7 ms·

Elon Musk: human drivers are 'too dangerous'

45 points by riaface 12y ago

Methusalah 12y ago
Sensationalist bs. It'd take some pretty impressive programming to deal with issues like road construction, missing or faded road markings, unmarked dirt roads, and potentially outdated or incorrect maps.
- jiballer 12y ago
  I don't think those are programmed into the driving software as much as it has learned them through machine learning. Most of these systems have gone through the equivalent of millions of years of humans driving so I'd trust it over a human driver any day.
  - a8da6b0c91d 12y ago
    I don't think any of the systems really use machine learning. It's more classical stats and control theory. Maybe the object recognition stuff uses some trained classifiers, but the higher level systems doing the driving to my knowledge do not.
  - 0xdeadbeefbabe 12y ago
    Well these systems weren't ever human, so I'm not sure a million human years would help them as much as you hope.
- lukasm 12y ago
  They're gonna cover 98% of cases. The rest can be picked up be remote operators.
  - Mithaldu 12y ago
    Remote operators work for flying drones, since those don't have to evade children running onto the road with sub-second reaction times. For cars there is no way electron-based signal transmission technology can control them with an even marginally acceptable latency.
    - lukasm 12y ago
      Latency doesn't matter. What matter is the final outcome. Better skill, software, data etc can compensate that.
      - astazangasta 12y ago
        Drone pilots can have several seconds of latency. This means remote operators are useless as a backup to the machine if a split second reaction is required.
    - jsprogrammer 12y ago
      What do you mean by electron-based signal transmission technology? If you're talking remote control, those signals will be photon based. If you're talking internal control, electrons on wires move pretty fast and many cars already utilize them. Is there much latency difference between mechanically closing the throttle with your foot and signaling a servo to close the throttle? Presumably some situations can be analyzed by a processor and trigger the servo before the human controller has even been able to consciously perceive a scene.
      - Mithaldu 12y ago
        One question: How do you plan to remote control a moving car with photons? Edit: Huh, ok. I didn't know radio signals are also photons. Still, the speed of light is significant with sub-second decisions. So even for photon transmission remote control of a moving car in live traffic is not feasible.
  - plikan13 12y ago
    You still need an algorithm to decide when it's time to handle control to a remote driver. Who's going to handle the edge cases of that algorithm?
  - ams6110 12y ago
    So a remote operator is going to be given command of a vehicle which has encountered an unusual situation that the automatics have bailed out of? Ask the pilots of Air France 447 how well that works out (and they weren't even remote). Oh you can't... they're dead.
- sesutton 12y ago
  I don't think we'll go straight from total human control to total automation directly. I imagine that the first versions of automated cars will require some supervision and only work in certain conditions. This will likely require some sort of system to detect the driver is paying attention, like steering wheel sensors, to prevent people from just going to sleep.
  - lojack 12y ago
    The first versions of automated cars already exist. Mercedes, BMW, and Audi have versions of their cars that maintain a safe distance from the cars in front of you. Mercedes also has a system which will automatically steer the car to keep it in a lane for you. Prius and Lexus have cars that'll automatically parallel park themselves.
rplst8 12y ago
He's right. Driving a car is something like 1000 times riskier than commercial aviation. We still have about 30,000 vehicle fatalities per year.
- refurb 12y ago
  1000 times more risky as measure by what? People drive a car far more often than fly in a plane. I'm sure flying is still safer, but the comparison should take into account the differences.
  - rplst8 12y ago
    The comparisons I've seen are usually on a person-mile basis. Commercial aviation can move more people a further distance with fewer fatalities.
drzaiusapelord 12y ago
Musk is pretty self-serving here. He writes about how advanced AI will "destroy humanity" to keep interest in manned space and to discredit robot and science missions. Now he's cheerleading automotive AI because Tesla wants to sell it. I really am starting to see him as an empty suit. He just does PR for his companies' current projects. That's fine, but he's not a visionary. He's salesman. I can't wait for his celebrity to wind down. Its more than a bit over the top now. Tesla seems like a failure to launch an affordable electric or an electric with realistic range at a sane pricepoint. SpaceX is doing well, but LEO theatrics are boring considering the SLS is geared to take us way past that. I wish he'd buy into causes that don't personally enrich him once in a while. I'd love to see someone of his caliber stump for a radio telescope on the far side of the moon or on replacing cars with more efficient tram systems powered by electric wires instead of the unsolved and perhaps unsolvable problem of batteries that can compete with the convienance of gasoline.
