TMC is an independent, primarily volunteer organization that relies on ad revenue to cover its operating costs. Please consider whitelisting TMC on your ad blocker and becoming a Supporting Member. For more info: Support TMC

Firmware 9 in August will start rolling out full self-driving features!!!

Discussion in 'Model S' started by diplomat33, Jun 10, 2018.

Tags:
  1. BigD0g

    BigD0g Active Member

    Joined:
    Jan 12, 2017
    Messages:
    1,653
    Location:
    Somewhere
    I apologize, I was truly speaking what I thought to be true, since my AP2 car does not do local road lane changes.
     
  2. conman

    conman Member

    Joined:
    Aug 19, 2017
    Messages:
    846
    Location:
    Melbourne, Australia
    That's fine, but it's odd that it doesn't for you? Every since V9, mine has done it everywhere that it recognises two or more lanes and I use it on a daily basis (since I use AP in urban roads all the time.) Note, I'm not talking about NoA, but just assisted lane change.
     
  3. dgatwood

    dgatwood Member

    Joined:
    Dec 20, 2017
    Messages:
    144
    Location:
    Sunnyvale, ca
    It's important to determine the size of the object. The nature of the object is (mostly) moot.
     
  4. MarcusMaximus

    MarcusMaximus Active Member

    Joined:
    Jan 2, 2017
    Messages:
    1,686
    Location:
    San Jose
    #1644 MarcusMaximus, Jan 8, 2019
    Last edited: Jan 8, 2019
    This, notably, doesn't seem to be the case anymore. Don't have immediate access to a link, but a test done in Europe a few months ago found that Tesla's system will stop for a fully stopped vehicle at the highest tested speed(~80mph). Every other car they tested failed to stop at much lower speeds.

    EDIT: found it again, Euro NCAP: 2018 Automated Driving Tests | Euro NCAP
    If you download the Model S PDF, you'll note it safely stops for a stationary car at 130kmph(roughly 80mph)
     
  5. OPRCE

    OPRCE Member

    Joined:
    Jun 16, 2018
    Messages:
    236
    Location:
    Bern, Switzerland

    Thanks, had been looking for this!

    N.B. from video that in worst case scenario, the cut-out to stationary vehicle test, it worked at 70kph but failed to brake at all from 90kph, requiring swift evasive action.

    From IC images that was still on v8.1 software, though I suspect v9 is as yet no better in this test.
     
  6. chillaban

    chillaban Active Member

    Joined:
    May 5, 2016
    Messages:
    3,387
    Location:
    Bay Area
    Yeah absolutely. Those improvements have been HIGHLY encouraging and AP2 performs very well these days when approaching stopped vehicle like objects. Way better than ap1. But the problem is this seems to be specifically trained against vehicles at various angles.

    There’s still no attempt to recognize a piece of furniture, mattress, bucket, etc etc etc.

    To me it seems like they need a general “can I safely drive over this thing” classifier. That’s how humans generally work. You are trained on, or learn the hard way, what things are safe to drive over and what things you should avoid hitting at all costs. It’s not just a simple matter of size either — a large plastic trash can is very different from a small piece of sharp metal debris.
     
    • Like x 2
  7. OPRCE

    OPRCE Member

    Joined:
    Jun 16, 2018
    Messages:
    236
    Location:
    Bern, Switzerland
    #1647 OPRCE, Jan 8, 2019
    Last edited: Jan 8, 2019
    Excellent point!
    I noticed my car in AP (2018.36.2) about 2 months ago in daylight would stop in a village at a temporary traffic-light roadworks barricade behind a stationary vehicle from 50kmh, but if no car was there already in same conditions 2 minutes later it completely ignored the massive wooden plank barricade perpendicular across the full lane width and would have ploughed straight on through it.

    It is also remarkable, even in AP v9 2018.48.2, that when approaching trucks in same lane at high differential speed [e.g. 60kmh] from the rear, they are recognised considerably later and usually with a few hiccups in speed compared to a standard car shape in same test.

    This must reflect that Tesla's training data thus far has been heavily biased towards recognition of the basic car silhouette.
     
  8. rnortman

    rnortman Member

    Joined:
    Aug 31, 2017
    Messages:
    644
    Location:
    Earth
    Tesla warns ‘self-driving package’ buyers that activation is ‘very far away’ due to regulations

    Yeah, "regulations". That's the problem. That's always been the problem, doncha know? Those pesky regulators forcing Tesla to keep their amazing technological capabilities locked away where nobody can see or test or verify them...

