"Tesla expects to enable full self-driving by the end of 2017"

[whitex]

You totally lost me here. There is no cure for cancer coming in 12 months. There are videos of Tesla's AP2.0 driving itself (with a driver in the seat to oversee only). So I don't get your point.

You were using the stats of how many people die as if it was a justification of why it's coming sooner. If not, why even mention it in this thread?

You can't use quotation marks around words I never said. I never said there will "always" be a driver in the seat. This thread is about the wiki quote: "Tesla expects to enable full self-driving by the end of 2017" and people taking issue with that statement. I said "While 12 months may be somewhat optimistic for full autonomous" it is not for self-driving. There is a difference. There's no way the regulators will go directly to autonomous. The system will need to prove itself with a driver in the seat first. I see that happening with 12 months. The ride sharing, etc. will come later.

Agreed, overuse of quotation marks. So you do think AP2 will be able to drive without anyone in the car, yes or no? If yes, any weather, any conditions? What happens if weather changes, construction happens, accidents require driving in the wrong direction highway, or any other myriad of things that can happen while the car is driverless?

Yes, of course. I am really surprised you don't. It seems so obvious to me.

Scenario 1: drunk person gets into the car, crashed into a fence in the parking lot, is arrested.
Scenario 2: drunk person gets into the car, tell AP to start driving home. On the way the car gets confused and starts driving in the wrong lane (which was the right lane yesterday but today there is construction so the opposite traffic is routed through the left lane of the highway). The drunk person is sleeping and the car plows head on into the car with 3 teenagers in it.

Seems obvious to me too.

You've been tainted badly by AP1.0. Throw what you know about AP out the window. The lack of hardware and slow processing power are nothing at all like AP2.0. To suggest AP2.0 will cause more accidents that humans who cause 94% of deaths is ludicrous to me.

You'll probably be telling the same story when AP3 comes out, and AP4. The current sensors have no redunancy do deal with obstructed camera (see any wipers on cams except for the front ones, even those I suspect will freeze up as AP1 did as I don't see any heating elements there).

And therein lies my fear -- as I said above -- which is that more people with your type of mindset make up the regulatory bureaucracy than those with my type.

Please don't label. Let me clarify my mindset:
1. I am a pragmatic, I would prefer 25 vs 100 people dead.
2. I am not a bureaucrat, never have, never plan to be.
3. As a pragmatic, I can recognize how people panic and over-react. In the city where I grew up a pit-bull mauled a small child. The city banned all bit-bulls from being owned in city limits. Completely irrational, but it happens. If you want more irrational, look up bans on dihydrogen-monoxide. One autonomous car kills a little child and it's game over - industry set back by 10+ years as people freak out.
4. As a pragmatic I can recognize other issues which will need to be solved, but which are hard - such as liability in case of an accident if there is no driver in the car. How to address maintenance? I don't foresee cars becoming the next aircraft industry with maintenance log requirements, driven hours per part, mandatory rebuilds, etc - all needed for aircraft manufacturer to guarantee safety.
5. Given Tesla history of putting out hardware, over-promising and under-delivering, it is not difficult to extrapolate that this will be another one of those. Common sense tells me if I cover the camera's with mud, all I have is AP1.0 sensors - front cam, radar and ultrasonic - yes there is more processing power but so what? And yes, they can get covered in mud in one mud splash while driving without anyone inside, now what?
6. I think Tesla is making great progress, just overselling things by a lot (another good example, 691hp they sold P85D customers on - it took 3 hardware generations to actually get there, and even now apparently if you use it too much you get permanently throttled back to protect the drivetrain).

If Tesla was really so sure they will get to FSD with AP2, they would offer warranty until such time. Personally, if that was the case, I would not only activate AP2 on my recent car, but also get one more Tesla to replace my AP1 car. I asked when buying my most recent MS by the way, they are not willing to guarantee FSD before bumper to bumper warranty expires, or even before the drivetrain warranty expires, so pretty much no guarantees whatsoever.

What do you say we pick up this discussion when I can have my car drop me off at Seattle airport and then meet me the next day in San Jose? Assuming TMC will still be around and this thread will not be taken offline. I do hope I'm wrong, I just don't think so.

