Tesla's Autopilot Reduced Crashes by 40 Percent, but Not Really

Discussion in 'HardForum Tech News' started by Megalith, Feb 17, 2019.

  1. Megalith

    Megalith 24-bit/48kHz Staff Member

    Messages:
    13,003
    Joined:
    Aug 20, 2006
    The National Highway Traffic Safety Administration released a report in 2017 claiming Tesla’s Autopilot technology reduced crashes by as much as 40 percent, but an investigation by research firm Quality Control Systems Corporation suggests otherwise. “The majority of the vehicles in the Tesla data set suffered from missing data or other problems that made it impossible to say whether the activation of Autosteer increased or decreased the crash rate.” Ars Technica alleges the NHTSA is more concerned with protecting Tesla from embarrassment than the public from “potentially unsafe” technologies.

    …we discovered that the actual mileage at the time the Autosteer software was installed appears to have been reported for fewer than half the vehicles NHTSA studied. For those vehicles that do have apparently exact measurements of exposure mileage both before and after the software's installation, the change in crash rates associated with Autosteer is the opposite of that claimed by NHTSA - if these data are to be believed. For the remainder of the dataset, NHTSA ignored exposure mileage that could not be classified as either before or after the installation of Autosteer.
     
  2. travisty

    travisty Gawd

    Messages:
    816
    Joined:
    Feb 3, 2016
    I will be the first to admit i do not often read statistical analysis. I tried reading this and what I go was they took a subset of data, transformed it, and spit it on a wall as saw what stuck...

    The NHTSA data is also questionable as they took a subset of data as well - roughly 1% of total autopilot miles driven

    Let's look at who actually writes these reports on QCS (website that's linked in this post). A single man named Randy Whitfield, who is an epidemiologist and statistician.

    There's also Tesla's data for each quarter - https://www.tesla.com/VehicleSafetyReport
     
    Revdarian and Wrecked Em like this.
  3. c3k

    c3k 2[H]4U

    Messages:
    2,058
    Joined:
    Sep 8, 2007
    Integrating all the tech into the car, including autopilot, is blamed for a 40% production shortfall.

    QED. 40% fewer vehicles means 40% fewer accident.
     
  4. PeaKr

    PeaKr Gawd

    Messages:
    675
    Joined:
    Sep 6, 2004
    The NHTSA is supposed to protect us. A few years back they said this tech would need a decade to mature then all of sudden we're good to go. I wonder if there are any gov orgs we can trust anymore. Oh and ars is a rag now, it happened sometime after conde nast bought it.
     
  5. BloodyIron

    BloodyIron 2[H]4U

    Messages:
    3,460
    Joined:
    Jul 11, 2005
    The statistical majority of vehicular collisions are due to human error. Auto pilot is for sure about reducing vehicular collisions. Ignoring this stat is akin to ignoring another big killer... Cancer.
     
  6. Krenum

    Krenum [H]ardForum Junkie

    Messages:
    15,151
    Joined:
    Apr 29, 2005
    Well no shit, cars don't drive themselves. Maybe they need to for some people. I passed a dumbass on the freeway the other day driving with his knees, head down fumbling with his phone. Too bad they don't install rocket launchers on cars from the factory.
     
  7. macksomerville

    macksomerville [H]ard|Gawd

    Messages:
    1,966
    Joined:
    May 18, 2000
    For all of Tesla's foibles, I laude the day fucking autopilot is standard on all cars... 99.9% of humans have no reason to be in control of a 4 ton death machine.
     
    Revdarian and FreeLow like this.
  8. Burticus

    Burticus 2[H]4U

    Messages:
    3,745
    Joined:
    Nov 7, 2005
    Well I see the problem right here. Dude is sitting on the wrong side. /s

    1550434603ka1mdpanc6_1_1.jpg
     
    Uvaman2, motomonkey and filip like this.
  9. Corporal Hicks

    Corporal Hicks n00b

    Messages:
    9
    Joined:
    Nov 27, 2018
    Then autopilot will be in control of a 4 ton death machine, I'm not certain there will be much of a difference except who or what is controlling the 4 ton death machine, death is always the outcome of a 4 ton machine made for death.
    Also, 4 tons is quite a large car. We are talking plus size pickup trucks like Ford F250, Ford F350, Ram 2500, Large Vans, lot's of death in those. 2 tons is the sweet spot for 99.9% of humans.
     
  10. katanaD

    katanaD 2[H]4U

    Messages:
    2,050
    Joined:
    Nov 15, 2016

    you can ALWAYS trust that the IRS wants more of your money

    that whole death and taxes thing...
     
  11. macksomerville

    macksomerville [H]ard|Gawd

    Messages:
    1,966
    Joined:
    May 18, 2000
    I meant, 5 ton death machine, of death ;)
     
  12. motomonkey

    motomonkey [H]ard|Gawd

    Messages:
    1,395
    Joined:
    Jan 17, 2009
    Spin it however you want, they cooked the data to reach the conclusion they wished for, not what it actually represented.

    Eventually, when autopilots are perfect, your supposition will be true, it's certainly not true at this time.
     
  13. Uvaman2

    Uvaman2 2[H]4U

    Messages:
    2,887
    Joined:
    Jan 4, 2016
    Let's start with making this technology understand how to deal with stationary objects, as opposed to just assuming they don't exist... How about that?