    “So even when we have the hardware ready, and your car would have it, you would most likely not be able to use it for a very long time.”​

    Yup, sounds like we are completely on-track for the first FSD features in August... 2020.

    Here's my prediction: HW3 will be required to deliver on the original EAP features. This is why they've toned down the EAP description. Tesla will not provide free upgrades to HW3 for EAP buyers, though, until somebody sues them, because they simply can't afford to give free hardware that probably costs $1.5-$2k to 200,000 customers (which is what it will be by then). So we'll all get $20 checks as a consolation prize, which we can choose to apply toward the $5000 HW3 upgrade, which will get us not FSD but "EAP++", which is really just something almost but not quite fulfilling the original EAP promises of on-ramp-to-off-ramp, smart summon, and self-park. With steering wheel nags in full force.

    A few FSD features, like stop sign and traffic light detection, will be released on HW3, but they will be driver assistance features. The human driver will be responsible. Most actions will require confirmation from the driver. The steering wheel nags will still be present at least on local roads. Highway may creep slowly toward an L3 system with severe limitations and the requirement that the driver remain aware, if not hands-on-wheel. But they have no way to ensure that the driver is aware, so this will be merely a legal CYA measure. People will read books on the highway, and people will die.

    And then the regulators will put a stop to it, and there will be no more "FSD" on currently-produced vehicles. And then Elon's pure raw hatred of regulators will be shown to be completely justified!
     
    • Like x 5
    • Love x 1
  9. OPRCE

    OPRCE Member

    Joined:
    Jun 16, 2018
    Messages:
    236
    Location:
    Bern, Switzerland
    A not too unlikely scenario, though L3 is perfectly legal in Europe right now, as evidenced by the 2019 Audi A8 already going on sale, so Tesla's concocted 'regulations' excuse holds no water at all.

    Furthermore, if they have FSD working, all they have to do is demonstrate it to the regulators and approval will be quickly granted if it is up to spec, even if at restricted speeds, like the Audi [60kmh AFAIK].

    In fact they have not yet finished writing the software, but side-windingly refuse to admit that to potential customers.

    It's like saying "you could read my amazing book if the bloody printers would only get their finger out", while still stricken with writer's block early on in chapter two of ten!
     
    • Like x 2
  10. selfbp

    selfbp Member

    Joined:
    Aug 4, 2017
    Messages:
    211
    Location:
    East, TN
    This thread aged like butter in the sun :D
     
  11. emmz0r

    emmz0r Member

    Joined:
    Jul 12, 2018
    Messages:
    263
    Location:
    Earth
    It's a clear double communication here:

    • Elon says FSD is progressing nicely
    • Elon earlier said FSD was off menu since it was "confusing"
    • Salespeople say that it's due to regulatory burden
     
    • Helpful x 1
  12. OPRCE

    OPRCE Member

    Joined:
    Jun 16, 2018
    Messages:
    236
    Location:
    Bern, Switzerland
    Getting it right has certainly been confusing for Mr Musk!

    The salespeople, however, say nothing meaningful, they merely imply some imaginary regulations *could* delay the customers' enjoyment of the currently non-existent FSD features, were they ever to be publicly released.

    In short, this leaked pitch is a transparent machination to use waffle to discourage people ordering the hidden option for FSD now, done only in order to be able to charge them more for it and/or the necessary hardware upgrades [including sensors] at a later date.

    If Tesla do not want to pre-sell the promise-of-future-FSD option they should simply remove it altogether instead of drizzling customers with contradictory BS!
     
  13. emmz0r

    emmz0r Member

    Joined:
    Jul 12, 2018
    Messages:
    263
    Location:
    Earth
    Yes I am getting "blueballed" all the time :p

    One thing though. Like NoA, I fear that FSD will wait for regulatory approval in US, even if it's fine and dandy in many EU countries. So even if FSD has some good features ready, I as a second class European won't see them, even if those features are green-lighted already.
     
  14. electronblue

    electronblue Member

    Joined:
    Oct 1, 2018
    Messages:
    362
    Location:
    Earth
    It would seem to me Tesla is simply walking the tightrope set up by Musk’s lofty promises back in 2016 and the retrofit requirements that have been confirmed later — as well as PR needs.

    Seems they can not really just stop selling FSD without a PR backlash (got to keep that autonomous storyline running?) but it has also become evident apparently that the facts are such that they’d prefer not to increase their actual legal and fiscal commitments on the matter either.

    So this happens.
     
    • Like x 5
  15. OPRCE

    OPRCE Member

    Joined:
    Jun 16, 2018
    Messages:
    236
    Location:
    Bern, Switzerland
    Naturally it would make no business sense to delay release in e.g. Europe and China while waiting for the US to homologise regulations federally, but yeah, I wouldn't put it past them either.

