-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FIRMWARE UPDATE! AP2 Local road driving...and holy crap
- Thread starter Skalinator
- Start date
HumanGenome
Member
The bridge issue is a known issue and in that video the driver is unsafely driving 50mph when the rate of traffic is much higher. He should not have been using autosteer there at all. Also not sure where you saw him nearly getting rear-ended, there's no proof of that at all in the video.
Joe F
Disruption is hard.
Anecdotal data? Until you find yourself on the losing side of a statistic.
When it happens to you, it's not anecdotal data. It's real. Trust AP2 right now, but know its limitations, and only use it under ideal conditions. Even Elon cautioned that. But I guess you discount him, too.
The NHTSA report compared data for "all MY2014 through 2016 Model S and 2016 Model X vehicles equipped with the Autopilot Technology Package, either installed in the vehicle when sold or through an OTA update." https://static.nhtsa.gov/odi/inv/2016/INCLA-PE16007-7876.PDF. I don't have the exact number but that's a lot of vehicles and vehicle miles. I am confident that if it did not rise to the level of statistical significance the NHTSA would have said so.
I agree that AP2 should be used as instructed by Tesla and that learning its limitations and any quirks is valuable. Having said that, I stand by my statements about the lack of value of anecdotal data in assessing overall safety of the system. You are of course free to have a different opinion.
It happend to me at 75mph multiple times on TACC. Some are reporting decelerations from 75 to 45.
Unfortunately I don't have a video.
Unless I am mistaking they are comparing Automatic Emergency Braking (AEB) to non AEB vehicles. Of course you will get better results with AEB. Even with AEB having false positives, you will get better results as the person behind you might be able to stop in time or maybe his car has an AEB as well.
Check the mirror. He also says it in the description. Hard to say how close it was.
Maybe?
Figure 11 shows the rates calculated by ODI for airbag deployment crashes in the subject Tesla vehicles before and after Autosteer installation.
I think the biggest reason for the airbag to be deployed is a frontal collision. Also the report was done because of a frontal collision.
Figure 11 shows the rates calculated by ODI for airbag deployment crashes in the subject Tesla vehicles before and after Autosteer installation.
I think the biggest reason for the airbag to be deployed is a frontal collision. Also the report was done because of a frontal collision.
NerdUno
Member
Comparing AP1 development to AP2 is apples to oranges. Tesla used Mobileye technology and hardware for AP1. It already had been under development and testing for over 10 years. Nothing even close to comparable exists for AP2 hardware. Many AP2 purchasers didn't know the Mobileye hardware and software would be disappearing. Tesla had made it sound like they would simply be embellishing existing software in house for AP2 WITH ENHANCED HARDWARE not replaced hardware.. It turns out they were starting all over on a completely different platform. When some are suggesting the development cycle will be similar, may I inquire what factual basis you have for your assertions??
Last edited:
I don' think so. AEB effectiveness has been studied extensively. A recent IIHS study concluded that AEB plus Forward Collision Warning reduced overall accidents by only 6%, which did not even rise to the level of statistical significance. http://orfe.princeton.edu/~alaink/S...s/IIHS-CicchinoEffectivenessOfCWS-Jan2016.pdf (see pages 1 and 15.)
Compare the AP2 buyers with TMC join dates in late 2016 and in 2017 compared to the buyers who already had an AP1 or classic, and traded up to AP2. Are any of the latter similarly outraged as the former are? I don't think I see any.
What this whole episode really illustrates is that Tesla is moving beyond the early adopter crowd who tend to be more informed and tolerant of the development of the technology in the vehicle.
It wasn't a sample..It was just ALL of the cars.
What this whole episode really illustrates is that Tesla is moving beyond the early adopter crowd who tend to be more informed and tolerant of the development of the technology in the vehicle.
It wasn't a sample..It was just ALL of the cars.
My point is that only anecdotal information of no statistical value is being cited in support of sweeping claims that AP2 is unsafe. Anecdotes are just that -- anecdotes. Not only is anecdotal information statistically meaningless, there is a track record of similar types of information being incorrect in a very similar situation (AP1).
Tesla also has a track record of relying on solid data to roll out AP in a safe manner. I believe their data and track record have proven reliable, and they appear to be exercising restraint and caution in rolling out AP2. In the past, anecdotal information about shortcomings of Autopilot have been not only totally unreliable in predicting overall system safety, but very misleading. So I believe Tesla's data and track record are more trustworthy than the anecdotes. Even if you don't accept that, making blanket assertions that AP2 is "unsafe" based solely on anecdotal evidence is unsupported and, in my opinion, irresponsible.
The problem with the anecdotal evidence is not only that it is statistically meaningless but ignores that the baseline -- average human drivers -- are very unsafe. They crash their cars in all sorts of ways with alarming frequency. AP2 doesn't need to be perfect, it just needs to be better than average drivers in similar circumstances.
NerdUno
Member
I disagree. AP2 buyers are much closer to being early adopters than AP1. Those cars had a proven platform with a 10 year track record. I'd be willing to bet that Tesla only pulled off that demo video by relying on some Mobileye technology that they couldn't actually resell in AP2 vehicles.
Last edited:
Exactly. AEB plus FCW is shown to reduce rear-end collisions by 39% (not clear how much is due to FCW versus AEB). But rear-end collisions are only about a quarter of total accidents, and when the IIHS looked at the big picture it found that AEB plus FCW reduced overall accidents by only 6%, which did not reach the level of statistical significance. Ironically, adding AEB to FCW actually resulted in more accidents (12% reduction with FCW alone v. 6% reduction with FCW and AEB) and only FCW alone reached the level of statistical significance. So based on the IIHS's Jan. 2016 report, it is not clear whether AEB by itself reduces overall accidents at all.
Sorry, there is no basis for asserting that the 40% reduction in airbag deployment accidents after Autosteer is attributable to AEB.
I am curious what else do you think it is?
I'm also gonna leave this here if someone is interested. It shows that most airbag deployments are frontal crashes by a lot. I don't have time to crunch the numbers.
Edit:
IMHO what the report showed was expected.
Attachments
Last edited:
I had the same understanding as you in November/December. I posted many messages saying I thought we might get AP1 parity by the end of the 2016 or beginning of 2017. Many tried to tell me that we'd have full EAP but I never thought it would be out that quickly.
I'm still eagerly awaiting additional AP2 functionality but have been happy to get TACC and other features added every couple of weeks since taking delivery in December.
Last edited:
There are over 1 billion cars on the road throughout the world, so I think the relative handful of AP1 and AP2 owners all qualify as very early adopters in my book.
I think others have addressed the other points in your post so will leave it at that.Let me put it this way. As mentioned upthread, I predict that once the data is in, as with AP1, AP2-enabled cars will be shown to be much safer than driving without it. The exact mechanisms by which this will take place, and which components of AP2 make the biggest contribution, are of less importance to me personally, if that can even be determined.
I also predict that no one on this thread or any other saying that AP2 is "unsafe" will acknowledge that they are wrong, just like, to the best of my knowledge, no one making similar claims with AP1 right up until the NHTSA report issued has acknowledged that they were wrong. I would love to be proven incorrect on this second prediction but am not holding my breath.
Let me acknowledge it right now. There is no way AP2 will get worse with more data, but as it is right now, I won't use it. Personally, it feels "UNSAFE" and takes more attention to use it than not to use it.
Similar threads