- matthewmacleod 12y ago
  This just seems needlessly picky, and I can't help but feel that you're expecting a little much. Of course Elon Musk is self-serving, in the sense that he's talking about and promoting the benefits of projects that he's working on. They're more impressive than you seem to be implying; Tesla's approach to tackling the existing automotive industry and introducing affordable vehicles is a long-term project, and considering we're about 11 or 12 years in, they seem to be making good progress. It's not something that many other companies have had the vision and drive to execute. Likewise with SpaceX. There's a long-term plan to colonise Mars – that does seem pretty visionary, no? And in terms of other projects, look at something like Hyperloop. You seem to imply that "efficient tram systems powered by electric wires" (a technology used everywhere throughout the world) would be more 'visionary' or something? That seems like an odd claim to make. People treat Musk like a bit of a celebrity because he talks unashamedly of large-scale, long-term visions for things he wants his projects to achieve. That's amazingly attractive in a world where there is so much focus on the short term, and what can't be achieved. I'm not sure why you react so badly to that.
  - drzaiusapelord 12y ago
tokenadult 12y ago
Here are some facts and figures about driving versus flying for travel safety per passenger mile over a time period carefully measured by official statistics. "In absolute numbers, driving is more dangerous, with more than 5 million accidents compared to 20 accidents in flying. A more direct comparison per 100 million miles pits driving's 1.27 fatalities and 80 injuries against flying's lack of deaths and almost no injuries, which again shows air travel to be safer."[1] I thought everybody knew that airline travel is by far the safest form of travel.[2] The basic issue about self-driving cars is that setting the standard of performance at "better than most human drivers most of the time" is a moderately low technical bar to clear, but would still result in a HUGE reduction of deaths, injuries, and damage to vehicles and roadside property. But I'm still not sure how soon self-driving cars will be good enough to drive on snowy roads (Google's self-driving cars have never been tested in snowy conditions, as of the last time I checked) or even in rain.[3] The story posted today is more nuanced than a story posted yesterday about Elon Musk's predictions, and reports, "He said he believed that we're still 20 years out from the roads being full of autonomous cars." [1] http://traveltips.usatoday.com/air-travel-safer-car-travel-1581.html http://traveltips.usatoday.com/air-travel-safer-car-travel-1... [2] http://www.businessinsider.com/flying-is-still-the-safest-way-to-travel-2013-7 http://www.businessinsider.com/flying-is-still-the-safest-wa... [3] http://spectrum.ieee.org/cars-that-think/transportation/advanced-cars/japan-competition-shows-weather-still-a-challenge-for-autonomous-cars http://spectrum.ieee.org/cars-that-think/transportation/adva...
- jowiar 12y ago
  Regarding poor-weather performance, I'm assuming self-driving cars will outperform human drivers sooner than in good conditions, rather than later (as long as it is made a priority). The information a self-driving car can gather and process about wheel traction is far more detailed than can be gathered by human drivers.
- Alupis 12y ago
  We're largely talking about only the US market here -- in which only (and I say that word relatively) 32,000 people died in any sort of automobile related accident last year. To put this in perspective, the common Flu (a totally vaccinatable, preventable, and treatable illness) killed more than 52,000 people in the US last year alone. In reality, automobiles in the US are the safest they have ever been since 1949, and automobile related deaths have dropped more than 26% since 2005 alone. According to the CDC, automobile related deaths are not even in the top 10 leading causes of death in the US. Last year, heart disease was responsible for 596,577 US deaths last year alone. Stroke claimed 128,932 US lives. Diabetes took 73,831 american lives. Suicide robbed 39,518 people their life last year -- in excess of 7,000 more than died by automobile related accidents. In 2009, approximately a grand total of 2,436,652 died in the US of all causes (including natural causes). Of that number, automobile related deaths only accounts for 1%. Of the entire US population, we're talking about 0.0096%... According to the CDC, in 2012, 10,322 people (or 