3 ms·

If you go look at IIHS, they have tests that will tell you exactly how good non-Autopilot cars are at this kind of crash avoidance: I cannot find any current (

If you go look at IIHS, they have tests that will tell you exactly how good non-Autopilot cars are at this kind of crash avoidance: I cannot find any current (2018) car that crashes into the parked car. (I only spot-checked prior years). I found one in 2014 (audi q5) that hits the car at 1mph when going 12mph. http://www.iihs.org/iihs/ratings/vehicle/v/audi/q5-4-door-suv/2014 http://www.iihs.org/iihs/ratings/vehicle/v/audi/q5-4-door-su... By contrast, the 2014 subaru outback avoids collision even at 25mph. http://www.iihs.org/iihs/ratings/vehicle/v/subaru/outback-4-door-wagon/2014 http://www.iihs.org/iihs/ratings/vehicle/v/subaru/outback-4-...

mikeash 8y ago
Their report for the Model S says it avoided the collisions. Clearly there’s more to it than that.
- DannyBee 8y ago
  You asked about non-autopilot cars and how common it is. I gave you data. I can also find no data that suggests any other TACC is crashing into vehicles (no NTSB studies, investigations, etc) at all. So yeah, i'm going to conclude, based on what i've seen so far, that autopilot is part of the problem. If you have data that says autopilot is not worse, i'd love to see it.
  - mikeash 8y ago
    I asked about actual crashes. You gave me info about a controlled test where the exact car discussed in this article avoided the test collision. Given that the car discussed in this article crashed in real life, the controlled test is clearly not the final word in whether these other cars are experiencing crashes like this in the real world. Also, I wasn't asking about how Autopilot compares to similar systems from other manufacturers, I was asking how it compares to the Mark I Human, which is a driving system that crashes into stationary vehicles with distressing frequency.
    - DannyBee 8y ago
      I'm just going to leave you to your non-data driven skepticism of "everything other than tesla", because this is clearly not a useful conversation. That isn't what you originally asked (and as you see, i wasn't the only one who interpreted it this way), and imho it's not a useful comparison. I have no idea why one would ask "is autopilot worse" and then say "but only compared to humans, not taking into account the other systems that do the same thing and humans are also using on a regular basis" It doesn't answer the question whether autopilot is really worse or not, because nobody cares if it's better or worse than a straw man human who has disabled the collision avoidance on their car.
      - mikeash 8y ago
        I'm not talking about people who disable collision avoidance. I'm talking about people driving cars that don't have it at all. These are, as far as I know, still the vast majority of cars out there. The data you provided is clearly not sufficient to reach conclusions about actual crash rates, so I really don't see how it's useful to present.
        DannyBee 8y ago
        and again, you literally have no data at all, sufficient or not, to back up your claims. Instead, you just shit on all data that doesn't agree with your pet theory as insufficient. I'll take insufficient data over zero data any day of the week.
mentat 8y ago
Subaru Outback is vastly superior collision avoidance than AP which probably says a lot about software quality. It certainly doesn't have better sensors.
- mikeash 8y ago
  They're equally capable according to that test, and it's not optional equipment on the Tesla.
  - mentat 8y ago
    My experience says they're not with a few thousand miles of driving but sure, whatever.
    - tgb 8y ago
      How often are you attempting to drive into parked cars?
      - mentat 8y ago
        Cars in front of you decide to do random things like slam on the brakes or make sudden turns. Eye sight consistently would warn early then brake as much as necessary. I only had it do emergency braking 2 times and both would have been accidents as I'd misjudged the distance and speed of other objects. They have you test driving towards a trash can as part of the sale, it works, so not just cars in a lane. More interestingly though, the warnings, especially as early as they happen, made me a better driver fairly quickly. Tangentially, the warning sound on the Outback is far more compelling than on the Tesla. Weird UX thing.
Maxion 8y ago
But that's not actual data on crashes. What op is suggesting is that crashes per mile driven is a more relevant statistic. Obviously self driving cars will be involved in accidents, the question is, is it happening more frequently than with human drivers? If yes, then there's a problem and an indication that the technology is not yet mature.

mikeash 8y ago

Their report for the Model S says it avoided the collisions. Clearly there’s more to it than that.

DannyBee 8y ago
You asked about non-autopilot cars and how common it is. I gave you data. I can also find no data that suggests any other TACC is crashing into vehicles (no NTSB studies, investigations, etc) at all. So yeah, i'm going to conclude, based on what i've seen so far, that autopilot is part of the problem. If you have data that says autopilot is not worse, i'd love to see it.
- mikeash 8y ago
  I asked about actual crashes. You gave me info about a controlled test where the exact car discussed in this article avoided the test collision. Given that the car discussed in this article crashed in real life, the controlled test is clearly not the final word in whether these other cars are experiencing crashes like this in the real world. Also, I wasn't asking about how Autopilot compares to similar systems from other manufacturers, I was asking how it compares to the Mark I Human, which is a driving system that crashes into stationary vehicles with distressing frequency.
  - DannyBee 8y ago
    I'm just going to leave you to your non-data driven skepticism of "everything other than tesla", because this is clearly not a useful conversation. That isn't what you originally asked (and as you see, i wasn't the only one who interpreted it this way), and imho it's not a useful comparison. I have no idea why one would ask "is autopilot worse" and then say "but only compared to humans, not taking into account the other systems that do the same thing and humans are also using on a regular basis" It doesn't answer the question whether autopilot is really worse or not, because nobody cares if it's better or worse than a straw man human who has disabled the collision avoidance on their car.
    - mikeash 8y ago
      I'm not talking about people who disable collision avoidance. I'm talking about people driving cars that don't have it at all. These are, as far as I know, still the vast majority of cars out there. The data you provided is clearly not sufficient to reach conclusions about actual crash rates, so I really don't see how it's useful to present.
      - DannyBee 8y ago
        and again, you literally have no data at all, sufficient or not, to back up your claims. Instead, you just shit on all data that doesn't agree with your pet theory as insufficient. I'll take insufficient data over zero data any day of the week.

mentat 8y ago

Subaru Outback is vastly superior collision avoidance than AP which probably says a lot about software quality. It certainly doesn't have better sensors.

mikeash 8y ago
They're equally capable according to that test, and it's not optional equipment on the Tesla.
- mentat 8y ago
  My experience says they're not with a few thousand miles of driving but sure, whatever.
  - tgb 8y ago
    How often are you attempting to drive into parked cars?
    - mentat 8y ago
      Cars in front of you decide to do random things like slam on the brakes or make sudden turns. Eye sight consistently would warn early then brake as much as necessary. I only had it do emergency braking 2 times and both would have been accidents as I'd misjudged the distance and speed of other objects. They have you test driving towards a trash can as part of the sale, it works, so not just cars in a lane. More interestingly though, the warnings, especially as early as they happen, made me a better driver fairly quickly. Tangentially, the warning sound on the Outback is far more compelling than on the Tesla. Weird UX thing.

Maxion 8y ago

But that's not actual data on crashes. What op is suggesting is that crashes per mile driven is a more relevant statistic. Obviously self driving cars will be involved in accidents, the question is, is it happening more frequently than with human drivers? If yes, then there's a problem and an indication that the technology is not yet mature.