Live data from Hacker News

Study: Tesla Autopilot misleading, overestimated more than similar technology

mercurynews.com

71–80 of 159 posts

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#71

"Survey participants were asked questions including whether they thought it was safe to take their hands off the steering wheel" I think this is misleading, Autopilot never told it's safe enough to let hands off the steering wheel. To this extend of course there is no difference than other solutions.

What's misleading? If they think it, they think it, and the intentional implications of advertising are just as important as the words it uses.

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#72
post #5

Earlier quoted context omitted.

I see the use of “autopilot” name bring brought up a lot. Thinking about what an autopilot in an airplane does, labeling Tesla’s system autopilot is actually pretty apt. It’s unfortunate that there’s a disconnect between what autopilot means colloquially vs what it actually does on a plane.

> what an autopilot in an airplane does ...Flies the plane safely with zero input from the pilot (hands off the yoke) indefinitely? Lands the plane on its own [1]? I'm well aware that autopilot systems need input directions for desired altitude and air speed, but the semantic distinction people try to draw between "real" autopilot and Tesla's autopilot capabilities have never made sense to me. [1] https://en.wikipedi…

I think what you're missing is that Tesla's Autopilot is not done. It's clearly labeled "Beta" when you enable it in the car.

The goal of Tesla's Autopilot is to take you from your garage to your parking space with zero input. Therefore it's exactly the same as a plane's which takes you from runway to runway.

Just because it's a work-in-progress doesn't mean it's named wrong.

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#73

Earlier quoted context omitted.

> they're helping to lay the ground work for safer roads by forcing the industry to integrate more and more automated safety systems. They focus more on driver-aid systems that encourage not paying attention. Their automated safety systems aren't anything special compared to what most manufactures offer.

> Their automated safety systems aren't anything special compared to what most manufactures offer. When I'm trying to figure out which of two really vocal, opposed camps might be more right on some issue, I look for the really hyperbolic statements that are easily refuted but still repeated. 2016: A Tesla's Forward Collision Warning System warns the driver about an impending collision he can't even see yet and starts…

Individual reports are not data.

https://www.iihs.org/topics/advanced-driver-assistance

"study found that systems with forward collision warning and automatic braking cut rear-end crashes in half, while forward collision warning alone reduces them by 27 percent. The autobrake systems also greatly reduce rear-end crashes involving injury."

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#74

Earlier quoted context omitted.

My Model 3 did exactly that this morning. I was in the act of changing lanes, expecting the car behind me in the target lane to either maintain speed or slow to avoid hitting me. Instead it sped up, my car alerted me, and as I glanced in the side view to see the car approaching, the steering wheel gently tugged me (us??) back into our lane, avoiding collision. I’ve never felt so grateful to an object before.

Whereas, i presume, in your previous non-Tesla cars you were regularly getting into pileups?

How reductive. Let's try looking at it as getting closer to accidents before managing to fix the issue. So now with this system, one-in-a-million odds of that type of accident might become one-in-ten-million odds.

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#75

"Cruise Control" doesn't mean I can ignore the ~3200lbs of sports car I drive, just like "Tesla Autopilot" doesn't mean drivers can ignore the ~5000lbs of battery under them. As an American who spends considerable time driving cars in various performance events (even some fun wheel to wheel endurance stuff) I think I can safely estimate at least ~25% of drivers on the road today, probably shouldn't be. Tesla is makin…

>Tesla is making progress, and regardless of whether they survive these various courts of public opinion, they're helping to lay the ground work for safer roads by forcing the industry to integrate more and more automated safety systems.

They're also wilfully misrepresenting the quality and safety of their technology to their customers, presenting intentionally misleading statistics to the media about it, and attempting to suppress all criticism by insisting it's part of some big oil / short seller conspiracy.

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#76
post #44
post #26

Earlier quoted context omitted.

It's not about how easy it is to put your hand on wheel so that it can detect a bit of force (you can use bottle of water to do that). The difficult part is to stay focused when doing that. If you spent 30 minutes driving to work on a highway, and your car was able to drive itself every day for 2 months, how likely are you to take any reasonable action on 3rd month when something unexpecting happens?

So the problem is the human attention span? Cars without any always-on computer aids such as automatic braking must be really dangerous then.

Driving a car is absolutely the most dangerous activity most people regularly participate in. I don't think you're saying anything controversial there.

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#77
post #38

Earlier quoted context omitted.

If tesla’s autopilot were available to all vehicles for use only on highways (which is where most Tesla owners only use autopilot), the number of highway deaths due to inattentive drivers causing rear-ending accidents would drop to near zero overnight. It is so vastly superior to distracted humans on a highway, it is a “seatbelt” type revolution in terms of human safety. But Tesla doesn’t seem to push this specific u…

That is nothing unique to Tesla. If they were serious about full self driving there wouldn't be any reason to bother with autopilot.

I'm amazed that you can agree it would drop rear-endings to near zero, and then immediately say there is no reason to bother with it.

Do you think a level 4/5 system would be done already if they hadn't made a level 2?

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#78
post #5

Earlier quoted context omitted.

I see the use of “autopilot” name bring brought up a lot. Thinking about what an autopilot in an airplane does, labeling Tesla’s system autopilot is actually pretty apt. It’s unfortunate that there’s a disconnect between what autopilot means colloquially vs what it actually does on a plane.

> what an autopilot in an airplane does ...Flies the plane safely with zero input from the pilot (hands off the yoke) indefinitely? Lands the plane on its own [1]? I'm well aware that autopilot systems need input directions for desired altitude and air speed, but the semantic distinction people try to draw between "real" autopilot and Tesla's autopilot capabilities have never made sense to me. [1] https://en.wikipedi…

Real plane autopilot is basically lane-following.

Tesla autopilot is theoretically capable of indefinite highway driving and emergency 'landings' (on the shoulder).

Pretty close in actual capability. The difference is the environment.

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#79

Earlier quoted context omitted.

> So it's steering... but you're supposed to leave your hands on the wheel anyway? But not actually steering, just resting there just in case? I'm not sure how realistic this is in terms of human behavior. I have an Acura, so no autopilot, but it allows my car to steer, accelerate, and brake automatically within its lane. Also, if I don't apply control to the steering wheel (let alone have my hands off of it) for mor…

I was worried about the “diffusion of responsibility” issue, so I rented one on Turo for the day and drove around 200 miles on freeways in Northern California. The user interface is really good, you learn pretty quickly the right amount of force to hold the wheel - then you can let it “autosteer”, but taking control is instantaneous, you just make the movement you want and the car disengages autosteer instantly. I pr…

Tesla is betting they can do full self driving without LIDAR. It isn’t clear that this bet will pan out.

Re: Study: Tesla Autopilot misleading, overestimated more than similar technology

#80

I know that the name is a good one, but 'Autopilot' is a troublesome name. I think a better way forward would be to do something fun with the 'Autopilot' name, e.g.: "Zanzibar (formerly known as Autopilot)" Nobody need know what "Zanzibar" actually is other than some place a long way from California. You could easily talk about your Tesla to a non-owner saying 'yeah, it's got Zanzibar...' or to a fellow owner 'did yo…

I'm confused why there is such a dissonance on the definition of "autopilot." Most people are fully aware that commercial planes have at least one human pilot in addition to an "autopilot," and I think the passengers on the plane would be very unwilling to fly if the human pilots weren't present. "chauffeur" seems like it would be misleading for partial self-driving software, but "autopilot" seems spot on: helpful, but not good enough to replace the human.
Post reply on HN