31%) of all automobile related deaths were alcohol-related deaths. Another 18% of automobile related deaths are attributed to drugs other than alcohol in the same year. That accounts for just about 50% of all vehicle related deaths, leaving the true automobile related deaths due to accident at a shockingly low number (relatively) of about 15,000 people per year. All of this is to say, relatively speaking, automobile related deaths are not a significant concern. It does not make sense to spend significant effort in this sector while ignoring or not increasing efforts in these other much more problematic areas. Sources: [1] http://www.cdc.gov/nchs/fastats/leading-causes-of-death.htm http://www.cdc.gov/nchs/fastats/leading-causes-of-death.htm [2] http://www.usatoday.com/story/news/nation/2013/01/09/guns-traffic-deaths-rates/1784595/ http://www.usatoday.com/story/news/nation/2013/01/09/guns-tr... [3] http://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_in_U.S._by_year http://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_in... [4] http://www.businessinsider.com/top-causes-of-death-united-states-2011-11 http://www.businessinsider.com/top-causes-of-death-united-st... [5] http://www.iihs.org/iihs/topics/t/general-statistics/fatalityfacts/state-by-state-overview http://www.iihs.org/iihs/topics/t/general-statistics/fatalit... [6] http://www.cdc.gov/Motorvehiclesafety/impaired_driving/impaired-drv_factsheet.html http://www.cdc.gov/Motorvehiclesafety/impaired_driving/impai...
astazangasta 12y ago
The thing I keep emphasizing in this conversation is how terribly good human drivers are, on average. In the US, there are three fatalities for every 10 billion passenger miles traveled. That's tremendous. A machine that is currently incapable of driving in rain and unable to see a child in the street has a long way to go to hit this mark. Musk is here to sell his car. He'll say all sorts of bullshit to make it look sexier.
- FranOntanaya 12y ago
  Miles per passenger depend on too many factors. I'd stick with 1 death per 6,200 driver's licenses (US, 2009)
  - jeffdavis 12y ago
    Deaths per mile is more useful because it compares the cost (deaths) with the benefit (miles traveled per passenger). Additionally, we can compare directly against alternatives, such as buses, air travel, and self-driving cars. Deaths per driver's license doesn't tell us the benefit we are getting, and doesn't give us a way to make comparisons with alternatives.
    - omellet 12y ago
      A better measure would be deaths per driver decision, if you could estimate such a thing.
      - jeffdavis 12y ago
        That may measure the average driving ability; but it doesn't give you any hint as to the benefits (more driver decisions don't seem like a benefit to me), and it also doesn't give a good way to compare with alternatives.
    - Retric 12y ago
      Self-driving cars seem to be ~1:1 with driver licenses. A licenced driver drives at most one car at a time.
throwaway43 12y ago
I remain extremely skeptical that 95 % will be good enough and its the edge cases that really matter. It doesn't matter that they're better most of the time because the failure modes are different. However since Elon is a really smart guy who has mastered multiple disciplines , I'll take his word. I'm waiting for my self driving car.
lasermike026 12y ago
What about public transportation? We can make improvements to public transportation today.
matthewmacleod 12y ago
I'm a little bit confused about why there's such a negative reaction to this on HN - it seems like a pretty reasonable statement to make, no? Sure, I don't think there's any firm idea of when 'robotic' drivers will overtake humans in terms of safety. But it does seem pretty clear that it will eventually happen. Self-driving cars have a few massive potential advantages relative to humans, such as faster reaction times, no distractions, the ability to directly communicate car-to-car, and so on. It certainly seems likely that those advantages will far outweigh the adaptability of humans – after all, aren't most accidents caused by distraction, driving too fast for the conditions and the like?
- cheatsheet 12y ago
  The only thing I am concerned with is how secure / exploitable they are from a computer security perspective. I assume there will be some kind of link between the vehicular peripherals to vehicle control and configuration. Then I worry about who is actually controlling the car. I'm open to AI driven cars with redundant intelligent networks, but people are smart too, and some people are not nice people. How do you know that the entire system is functioning as expected? I'm pretty sure that's NP hard.
  - matthewmacleod 12y ago
    Definitely a concern, but I'm not sure what can be done about that – except for "let's develop the technology so that can't happen", even though that seems unlikely.
  - undersuit 12y ago
