Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Autopilot lane keeping still not available over 6 months after delivery

This site may earn commission on affiliate links.
On the way to work this morning an 18-wheeler didn't see me and pulled over on top of me. Hit the brakes and the horn and avoided death, but it made me wonder - what would auto-pilot have done in that circumstance? I'm guessing nothing. The driver is still responsible for taking over and saving his own life. I wonder if they'll have some sort of instructions for auto-pilot, or maybe a demonstration video showing instances like this? If they just put it out there and say "auto-pilot is not responsible for reacting in emergency situations" that's going to be a problem. Or if they don't say anything - even bigger problem. I wonder if this is a "corner case" they have programmed it for? We need to have a clear understanding of what the car will and will not do in emergencies.

First, I, too, am glad you're OK.

Second, I don't see why this is an autopilot problem. Aren't we always expected to be attentive while driving, even with autopilot engaged?
 
Autopilot != Autonomous driving

Lots of people still seem confused about this.
I am not confused at all. And in emergency situation, when I can (enough response time) I would take over from AP and attempt to avoid an accident.

Quote: Would the auto-pilot do what I did given the same condition?

My question was not to suggest that Autopilot should be able to do the emergency avoidance 100% of the time (specially where other human drivers are involved), but only asking a question on what might happened. And this is not to raise the expectation in anyway. Just curious.
 
I am not confused at all. And in emergency situation, when I can (enough response time) I would take over from AP and attempt to avoid an accident. My comment was not to suggest that Autopilot should be able to do the emergency avoidance 100% of the time (specially where other human drivers are involved), but only asking a question on what might happened. And this is not to raise the expectation in anyway. Just curious.

My comment wasn't targeted at only you.

But to your question, I don't think anyone outside of Tesla can predict what the algorithms are tuned for; considering the car can not (and with current hardware will never be able to) change lanes on it's own and the blind spot monitoring software sucks, IMO I don't see how autopilot would have prevented a side impact whatsoever.
 
I am not confused at all. And in emergency situation, when I can (enough response time) I would take over from AP and attempt to avoid an accident.

Quote: Would the auto-pilot do what I did given the same condition?

My question was not to suggest that Autopilot should be able to do the emergency avoidance 100% of the time (specially where other human drivers are involved), but only asking a question on what might happened. And this is not to raise the expectation in anyway. Just curious.

I second this. We can find out by trial and error, kind of like we all learned TACC would stop the car for you. If you weren't scared the first time you depended on it to stop and not rear-end the car in front of you then you need to apply for secret agent duty because you have nerves of steel. I'd rather not learn how auto-pilot works (not autonomous pilot, auto-pilot - what we're getting) by experience. If you depend on it to hit the brakes when a car starts changing lanes into you and it doesn't do it, you're in trouble. And the car has ultra-sonic sensors reaching out to 16 ft, 360 degrees fully around the car (doesn't it?), so yes it could prevent a side-impact like this. But will it? That's the question I hope Tesla answers with some sort of info provided before roll-out.
 
I have shivers just reading this.

I'm guessing that lane keeping will be most useful in long distance lonely cross-country highway driving -- not so much in the cities. Better than nothing.

Once they release it, people are going to use it - everywhere. And invariably people are going to be paying less attention. Going back to spending hours in the car paying attention to staying in your lane would be worse than MS owners going back to driving an ICE. Why do you think all other manufacturers require you to have your hands on the wheel ect. Because the system isn't perfect and they don't want the liability if it fails.

So Tesla needs to make sure it works well before releasing it.
 
I am frightened by auto pilot. I don't know if it will be good or bad.

I commuted on a motorcycle for years (lane splitting is not against the law in California). I stopped due to cell phones. I my experience I believe 30% of the commuters were texting or emailing while they drove to work.

Either auto pilot will be safer or it will make it worse - I don't know what will happen.
 
Unfortunately, I'm afraid one of the side effects of Autopilot will be less attentive drivers.
I don't know. I feel like I've heard that my entire life. "ABS will mean people don't know how to use the brakes." (Well, the irony there is that older folks still pump ABS brakes and limit it's usefulness) "Navigation will mean people won't learn the neighborhoods." (I feel it's the opposite, I now intimately know all the little side streets I didn't pay attention to with my paper maps) "Airbags will kill the driver in any accident" "Retracting door handles will cut your fingers off"

I think people feared every single advance in the vehicular world. Steering wheel audio controls, power windows, digital stereos, etc. I wouldn't be shocked to find an article from Horseless Buggy Monthly that talks about how switching from candlepower to electric light in the headlights will kill everyone and bring on the end of days.
 
I don't know. I feel like I've heard that my entire life. "ABS will mean people don't know how to use the brakes." (Well, the irony there is that older folks still pump ABS brakes and limit it's usefulness) "Navigation will mean people won't learn the neighborhoods." (I feel it's the opposite, I now intimately know all the little side streets I didn't pay attention to with my paper maps) "Airbags will kill the driver in any accident" "Retracting door handles will cut your fingers off"

I think people feared every single advance in the vehicular world. Steering wheel audio controls, power windows, digital stereos, etc. I wouldn't be shocked to find an article from Horseless Buggy Monthly that talks about how switching from candlepower to electric light in the headlights will kill everyone and bring on the end of days.

Speak it, brother!
 
And that's really the rub, isn't it? If ALL cars were instantly converted to an auto-pilot like system, and even better, were aware of each other, there'd be no problem. The challenge is the irrational, sudden and inexplicable movement of another vehicle - most likely human - that cannot be pre-programmed to avoid.

This is deeply naive in my opinion. We have no idea what the behavior will be of auto-pilot like systems in reaction to each other. Especially when you consider that different cars are going to have different implementations and different behaviors. Driving being a 2D problem is somewhat easier to solve that flying an airplane, but I think that the situation with Traffic Collision Avoidance Systems in airplanes is instructive. Granted TCAS is not executing actions to avoid collisions on its own, it is still specifying the actions that should be taken and it still sometimes can provide instructions that would result in a collision. This is despite the fact that the systems are implemented per a standard as far as what the behavior should be, so the suggested reactions are coordinated between implementations. Something that at this point isn't going to be the case with auto-pilot type systems.

In short complex systems like these may have unexpected behaviors in the real world. So no I don't believe that a complete conversion to auto-pilot like systems of the entire fleet would actually eliminate all accidents.
 
This is deeply naive in my opinion. We have no idea what the behavior will be of auto-pilot like systems in reaction to each other. Especially when you consider that different cars are going to have different implementations and different behaviors. Driving being a 2D problem is somewhat easier to solve that flying an airplane, but I think that the situation with Traffic Collision Avoidance Systems in airplanes is instructive. Granted TCAS is not executing actions to avoid collisions on its own, it is still specifying the actions that should be taken and it still sometimes can provide instructions that would result in a collision. This is despite the fact that the systems are implemented per a standard as far as what the behavior should be, so the suggested reactions are coordinated between implementations. Something that at this point isn't going to be the case with auto-pilot type systems.

In short complex systems like these may have unexpected behaviors in the real world. So no I don't believe that a complete conversion to auto-pilot like systems of the entire fleet would actually eliminate all accidents.

Well don't leave me hangin - do you think it would make it worse or better? If better - how much? Or are you saying its a total and complete crap-shoot and could go either way?

All the experts I've read say conversion of cars to autonomous driving would be much safer - and I think (don't quote me) Elon even said someday humans won't be allowed to drive cars. I've heard that from multiple people, so maybe it wasn't Elon.