 

[3Victoria]

@whitex. You may be right. Bringing up hypothetical examples to disprove self-driving is contrived. If something happens that disables the car, such as mud, it will be disabled -- so what. If you are driving and have a stroke, how will you keep driving? Obviously you won't. Tesla isn't saying that AP2 will be all singing and dancing and the best thing since sliced bread. It will do certain things and will have limitations ... Just like everything else. No one is forced to pay for the enhancements, just forego them. I remain sceptical but hopeful Tesla will achieve what they are promising. They have achieved the seemingly impossible before.

 

[whitex]

@whitex. You may be right. Bringing up hypothetical examples to disprove self-driving is contrived. If something happens that disables the car, such as mud, it will be disabled -- so what. If you are driving and have a stroke, how will you keep driving? Obviously you won't. Tesla isn't saying that AP2 will be all singing and dancing and the best thing since sliced bread. It will do certain things and will have limitations ... Just like everything else. No one is forced to pay for the enhancements, just forego them. I remain sceptical but hopeful Tesla will achieve what they are promising. They have achieved the seemingly impossible before.

The "contrived example" is to give an example of issues that will come and need to be resolved. If you have a heart attack and you plow into traffic killing someone, people feel bad for you and sympathize with you and the victim - nothing could have been done to prevent it. If your AP car gets blinded by the sun or weather conditions and plows into traffic, people sue the car-maker for not making it better, they don't feel bad for them, they feel angry and will fight to hold someone responsible (by making them pay and/or legislating what they can or cannot do in the future).

AP2 today does absolutely nothing, so any discussion about what it will or will not do is purely hypothetical. The self-driving video tells us nothing other that someone had built a prototype which worked once in a specific set of conditions. The trick to get it right is to get it to work in most conditions, so another 95% of the work. The only things that are not hypothetical and relate to the possibility that FSD will be here in 12 months (the topic of this thread) are:
1. AP2 is not even on par with AP1 yet
2. Tesla offers no guarantees that FSD will ever do everything they say it will.
3. Tesla has a history of over-promising and under-delivering. In 5 days we'll know if they deliver on their first promise of basic EAP in December.

Everything else remains to be seen. I wish Tesla well, support them by buying their products and highly recommending their products on their current merits to anyone who can afford it. I just don't buy into the hype of big, overhyped features coming "soon" anymore.

 

[lunitiks]

Great discussion going on here!

And hey: Only a few days left of December

 

[Doug_G]

I see that most people who do not understand exponential technology consistently give
distant timeframes for its deployment.

People who understand technology development will be the most skeptical.

The edge cases will take years to resolve. I can't imagine a self-driving car to manage blizzard conditions any time soon.

 

[lunitiks]

Besides the 1% corner cases, snow (/"blizzards") might be the ultimate challenge for this system in practice. Of course, the car must be able to safely pull over and stop, but in some areas this fail safe move will happen far to often to be acceptable

 

[msnow]

People who understand technology development will be the most skeptical.

The edge cases will take years to resolve. I can't imagine a self-driving car to manage blizzard conditions any time soon.

My field is both technology and regulatory compliance. We are many years away, many.

 

[Finder]

Airbags are the closest parallel to the FSD safety paradox. Just read a stat here:
Air Bags Save Many, Kill Some

Estimated 15,000 lives saved and 242 killed in the first 20 years after airbags were mandated.

Some automatic process killed those 242 people though - similar to how FSD might if it runs into a problem. They are still mandated by law.

Elon is smart to run this in shadow mode for a while. If they can truly prove how much safer it is... it is a numbers game.

Some of these examples - disabled cameras, buzzards, etc. if that happens the car should pull over. Just like a normal person. What happens if the power steering fails in a car? You pull over. What if a tire blows out? You pull over. What about if the motor dies? Pull over.

The description of FSD is very specific - twice as good as a human. Most humans probably drive terribly in blizzards. Honestly most humans drive terribly almost all the time - the bar is low. Agreed that when a FSD Tesla runs somebody over they will sue Tesla - but as they do now, Tesla will have the data to explain what happened.

Said this in another thread a while back - but this will start with Tesla releasing level 4 autonomy (human may need to take over in some situations). With enough data and laws being passed we get to level 5 (no human needed - ever).

I really don't know what law prevents Tesla from releasing Level 4 autonomous cars as long as the driver accepts liability.

A friend of mine has the complete opposite fear as everyone on the thread. His belief is once automous driving improves enough that manual driving will be outlawed. Regulators have every incentive to get self driving cars on the road - Human drivers are terrible. It will happen sooner than everyone thinks. FSD will be the mandatory airbag of tomorrow.