    > people are smart too, and some people are not nice people. What did you do to these people? Because there are two reasons these people aren't going to be nice to you: you offended them, or you're a random target of violence that we already experience, but now it takes the form of a homicidal self driving car instead of a boy in a movie theater with automatic weapons. Are you advocating we stop pursuing self driving cars just because security is an open ended problem?
ecobiker 12y ago
The way I see it - it's hard to "transfer" all the learnings we get from our everyday driving experiences to other fellow drivers but a machine can do that. It's going to be a sum total of all of its learnings and its neighbours (all over the world). I like the idea of it.
mixmastamyk 12y ago
I'm glad they are working on this in general, however unfortunately the Model S is not able to slow or stop when encountering a car stopped in front of it. I was quite disappointed when I found out they don't have this important technology, in a car packed to the hilt with it. http://my.teslamotors.com/forum/forums/collision-avoidance http://my.teslamotors.com/forum/forums/collision-avoidance
gaius 12y ago
Heh, I have a friend who is all rah-rah for self-driving cars. He also likes to go camping. So I asked him, when you show up at the site and the farmer says, top field, past the sheep, turn left at the tractor, what's your self-driving car going to do? Hmm, he said, well it would need some sort of touch screen interface for that. Or I said, some sort of "steering wheel"...
- elliottkember 12y ago
  I don't think "self-driving car" and "driveable car" are mutually exclusive. Self-driving mode for roads, highways, parking, commuting... manual override for the countryside.
- dragonwriter 12y ago
  > Heh, I have a friend who is all rah-rah for self-driving cars. He also likes to go camping. So I asked him, when you show up at the site and the farmer says, top field, past the sheep, turn left at the tractor, what's your self-driving car going to do? I don't see any problem with that, even with no direct controls. That's probably a situation that requires more detailed navigational input from a driver than providing a destination for road driving and letting the system handle the routing, but doesn't necessarily require actually manual driving. And, even if it did, even fully automated cars that are designed for personal ownership and (especially) for off-road use are likely to have full-control overrides available for some time. Of course, camp sites and other destinations that people are expected to drive to will probably also have navigation information or guides available that self-driving cars can follow when self-driving cars are common, rendering the whole hypothetical largely moot.
  - gaius 12y ago
    I can imagine the look on the average farmer's face when some city folk in their self driving car ask him to lead them up to the field himself. He'd probably just leave them there and drive off in his tractor, chuckling to himself.
    - dragonwriter 12y ago
lordnacho 12y ago
Humans are not going to stop needing sleep, get distracted, or react slowly. All we can really do (and have done) is mitigate. Seat belts, driver training, etc. Whatever problems a machine has, we can improve on them gradually. We can spend loads of time looking at test cases, and improvements are ratcheting. I can't wait for self-driving cars. It will be a massive change to be able to go to sleep in your own comfy bed and wake up in time for a meeting in another town. I also imagine traffic will be smoother, as tolerances don't have to be so wide.
clarle 12y ago
FWIW, Elon Musk did say the following on Twitter: > To be clear, Tesla is strongly in favor of people being allowed to drive their cars and always will be. Hopefully, that is obvious. > However, when self-driving cars become safer than human-driven cars, the public may outlaw the latter. Hopefully not. https://twitter.com/elonmusk/status/577946893646364673 https://twitter.com/elonmusk/status/577946893646364673 So his goal isn't to shore up support for legislation against non-autonomous cars anytime soon.
- closetnerd 12y ago
  He made that clarification because he was met with unexpectedly large audience which was critical of his statements. He was quite obviously supporting future legislation against humans driving cars: > In fact, in the distant future, I think it's probably going to be... people may outlaw driving cars, because it's too dangerous. > You can't have a person driving a two-ton death machine.
mudil 12y ago
Actually, human pilots are too dangerous as well. Analysis of airplane disasters consistently shows that human errors are responsible for majority of crashes. I would feel much more comfortable flying in a completely automated airplane.
panglott 12y ago
No question, if robot cars will obey speed limits and not blow through red lights. The biggest safety improvements will come from convincing motorists/passengers that it's OK to travel just a bit slower.
ocfx 12y ago
This is the natural progression of driving, dunno why people are getting so offended.

