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

60 mph; excellent road markings and AP2 tried to throw the car at the median

This site may earn commission on affiliate links.
1. 40% accident rate reduction was with AP1
2. 40% accident rate reduction was observed with Teslas with AP1 vs. Teslas without AP1. But Teslas before AP1 ALSO LACK AEB! I believe this 40% accident rate reduction is based solely on the simultaneous introduction of AEB.

#2 is a key point. We really have no idea how the autosteer/TACC combo compares to cars without the combo. Any statistics get washed out by AEB. My bet is that if you divide the cars with AP 1 hardware into two groups (I) the ones that don't have the AP option purchased and (ii) the ones that do, the first group has better accident statistics.

Same thing with a comparison of AP 2 hardware cars that do and don't have the EAP option activated.
 
  • Like
Reactions: Matias
screen-shot-2017-08-31-at-12-58-53-pm-png.244867

Well.

They ship something at scale, anyway.
 
Hi, I was trying to learn more about the eight camera system that your 2013 Audi A6 used for active lane assist, but I'm having a hard time finding any detail on an Audi system with those specs. Can you share a link, or maybe you have some paperwork at home about it's eight camera active lane assist? I'd love to know more! Thanks!!

Audi does not use eight (is it 8 or 6-7 anyway?) cameras for the lane assist (just one for that), they do use them for other things (night vision, 360 view etc.), though.
 
As ever the hyperbole kicks in.

No the car did not try to kill me.
Nor will it kill anyone if you drive the car as instructed byt the manufacturer with your hands on the wheel.
If I hadn't been paying attention I might have scraped the front corner is all.

The Autopilot a very good driver assistance system (arguably the best available today - where's the comparative reviews btw) and mostly works well, but it is a bit of a twitchy bugger in the current release that I have; which I notice is already three released builds behind the latest so no doubt it is already improved.

Overly agressive application of steering at speed is the bit I would really like to see them dial out soonest, so it become less of a test of how alert you are (although you should be anyway).

There's "hands on the wheel" so AP2 continues to work and "hands on the wheel" to prevent random acts of craziness where the car attempts to dart into another lane of traffic or a ditch. The two are very different levels of steering AND attention. When I'm driving an ordinary car, I don't hold onto the steering wheel for dear life anticipating that the car may, without notice, dart into another lane of its own volition. With a Tesla AP2 vehicle, you damn sure better make that assumption.

"Best available" option for self-driving doesn't really have ANYTHING to do with whether the Tesla AP2 system is safe. My AP2 is NOT SAFE without an extremely careful and attentive driver that knows and expects random, erratic driving behavior. YMMV! :eek:
 
Hey @NerdUno - the way my Disagree count is stacking up I think you must be going for the record for Disagrees in one thread ;)

I guess the difference between us, is that I am disappointed in the delays getting AP2 finessed but have faith Tesla will sort it, you kinda seem pretty sore about it and I can appreciate that.

I don't have FSD so I can't be disappointed about that, but if I did I think I would be getting somewhat pissed by now and looking for a statement from Tesla as nothing whatsoever has been released on that front.
 
aww now you gone and spoiled it :cool:

Actually I think Tesla have lowered my car's priority for updates as I haven't had any of the newer releases since I started this thread. That'll teach me to complain.


I feel the same way. Word was .32 and anything above had EAP updates meanwhile I had outstanding investigations into EAP failures and was stuck on .28 for 6 weeks. Finally had something minor looked at at my SC and got .34. I don’t know how long Tesla would have kept me at .28 otherwise.
 
  • Informative
Reactions: zmarty
So guys.. for us model 3 reservation holders... what's your opinion on EAP.. should we get it?

The problem with not getting it is that without EAP you will have only "dumb" cruise control, i.e. the kind that pays no attention to the car in front of it. Adaptive cruise control - what Tesla calls TACC - increases safety and convenience. Dumb cruise control decreases safety, especially in moments of inattention or drowsiness. In congested areas, it's impractical to engage.

So if you assume that the other features of EAP won't evolve to more than a hobbyist's toy - the current state, in my opinion - while you own the vehicle, the question is whether it's worth $5,000 to get adaptive cruise control. That's a significant chunk of cash, in percentage terms more expensive for the 3 than for the S. Anyone willing to spend over $100k for a car isn't likely to be too concerned about another $5k, but the supposed mainstream buyer the 3 is aimed at is likely to be more value conscious.
 
  • Informative
Reactions: xav-
Audi does not use eight (is it 8 or 6-7 anyway?) cameras for the lane assist (just one for that), they do use them for other things (night vision, 360 view etc.), though.
Haha, yea, I know... I was just passively poking a little at that guy. :p
There best system to date I believe has 6 cameras: 1 rear camera (backup and 360 view), 2 side cameras (360 view), 1 forward thermographic camera, 1 forward 360 view camera, 1 forward facing camera for all driver assist items.

For what it's worth, I haven't driven their cars, but on paper their driver assist features look pretty amazing.
 
  • Like
Reactions: AnxietyRanger
Learning to understand lines in various light situations should be an intern's project. Especially with so many cars driving around recording on so many cameras and sensors.
I'm not a programmer but can see multiple approaches for the mothership to understand lines an lighting, or even map faux markings that might be misinterpreted within current level of line recognition consistency.
Machine learning, if it existed in this particular dimension of Earth, would sort the line thing pretty much overnight. And it should probably be implemented before any more ambitious AI features are actively pursued. Understanding just a few messy roads and being about to hold lines over them would help so much towards a cross-continental FSD.
 
s/w 17.28.4 cf44833
[AutoPilot], when it's good it is very good but there are still too many occasions when it will behave completely unpredictably and this ruins its utility.
Sadly, AutoPilot will never be safe, period. Whether Tesla Autopilot, or any other car manufacturer's Level 3 self-driving. Imagine if Tesla team did so well, that your EAP only tried to hurt you like that once every 10,000 miles. How much do you want to bet that you wouldn't be ready for it? Months of driving during which it performs flawlessly will only build up your trust and complacency, then it drives you into a median - game over. Google (now Waymo) has realized that 3 years ago as described in this article: Robot Cars Can’t Count on Us in an Emergency
 
  • Informative
Reactions: xav-
Yes, this happens. AP2 - two versions ago - almost threw me into concrete blocks as well when it started following a vague, black tar-line on the road instead of beautiful, newly painted white lines on both sides of the lane.
AP1 exhibits similar behavior when the sun shines just right, except where it has good GPS data (i.e. it knows many other Tesla's when driving through this exact spot, did not follow the tar line).
 
  • Helpful
Reactions: AnxietyRanger
Haha, yea, I know... I was just passively poking a little at that guy.

@Moi? That's ok, I can use a good poking every so often.

Just to be clear, I wasn't pointing to Audi circa 2013 as the Holy Grail. My point was that the features Tesla hypes with non-standard names like autopilot, autosteer, and tacc aren't unusual and don't outperform other high-end (and even some low end) cars. Those who believe they do should test drive a few.

The features in the fantasy version of EAP, such as autonomously following a route across multiple freeways, using four or eight cameras, summoning that steers around obstacles to find the owner, etc, really are special. But dreams and reality are, sadly, two different things.