 

[3Victoria]

@Finder Agree.

... your spouse gets blinded by the sun or weather conditions and plows into traffic, people sue them for not driving it better, they don't feel bad for them, they feel angry and will fight to hold someone responsible (by making them pay and/or legislating what they can or cannot do in the future).

ssme story?

The edge cases will take years to resolve. I can't imagine a self-driving car to manage blizzard conditions any time soon.

AP is not expected to drive no matter what the conditions.

 

[Doug_G]

AP is not expected to drive ni matter what the conditions.

AP... nope.

Level 5 autonomous? What's it gonna do... driving down a remote highway and a blizzard blows in: Pull to a stop and waits for a car to slam into the back? Pull off the highway and wait for the occupants to freeze to death?

Around here on half the winter days you can't see the lines on the road, and the other half has many obscured lines and some lanes partially blocked by snowbanks. Nice high snowbanks make it hard to see cross traffic, pedestrians, etc. So now it can't drive half the year? These aren't trivial problems to solve.

 

[Finder]

AP... nope.

Level 5 autonomous? What's it gonna do... driving down a remote highway and a blizzard blows in: Pull to a stop and waits for a car to slam into the back? Pull off the highway and wait for the occupants to freeze to death?

Around here on half the winter days you can't see the lines on the road, and the other half has many obscured lines and some lanes partially blocked by snowbanks. Nice high snowbanks make it hard to see cross traffic, pedestrians, etc. So now it can't drive half the year? These aren't trivial problems to solve.

Don't regular people have trouble driving in those conditions? I know I would. You should see what happened in Atlanta with 1 inch of snow - was like the apocalypse - folks just abandoned their cars on the side of the road unable to drive. These things will happen in stages - FSD isn't going to be better than a human driver in every situation out of the gate. I am sure eventually it will be though as HW3, HW4, etc comes out.

They should have put some extra language - drives twice as safe as the average human driver.

Totally get what you are getting at with the snow in Canada though. I am sure HW2 with FSD (even when fully enabled) will have trouble in those conditions. I am sure your driving in those conditions is 100 times safer than someone used to driving in Georgia though.

 

[Canuck]

My field is both technology and regulatory compliance. We are many years away, many.

Great. You're exactly the person who can answer the questions I have but I don't know what question you answered above so please answer these:

1. How many years (or months) are we away from vehicles doing more than AP1.0? (i.e. anything more, such as stopping at lights/stop signs, or any one thing more than AP1.0 can do?)

2. How many years (or months) are we away from vehicles self-driving with a person in the seat overseeing and taking over whenever the vehicle requires?

3. How many years are we away from autonomous driving vehicles?

I didn't put "or months" in brackets for number 3 because I think that must be more than a year away, and that's why you said many (twice) but how many is many, many? But maybe you will tell me all of the above are more than one year away? I don't think so but you have much more knowledge than me on this issue so I look forward to your answers.

 

[Doug_G]

Don't regular people have trouble driving in those conditions? I know I would. You should see what happened in Atlanta with 1 inch of snow - was like the apocalypse - folks just abandoned their cars on the side of the road unable to drive.
...
I am sure your driving in those conditions is 100 times safer than someone used to driving in Georgia though.

Yep. The chaos in Atlanta has two major causes:

1. Most cars in Atlanta will have summer tires. They can be terrifying on ice. Even so-called "all season" tires don't do that well. No one in Atlanta has proper winter tires!
2. People in Atlanta have no clue about how to drive in snow and ice. They've never done it. Up here it's something every driver learns early on - it's unavoidable.

We get full-up no-holds-barred winter here. Yes there are days when it's unwise to venture out, but they are rare. We had an ice storm on Sunday that led to some major highway accidents. We went out only briefly on an errand, and the side streets were literally skating rinks (main roads were salted). I had no problem driving on it (in my RWD Model S), though I drove carefully of course. This evening I ran into black ice and had a moment... but I successfully avoided hitting another car that had stopped in front of me. I held the brake down and used the ABS to steer around it.

Around here, one inch of snow just makes the traffic worse. It's no biggie. But one inch of snow to a self-driving car will likely mean it can't see any lines on the road. Even if it can figure out how to control the car in those conditions, it won't necessarily know where to go. Even when the road surface isn't obscured, optical sensors get covered in salt, snow, ice, etc. I'm constantly wiping off my rear-view camera.