45 points by riaface 12y ago

Sensationalist bs. It'd take some pretty impressive programming to deal with issues like road construction, missing or faded road markings, unmarked dirt roads, and potentially outdated or incorrect maps.

jiballer 12y ago
I don't think those are programmed into the driving software as much as it has learned them through machine learning. Most of these systems have gone through the equivalent of millions of years of humans driving so I'd trust it over a human driver any day.
- a8da6b0c91d 12y ago
  I don't think any of the systems really use machine learning. It's more classical stats and control theory. Maybe the object recognition stuff uses some trained classifiers, but the higher level systems doing the driving to my knowledge do not.
- 0xdeadbeefbabe 12y ago
  Well these systems weren't ever human, so I'm not sure a million human years would help them as much as you hope.
lukasm 12y ago
They're gonna cover 98% of cases. The rest can be picked up be remote operators.
- Mithaldu 12y ago
  Remote operators work for flying drones, since those don't have to evade children running onto the road with sub-second reaction times. For cars there is no way electron-based signal transmission technology can control them with an even marginally acceptable latency.
  - lukasm 12y ago
    Latency doesn't matter. What matter is the final outcome. Better skill, software, data etc can compensate that.
    - astazangasta 12y ago
      Drone pilots can have several seconds of latency. This means remote operators are useless as a backup to the machine if a split second reaction is required.
  - jsprogrammer 12y ago
    What do you mean by electron-based signal transmission technology? If you're talking remote control, those signals will be photon based. If you're talking internal control, electrons on wires move pretty fast and many cars already utilize them. Is there much latency difference between mechanically closing the throttle with your foot and signaling a servo to close the throttle? Presumably some situations can be analyzed by a processor and trigger the servo before the human controller has even been able to consciously perceive a scene.
    - Mithaldu 12y ago
      One question: How do you plan to remote control a moving car with photons? Edit: Huh, ok. I didn't know radio signals are also photons. Still, the speed of light is significant with sub-second decisions. So even for photon transmission remote control of a moving car in live traffic is not feasible.
- plikan13 12y ago
  You still need an algorithm to decide when it's time to handle control to a remote driver. Who's going to handle the edge cases of that algorithm?
- ams6110 12y ago
  So a remote operator is going to be given command of a vehicle which has encountered an unusual situation that the automatics have bailed out of? Ask the pilots of Air France 447 how well that works out (and they weren't even remote). Oh you can't... they're dead.
sesutton 12y ago
I don't think we'll go straight from total human control to total automation directly. I imagine that the first versions of automated cars will require some supervision and only work in certain conditions. This will likely require some sort of system to detect the driver is paying attention, like steering wheel sensors, to prevent people from just going to sleep.
- lojack 12y ago
  The first versions of automated cars already exist. Mercedes, BMW, and Audi have versions of their cars that maintain a safe distance from the cars in front of you. Mercedes also has a system which will automatically steer the car to keep it in a lane for you. Prius and Lexus have cars that'll automatically parallel park themselves.

rplst8 12y ago

He's right. Driving a car is something like 1000 times riskier than commercial aviation. We still have about 30,000 vehicle fatalities per year.

refurb 12y ago
1000 times more risky as measure by what? People drive a car far more often than fly in a plane. I'm sure flying is still safer, but the comparison should take into account the differences.
- rplst8 12y ago
  The comparisons I've seen are usually on a person-mile basis. Commercial aviation can move more people a further distance with fewer fatalities.