    Re. NoAP I suspect there may be some justification for the delay in Europe due to the maps (or fleet-gleaned overlay data?) being less accurate than in USA, and the feature itself still being in very Alpha shape anyhow.
     
  16. Bladerskb

    Bladerskb Like how many times do i have to be right?

    Joined:
    Oct 24, 2016
    Messages:
    1,258
    Location:
    Michigan
    Xray checked out, nothing broken other than some sores and minor strains.
    Other car occupants weren't hurt either. 40MPH head on, definitely glad i walked away with no injuries.
     
    • Like x 6
  17. OPRCE

    OPRCE Member

    Joined:
    Jun 16, 2018
    Messages:
    236
    Location:
    Bern, Switzerland
    Good news, Bro.

    BTW, would you now think of replacing that written-off Beemer with something having AP or similar alternative?

    I've noticed v9 in recent times has notably improved at detecting LTAP (lateral turn across path) traffic, including bicycles approaching at an oblique angle at night ...
     
  18. dgatwood

    dgatwood Member

    Joined:
    Dec 20, 2017
    Messages:
    144
    Location:
    Sunnyvale, ca
    Yeah, but not in the way you seem to think, IMO.

    A small piece of debris can typically be avoided without leaving the lane, which makes that a very different case than something that requires serious avoidance.

    But given a choice between hitting a large plastic trash can and a small piece of sharp metal, I'll take the sharp metal. A plastic trash can poses a risk of going up over the hood and, if it hits in just the right way, shattering your windshield, which can result in complete loss of control. A small piece of sharp metal can only puncture your tires. With modern run-flat tires, this is little more than a nuisance, requiring a tow and slowing down your trip considerably, but having very little real risk to the driver or passengers.

    Size matters far more than the nature of the object. Weight ostensibly matters more, but realistically we aren't going to see cars swerving to avoid small falling anvils any time soon, so that's probably not worth worrying about. :)
     
  19. chillaban

    chillaban Active Member

    Joined:
    May 5, 2016
    Messages:
    3,387
    Location:
    Bay Area
    Those were just two arbitrary examples I pulled out of the air as example of things that might be okay to hit versus you need to take slight evasive action, vs you really really need to avoid it. Right now Tesla’s don’t have run flats, and even with run flat tires, a sudden catastrophic deflation at 70mph will still likely destabilize the car, even though RFTs can adequately support a car driving 50mph for dozens of miles.

    But with that said it’s not like humans are very good at this either. A recent example: I’ve always thought it was probably okay to hit one of those thick round construction barrels. They look kind of empty and I often see construction workers moving them around. But I have a friend who hit one erroneously left in the middle of his lane, and I could not believe how much damage it caused to the car. Apparently they’re fairly reinforced and also filled with sand!
     
  20. diplomat33

    diplomat33 Member

    Joined:
    Aug 3, 2017
    Messages:
    718
    Location:
    Terre Haute, IN USA
    New article on Elektrek where Musk gives more details on the regulatory pushback. Apparently, it's the new "smart summon" that needs regulatory approval still:
    Tesla is getting ‘some regulatory pushbacks’ over new Autopilot feature, says Elon Musk

    I know some of you think the regulatory push back is just an excuse to cover up that FSD is nowhere near ready yet. But I think the article does make some sense. I can totally see how Musk's "smart summon" could get a lot of regulatory push back since it would involve allowing owners to remote control their car with no one in the driver seat. I can totally see how regulators might be hesitant to approve something like that given the obvious safety concerns. So I don't think it is far fetched at all that Tesla has developed the software to do it but the regulators are wanting more before they approve it. The same might be true for the rest of Tesla's FSD program. Maybe Tesla has developed the software to do a lot of what "FSD" would do but regulators want more tests, proof etc that the system is safe before they authorize a public release. So I don't think it is completely far fetched to believe that Tesla may be more ahead than we think in terms of pure software capabilities but regulatory approval is holding it up.
     
    • Like x 1
    • Disagree x 1

Share This Page

  • About Us

    Formed in 2006, Tesla Motors Club (TMC) was the first independent online Tesla community. Today it remains the largest and most dynamic community of Tesla enthusiasts. Learn more.
  • Do you value your experience at TMC? Consider becoming a Supporting Member of Tesla Motors Club. As a thank you for your contribution, you'll get nearly no ads in the Community and Groups sections. Additional perks are available depending on the level of contribution. Please visit the Account Upgrades page for more details.


    SUPPORT TMC