[bro1999]

http://fortune.com/2017/10/25/steve-wozniak-tesla-autopilot-apple/

Echoes that GM engineer that called Musk "full of crap".

 

[Barry]

I think it goes without saying that grand claims about what AP will be able to do in the future are highly speculative, especially with respect to timing. Anyone who relies on those claims to make important decisions is foolish. And arguing that they are unrealistic makes people seem foolish since the argument is based on the premise that anyone would rely on them in the first place.

But the facts about what Tesla has already achieved with it's driver assistance features speak for themselves.

 

[bro1999]

I think it goes without saying that grand claims about what AP will be able to do in the future are highly speculative, especially with respect to timing. Anyone who relies on those claims to make important decisions is foolish. And arguing that they are unrealistic makes people seem foolish since the argument is based on the premise that anyone would rely on them in the first place.

But the facts about what Tesla has already achieved with it's driver assistance features speak for themselves.

Any lead Tesla had in self driving tech is quickly evaporating. And they may very well have been passed, looking at GM's AV efforts.
https://www.bloomberg.com/news/articles/2017-10-25/can-tesla-make-up-for-autopilot-s-lost-year

 

[Barry]

^ Yes, many companies are advancing this technology. It will be interesting to see how it all plays out. But as far as what Tesla has put into the hands of regular consumers actually on the road, Tesla has been the one breaking new ground so far.

 

[Loboc]

Tesla has been the one breaking new ground so far.

I think that comparing AP2.5 to GM Super Cruise, SC is a better end-user experience. Volvo, MB and others also have superior tech to AP.

As a software developer and long-time car and tech guy, I think Tesla is barking up the wrong tree. Level 4 and especially Level 5 are not attainable by enhancing Level 2. Levels 4 and 5 require way better computers and programming (learning AI) to get there. It can't be done using a hacked nVidia board and old-school chips-plus-programming technology.

Tesla themselves are getting the hint. They are on their 4th computer iteration just for AP. I see a huge class-action in their future when they cannot deliver Level 4 with their current hardware suite. Heck, they can't even deliver AP2 full feature set! Calling it AP2.5 doesn't make it work better.

Here's a recent story about AP and it's limitations.
https://www.theverge.com/2017/10/24/16504038/tesla-autopilot-self-driving-update-elon-musk

 

[Breezy]

But as far as what Tesla has put into the hands of regular consumers actually on the road, Tesla has been the one breaking new ground so far.

Tesla customers weren't regular consumers. Tesla's customers were unusually forgiving, but that's changing.

Tesla is going to have to adjust to being routinely sued by their customers like the rest of the industry.

 

[DonC]

I agree with Loboc that AP is never getting to Level 4 or Level 5. Just isn't enough hardware. You can assign talented software engineers the task of figuring out what a smudge is, but at the end of the day the smudge remains a smudge. The real solution is better hardware that can resolve what that smudge actually is.

While this isn't surprising, I didn't know Tesla was collecting $8K by promising to deliver the capability. The criticism seems justified.

 

[viking79]

They shifted the weight to AP to reduce their liability I think, of the 8k for Full Self Driving feature, 5k is for AP and 3k for FSD, so AP is a very expensive feature that they can deliver, the FSD is much less expensive but probably won't be able to deliver (in a timely fashion). Honestly, reading the FSD description on their website it clearly describes a level 5 car, but they have a disclaimer that they might not be able to deliver all the features due to regulations. They also give no timeline. I sense they will receive a class action lawsuit over that feature.

 

[ace]

I think that comparing AP2.5 to GM Super Cruise, SC is a better end-user experience. Volvo, MB and others also have superior tech to AP.

As a software developer and long-time car and tech guy, I think Tesla is barking up the wrong tree. Level 4 and especially Level 5 are not attainable by enhancing Level 2. Levels 4 and 5 require way better computers and programming (learning AI) to get there. It can't be done using a hacked nVidia board and old-school chips-plus-programming technology.

Tesla themselves are getting the hint. They are on their 4th computer iteration just for AP. I see a huge class-action in their future when they cannot deliver Level 4 with their current hardware suite. Heck, they can't even deliver AP2 full feature set! Calling it AP2.5 doesn't make it work better.