drzaiusapelord 12y ago

Musk is pretty self-serving here. He writes about how advanced AI will "destroy humanity" to keep interest in manned space and to discredit robot and science missions. Now he's cheerleading automotive AI because Tesla wants to sell it. I really am starting to see him as an empty suit. He just does PR for his companies' current projects. That's fine, but he's not a visionary. He's salesman. I can't wait for his celebrity to wind down. Its more than a bit over the top now. Tesla seems like a failure to launch an affordable electric or an electric with realistic range at a sane pricepoint. SpaceX is doing well, but LEO theatrics are boring considering the SLS is geared to take us way past that. I wish he'd buy into causes that don't personally enrich him once in a while. I'd love to see someone of his caliber stump for a radio telescope on the far side of the moon or on replacing cars with more efficient tram systems powered by electric wires instead of the unsolved and perhaps unsolvable problem of batteries that can compete with the convienance of gasoline.

matthewmacleod 12y ago
This just seems needlessly picky, and I can't help but feel that you're expecting a little much. Of course Elon Musk is self-serving, in the sense that he's talking about and promoting the benefits of projects that he's working on. They're more impressive than you seem to be implying; Tesla's approach to tackling the existing automotive industry and introducing affordable vehicles is a long-term project, and considering we're about 11 or 12 years in, they seem to be making good progress. It's not something that many other companies have had the vision and drive to execute. Likewise with SpaceX. There's a long-term plan to colonise Mars – that does seem pretty visionary, no? And in terms of other projects, look at something like Hyperloop. You seem to imply that "efficient tram systems powered by electric wires" (a technology used everywhere throughout the world) would be more 'visionary' or something? That seems like an odd claim to make. People treat Musk like a bit of a celebrity because he talks unashamedly of large-scale, long-term visions for things he wants his projects to achieve. That's amazingly attractive in a world where there is so much focus on the short term, and what can't be achieved. I'm not sure why you react so badly to that.
- drzaiusapelord 12y ago

tokenadult 12y ago

Here are some facts and figures about driving versus flying for travel safety per passenger mile over a time period carefully measured by official statistics. "In absolute numbers, driving is more dangerous, with more than 5 million accidents compared to 20 accidents in flying. A more direct comparison per 100 million miles pits driving's 1.27 fatalities and 80 injuries against flying's lack of deaths and almost no injuries, which again shows air travel to be safer."[1] I thought everybody knew that airline travel is by far the safest form of travel.[2] The basic issue about self-driving cars is that setting the standard of performance at "better than most human drivers most of the time" is a moderately low technical bar to clear, but would still result in a HUGE reduction of deaths, injuries, and damage to vehicles and roadside property. But I'm still not sure how soon self-driving cars will be good enough to drive on snowy roads (Google's self-driving cars have never been tested in snowy conditions, as of the last time I checked) or even in rain.[3] The story posted today is more nuanced than a story posted yesterday about Elon Musk's predictions, and reports, "He said he believed that we're still 20 years out from the roads being full of autonomous cars." [1] http://traveltips.usatoday.com/air-travel-safer-car-travel-1581.html http://traveltips.usatoday.com/air-travel-safer-car-travel-1... [2] http://www.businessinsider.com/flying-is-still-the-safest-way-to-travel-2013-7 http://www.businessinsider.com/flying-is-still-the-safest-wa... [3] http://spectrum.ieee.org/cars-that-think/transportation/advanced-cars/japan-competition-shows-weather-still-a-challenge-for-autonomous-cars http://spectrum.ieee.org/cars-that-think/transportation/adva...

