouts and recycle the existing ones for any useful materials or keep bits to show our kids how people used to live.

Lol, technology is there to help and improve our lives not to force us to redesign them, a 'designed' city woul be a horrible and uninspiring place.

eeguy wrote: »

If I live to 2070 and autonomous cars are released ASAP, then they'll save 1.1 million in the US alone. Considering the US accounts for about 3% of worldwide road deaths, if this was extrapolated out then about 36 million people total, equivalent to the entire population of Canada.

https://www.rand.org/blog/articles/2017/11/why-waiting-for-perfect-autonomous-vehicles-may-cost-lives.html

How can they possibly know that? These things aren’t on the road yet.

Complex software is never perfect and car software is complex already. . Machine learning - a buzz word - needs training. The cars need to crash to learn what not to do.

I’m sure that semi autonomous cars will be useful. Self driving cars will never be perfect.

Lantus wrote: »

There are 1.2million deaths worldwide each year due to road accidents. It's a simply staggering waste of human life that autonomous cars have the potential to massively reduce. Anyone who opposes cutting that disgusting statistic is unsane.

That is pretty much /thread right there.
You can stuff all your arguments about the trolley dilemma, because in the real world you don't have time to sit down with a group of philosophy students and spend a weekend at a conference to debate the issue.
It's more like "hey, whaBAM!" and it's over.
Except with the autonomous car there'd be no BAM because it will have reacted.
So there will be a few crashes due to the car reacting incorrectly, but millions of crashes that won't happen at all.
It is useful to debate possible scenarios which an autonomous vehicle could be faced with, but it's idiotic to make the argument that those fantasy and extremely unlikely scenarios are a reason they are not viable.
Because in the face of billions of people driving drunk, drugged (legal and illegal), tired, distracted, under the influence of stupid and just plain not watching where they're going, this is no contest. Autonomous vehicle wins 9999 to 1.
And to say the 9999 accidents is better than the one is an unsustainable argument.
It will happen, get used to it.

edit:
It's more like several million to one...

dr.fuzzenstein wrote: »

That is pretty much /thread right there.
You can stuff all your arguments about the trolley dilemma, because in the real world you don't have time to sit down with a group of philosophy students and spend a weekend at a conference to debate the issue.
It's more like "hey, whaBAM!" and it's over.
Except with the autonomous car there'd be no BAM because it will have reacted.
So there will be a few crashes due to the car reacting incorrectly, but millions of crashes that won't happen at all.
It is useful to debate possible scenarios which an autonomous vehicle could be faced with, but it's idiotic to make the argument that those fantasy and extremely unlikely scenarios are a reason they are not viable.
Because in the face of billions of people driving drunk, drugged (legal and illegal), tired, distracted, under the influence of stupid and just plain not watching where they're going, this is no contest. Autonomous vehicle wins 9999 to 1.
And to say the 9999 accidents is better than the one is an unsustainable argument.
It will happen, get used to it.

edit:
It's more like several million to one...

I mean you guys keep saying that but software fails all the time. And if not programmed for a scenario it will crap out.

Then there’s the legal aspect. The car companies, not the driver or his insurance will have to take responsibility for any crash.

Franz Von Peppercorn wrote: »

I mean you guys keep saying that but software fails all the time. And if not programmed for a scenario it will crap out.

Then there’s the legal aspect. The car companies, not the driver or his insurance will have to take responsibility for any crash.

Critical software rarely fails. This isn't a phone app knocked up over the weekend by some teenager.
The litigation cost if something goes wrong would easily sink a company so you can bet nothing will be released until it's bulletproof.

The likes of Volvo have already said they'd take the legal burden if an autonomous car crashed. I'd say others will follow suit. Again, it's not worth the cost to release crappy software.

eeguy wrote: »

Critical software rarely fails. This isn't a phone app knocked up over the weekend by some teenager.
The litigation cost if something goes wrong would easily sink a company so you can bet nothing will be released until it's bulletproof.

The likes of Volvo have already said they'd take the legal burden if an autonomous car crashed. I'd say others will follow suit. Again, it's not worth the cost to release crappy software.

Which doesn’t mean they won’t. Google and Apple (etc) are used to iterative releases. Google and Apple follow agile processes. The idea is to release early and often. Move fast and break things.

Which means version 1.0 won’t be any good of it follows this practice. As it stands this software isn’t written, as far as I can tell, in the extremely documented regimented way that, say, airplane autopilot software is written. It can take months to change a line of code in those systems and Silicon Valley doesn’t follow that model. As for the manufacturers they already have spaghetti code i their systems.

The supposed gold rush here is highly dependent on speed, but critical software can’t be speedy.

Your move fast and break things analogy doesn't work in automotive, which is why it's at least 5 years behind consumer software. . They'll need to comply with 26262 at a minimum to release anything, and autonomous systems will be in the highest safety category, ASIL-D.
They won't be able to meet regulations if their software doesn't pass QA. And, as I said before, the cost of recall or litigation is far greater than the profit to be made by releasing early.

Franz Von Peppercorn wrote: »

Which doesn’t mean they won’t. Google and Apple (etc) are used to iterative releases. Google and Apple follow agile processes. The idea is to release early and often. Move fast and break things.

Which means version 1.0 won’t be any good of it follows this practice. As it stands this software isn’t written, as far as I can tell, in the extremely documented regimented way that, say, airplane autopilot software is written. It can take months to change a line of code in those systems and Silicon Valley doesn’t follow that model. As for the manufacturers they already have spaghetti code i their systems.

The supposed gold rush here is highly dependent on speed, but critical software can’t be speedy.

You don't seriously believe that the autonomous software in self driving cars will be updated along with Google maps and gmail?
Or that the team who develops your music player app also develops the AI software?
Seriously? Really?
Nah.

dr.fuzzenstein wrote: »

You don't seriously believe that the autonomous software in self driving cars will be updated along with Google maps and gmail?
Or that the team who develops your music player app also develops the AI software?
Seriously? Really?
Nah.

You’re right. it’s been built by the really smart guys. They haven’t had much to do in decades because it was all music players, operating systems, search algorithms,cloud and the like.

But these guys are the real smart software devs and although they’ve been paid for years to do nothing now they will produce perfect code in version 1.0 even though nobody has ever done this before and the list of unknown unknowns is massive. It will be magical software. Produced by magical people.

(Oh and existing car software is in fact updated. And autonomous car software will have to be if it dangerous. In fact it will probably be forced updates as cars can hardly continue to drive with known issues).

Tell me, where do you get your amazing insights of how the software for autonomous vehicles is developed? You must be in the team. Because when I'm reading your description, it's like I'm in the room with them. :rolleyes:

bladespin

Franz Von Peppercorn wrote:

Which doesnâ€™t mean they wonâ€™t. Google and Apple (etc) are used to iterative releases. Google and Apple follow agile processes. The idea is to release early and often. Move fast and break things.

Won't happen, I'm sure they will have to validate their software just like the FDA/FAA demand and having dealt with both they won't be updating often, your iphone is not a life critical device.