 

[Doug_G]

Here's a pic of a friend's Model S. We had no trouble driving in these conditions. Just need snow tires. What would a self-driving car make of this?

 

[Canuck]

Here's a pic of a friend's Model S. We had no trouble driving in these conditions. Just need snow tires. What would a self-driving car make of this?View attachment 208117

I guess everything can be taken to the extreme. I drove home today on a snow covered highway (the No. 3 - "Crow's Nest" in BC) and two cars passed me at a high rate of speed. They sent slush up on windscreen and I said to my daughter that they were "yahoos" -- I was doing over the speed limit (at 110 km/hr - the limit was 100 - they passed me at over that speed). Right after I said that, the one in front lost control, spun around, and hit the one behind. Luckily, no one was injured. And lucky for me, I avoided the crash being right behind them. It was a good lesson for my 17 year old daughter to learn watching how easily that can happen. The first thing she said when we got home to her mother and sisters was about that accident. The next thing was how unfortunate that we didn't have it on the dashcam. The dashcam is in the Tesla but I was driving my Tahoe hybrid, which I barely ever drive unless there is snow over the mountain passes.

Long story short, yes - in snow it won't work. But most of the time we don't drive in snow and I'd much prefer AP2.0 (with lots of hardware and a super compter) be driving and exposing my kids to that on the road than the vast majority of humans. Sorry you humans, but the the number one cause of death for teenagers is motor vehicle accidents, and those are caused by humans 94% of the time. Given those stats, please bring on self-driving cars ASAP!

 

[whitex]

@Finder
ssme story?

Not at all. Unless you are Bill Gates, your spouse is not getting sued for 10's of millions for a traffic accident, if for no other reason than she doesn't have that much money - Tesla on the other had does have the money and there will be lawyers willing to take it on. Also, your spouse will not cause many of those accidents per day, Teslas would if they were widely deployed. Lastly, even if someone manages to ban your spouse from ever driving again, I think even you would agree that would me of much lesser impact to society than if Tesla FSD is banned. Do you still not see the difference between a large company and a single person?

 

[brucet999]

This evening I ran into black ice and had a moment... but I successfully avoided hitting another car that had stopped in front of me. I held the brake down and used the ABS to steer around it.

Can you 'splain to this SoCal native how one uses ABS to steer?

 

[msnow]

Great. You're exactly the person who can answer the questions I have but I don't know what question you answered above so please answer these:

1. How many years (or months) are we away from vehicles doing more than AP1.0? (i.e. anything more, such as stopping at lights/stop signs, or any one thing more than AP1.0 can do?)

2. How many years (or months) are we away from vehicles self-driving with a person in the seat overseeing and taking over whenever the vehicle requires?

3. How many years are we away from autonomous driving vehicles?

I didn't put "or months" in brackets for number 3 because I think that must be more than a year away, and that's why you said many (twice) but how many is many, many? But maybe you will tell me all of the above are more than one year away? I don't think so but you have much more knowledge than me on this issue so I look forward to your answers.

I said "many" twice because of the variables. For example no one knows if the technology is capable of consistently and safely achieving #1. Don't forget stop light/stop sign recognition was supposed to be part of AP 1.0 originally until the goal post was moved after Tesla realized it lacked the technological capability.

Point two, the regulators. This is perhaps the biggest obstacle to quick adoption. In the U.S. they are just now pulling together the knowledge to establish standards. Codifying standards is a time consuming process. Secondly and most importantly, regulators believe their mandate is to protect the public they serve so by definition the are slow and plodding.

I don't have a crystal ball but my sense is we are several years away from full automous driving if not longer and that assumes that the technology is capable enough to satisfy regulators.

 

[SW2Fiddler]

Some of these examples - disabled cameras, buzzards, etc. if that happens the car should pull over

Now I think maybe I understand the lag in Supercharging I-10 west of central TX...

 

[gavine]

I am not a Tesla owner but like the idea of my car driving me to work every day.

I would think that most owners would like their car to drive them to work everyday more so than any other reason, at least initially. With that in mind, wouldn't Tesla be able to "learn" your daily commute with you driving it manually a few times (several times...whatever) and then be fully autonomous for that going forward? I would think that would be pretty easy to accomplish with the full hardware suite and would be a great thing. I think people would be willing to pay for the full self-driving suite if they knew that it could learn their commute and drive them to work.