jowiar 12y ago
Regarding poor-weather performance, I'm assuming self-driving cars will outperform human drivers sooner than in good conditions, rather than later (as long as it is made a priority). The information a self-driving car can gather and process about wheel traction is far more detailed than can be gathered by human drivers.
Alupis 12y ago
We're largely talking about only the US market here -- in which only (and I say that word relatively) 32,000 people died in any sort of automobile related accident last year. To put this in perspective, the common Flu (a totally vaccinatable, preventable, and treatable illness) killed more than 52,000 people in the US last year alone. In reality, automobiles in the US are the safest they have ever been since 1949, and automobile related deaths have dropped more than 26% since 2005 alone. According to the CDC, automobile related deaths are not even in the top 10 leading causes of death in the US. Last year, heart disease was responsible for 596,577 US deaths last year alone. Stroke claimed 128,932 US lives. Diabetes took 73,831 american lives. Suicide robbed 39,518 people their life last year -- in excess of 7,000 more than died by automobile related accidents. In 2009, approximately a grand total of 2,436,652 died in the US of all causes (including natural causes). Of that number, automobile related deaths only accounts for 1%. Of the entire US population, we're talking about 0.0096%... According to the CDC, in 2012, 10,322 people (or 31%) of all automobile related deaths were alcohol-related deaths. Another 18% of automobile related deaths are attributed to drugs other than alcohol in the same year. That accounts for just about 50% of all vehicle related deaths, leaving the true automobile related deaths due to accident at a shockingly low number (relatively) of about 15,000 people per year. All of this is to say, relatively speaking, automobile related deaths are not a significant concern. It does not make sense to spend significant effort in this sector while ignoring or not increasing efforts in these other much more problematic areas. Sources: [1] http://www.cdc.gov/nchs/fastats/leading-causes-of-death.htm http://www.cdc.gov/nchs/fastats/leading-causes-of-death.htm [2] http://www.usatoday.com/story/news/nation/2013/01/09/guns-traffic-deaths-rates/1784595/ http://www.usatoday.com/story/news/nation/2013/01/09/guns-tr... [3] http://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_in_U.S._by_year http://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_in... [4] http://www.businessinsider.com/top-causes-of-death-united-states-2011-11 http://www.businessinsider.com/top-causes-of-death-united-st... [5] http://www.iihs.org/iihs/topics/t/general-statistics/fatalityfacts/state-by-state-overview http://www.iihs.org/iihs/topics/t/general-statistics/fatalit... [6] http://www.cdc.gov/Motorvehiclesafety/impaired_driving/impaired-drv_factsheet.html http://www.cdc.gov/Motorvehiclesafety/impaired_driving/impai...

astazangasta 12y ago

The thing I keep emphasizing in this conversation is how terribly good human drivers are, on average. In the US, there are three fatalities for every 10 billion passenger miles traveled. That's tremendous. A machine that is currently incapable of driving in rain and unable to see a child in the street has a long way to go to hit this mark. Musk is here to sell his car. He'll say all sorts of bullshit to make it look sexier.

FranOntanaya 12y ago
Miles per passenger depend on too many factors. I'd stick with 1 death per 6,200 driver's licenses (US, 2009)
- jeffdavis 12y ago
  Deaths per mile is more useful because it compares the cost (deaths) with the benefit (miles traveled per passenger). Additionally, we can compare directly against alternatives, such as buses, air travel, and self-driving cars. Deaths per driver's license doesn't tell us the benefit we are getting, and doesn't give us a way to make comparisons with alternatives.
  - omellet 12y ago
    A better measure would be deaths per driver decision, if you could estimate such a thing.
    - jeffdavis 12y ago
      That may measure the average driving ability; but it doesn't give you any hint as to the benefits (more driver decisions don't seem like a benefit to me), and it also doesn't give a good way to compare with alternatives.
  - Retric 12y ago
    Self-driving cars seem to be ~1:1 with driver licenses. A licenced driver drives at most one car at a time.

throwaway43 12y ago

I remain extremely skeptical that 95 % will be good enough and its the edge cases that really matter. It doesn't matter that they're better most of the time because the failure modes are different. However since Elon is a really smart guy who has mastered multiple disciplines , I'll take his word. I'm waiting for my self driving car.

lasermike026 12y ago

What about public transportation? We can make improvements to public transportation today.

matthewmacleod 12y ago

I'm a little bit confused about why there's such a negative reaction to this on HN - it seems like a pretty reasonable statement to make, no? Sure, I don't think there's any firm idea of when 'robotic' drivers will overtake humans in terms of safety. But it does seem pretty clear that it will eventually happen. Self-driving cars have a few massive potential advantages relative to humans, such as faster reaction times, no distractions, the ability to directly communicate car-to-car, and so on. It certainly seems likely that those advantages will far outweigh the adaptability of humans – after all, aren't most accidents caused by distraction, driving too fast for the conditions and the like?