Here's a recent story about AP and it's limitations.
https://www.theverge.com/2017/10/24/16504038/tesla-autopilot-self-driving-update-elon-musk

I agree that AP2 has its limitations and is unlikely to make it to level 5, but I have to give tesla credit for even attempting to deploy a system for consumer use. Super cruise is ok for what it is and its usefulness depends on the type of driving you do, but I would like to see more that just non-divided highways (I also don't like that it will require an additional subscription beyond 3 years). My issue with GM, Volvo, MB, etc. is that although they may have better software than tesla, the incorporation into consumer vehicles is painfully slow, that all of the EVs being developed are either compact (cars or SUVs) or performance oriented sports cars, and other then tesla there is no real thought to the charging network.

 

[Upper5Percent]

I agree with Loboc that AP is never getting to Level 4 or Level 5. Just isn't enough hardware. You can assign talented software engineers the task of figuring out what a smudge is, but at the end of the day the smudge remains a smudge. The real solution is better hardware that can resolve what that smudge actually is.

While this isn't surprising, I didn't know Tesla was collecting $8K by promising to deliver the capability. The criticism seems justified.

Criticism?...Try class action lawsuit...
http://www.businessinsider.com/tesla-class-action-lawsuit-autopilot-2-2017-4

http://fortune.com/2017/04/20/tesla-lawsuit-autopilot/

 

[Qinsp]

I had an interesting discussion with a contractor yesterday. His dad (60's) just bought a loaded Model X.
His dad loves the car, and took everyone out to dinner. He tried to demonstrate the Autopilot, and after 3 tries with it swerving on the freeway towards barriers, he said there must a problem and he will take it in for service, but still loves the car.

Perhaps it takes more practice, perhaps something was wrong with his, or perhaps freeway had confusing lane markers.
But it does not matter. He likes the car anyways. And isn't that what is important?

 

[srmarti]

I had an interesting discussion with a contractor yesterday. His dad (60's) just bought a loaded Model X.
His dad loves the car, and took everyone out to dinner. He tried to demonstrate the Autopilot, and after 3 tries with it swerving on the freeway towards barriers, he said there must a problem and he will take it in for service, but still loves the car.

Perhaps it takes more practice, perhaps something was wrong with his, or perhaps freeway had confusing lane markers.
But it does not matter. He likes the car anyways. And isn't that what is important?

I recall seeing a video of a Model S hitting one of those barriers on YouTube. Purported to be on autopilot.

More important to me would be Teslas not crashing or scaring the hell out of the driver or those around them. At least he decided to turn it off before something very bad happened.

 

[Shivetya]

I thought Tesla was the only auto manufacturer going without LIDAR, is that true?

youtube used to be replete with Tesla AP failure videos

 

[bro1999]

I thought Tesla was the only auto manufacturer going without LIDAR, is that true?

youtube used to be replete with Tesla AP failure videos

Used to be? Maybe Tesla threatened legal action and had them all removed. Just like many of the Model 3 videos that show them in a negative light.

 

[Qinsp]

I thought Tesla was the only auto manufacturer going without LIDAR, is that true?

youtube used to be replete with Tesla AP failure videos

Elon made some kind of comment about not needing LiDAR because Tesla can get by with RADAR. There apparently were many chuckles from engineers at that comment.

I'm no expert, but IIRC, the lower the frequency of electromagnetic radiation, the less resolution you get when using it as a sensor. RADAR lacks the frequency needed for safe automotive directional control. Everybody is using LiDAR because of that.

 

[Mister Dave]

Everybody is using LiDAR because of that.

Almost everybody.

 

[Steverino]

Almost everybody.

haha, a +1 on that wry observation, Mister.

 

[Mister Dave]

haha, a +1 on that wry observation, Mister.

I suspect they'll learn in time. Problem there is that the hardware is not pre-installed and can't be unlocked OTA.

 

[srmarti]

Elon made some kind of comment about not needing LiDAR because Tesla can get by with RADAR. There apparently were many chuckles from engineers at that comment.

I'm no expert, but IIRC, the lower the frequency of electromagnetic radiation, the less resolution you get when using it as a sensor. RADAR lacks the frequency needed for safe automotive directional control. Everybody is using LiDAR because of that.

The counter argument is light get blocked by rain, fog, snow. Both might be needed.