cheatsheet 12y ago
The only thing I am concerned with is how secure / exploitable they are from a computer security perspective. I assume there will be some kind of link between the vehicular peripherals to vehicle control and configuration. Then I worry about who is actually controlling the car. I'm open to AI driven cars with redundant intelligent networks, but people are smart too, and some people are not nice people. How do you know that the entire system is functioning as expected? I'm pretty sure that's NP hard.
- matthewmacleod 12y ago
  Definitely a concern, but I'm not sure what can be done about that – except for "let's develop the technology so that can't happen", even though that seems unlikely.
- undersuit 12y ago
  > people are smart too, and some people are not nice people. What did you do to these people? Because there are two reasons these people aren't going to be nice to you: you offended them, or you're a random target of violence that we already experience, but now it takes the form of a homicidal self driving car instead of a boy in a movie theater with automatic weapons. Are you advocating we stop pursuing self driving cars just because security is an open ended problem?

ecobiker 12y ago

The way I see it - it's hard to "transfer" all the learnings we get from our everyday driving experiences to other fellow drivers but a machine can do that. It's going to be a sum total of all of its learnings and its neighbours (all over the world). I like the idea of it.

mixmastamyk 12y ago

I'm glad they are working on this in general, however unfortunately the Model S is not able to slow or stop when encountering a car stopped in front of it. I was quite disappointed when I found out they don't have this important technology, in a car packed to the hilt with it. http://my.teslamotors.com/forum/forums/collision-avoidance http://my.teslamotors.com/forum/forums/collision-avoidance

gaius 12y ago

Heh, I have a friend who is all rah-rah for self-driving cars. He also likes to go camping. So I asked him, when you show up at the site and the farmer says, top field, past the sheep, turn left at the tractor, what's your self-driving car going to do? Hmm, he said, well it would need some sort of touch screen interface for that. Or I said, some sort of "steering wheel"...

elliottkember 12y ago
I don't think "self-driving car" and "driveable car" are mutually exclusive. Self-driving mode for roads, highways, parking, commuting... manual override for the countryside.
dragonwriter 12y ago
> Heh, I have a friend who is all rah-rah for self-driving cars. He also likes to go camping. So I asked him, when you show up at the site and the farmer says, top field, past the sheep, turn left at the tractor, what's your self-driving car going to do? I don't see any problem with that, even with no direct controls. That's probably a situation that requires more detailed navigational input from a driver than providing a destination for road driving and letting the system handle the routing, but doesn't necessarily require actually manual driving. And, even if it did, even fully automated cars that are designed for personal ownership and (especially) for off-road use are likely to have full-control overrides available for some time. Of course, camp sites and other destinations that people are expected to drive to will probably also have navigation information or guides available that self-driving cars can follow when self-driving cars are common, rendering the whole hypothetical largely moot.
- gaius 12y ago
  I can imagine the look on the average farmer's face when some city folk in their self driving car ask him to lead them up to the field himself. He'd probably just leave them there and drive off in his tractor, chuckling to himself.
  - dragonwriter 12y ago

lordnacho 12y ago

Humans are not going to stop needing sleep, get distracted, or react slowly. All we can really do (and have done) is mitigate. Seat belts, driver training, etc. Whatever problems a machine has, we can improve on them gradually. We can spend loads of time looking at test cases, and improvements are ratcheting. I can't wait for self-driving cars. It will be a massive change to be able to go to sleep in your own comfy bed and wake up in time for a meeting in another town. I also imagine traffic will be smoother, as tolerances don't have to be so wide.

clarle 12y ago

FWIW, Elon Musk did say the following on Twitter: > To be clear, Tesla is strongly in favor of people being allowed to drive their cars and always will be. Hopefully, that is obvious. > However, when self-driving cars become safer than human-driven cars, the public may outlaw the latter. Hopefully not. https://twitter.com/elonmusk/status/577946893646364673 https://twitter.com/elonmusk/status/577946893646364673 So his goal isn't to shore up support for legislation against non-autonomous cars anytime soon.

closetnerd 12y ago
He made that clarification because he was met with unexpectedly large audience which was critical of his statements. He was quite obviously supporting future legislation against humans driving cars: > In fact, in the distant future, I think it's probably going to be... people may outlaw driving cars, because it's too dangerous. > You can't have a person driving a two-ton death machine.

mudil 12y ago

Actually, human pilots are too dangerous as well. Analysis of airplane disasters consistently shows that human errors are responsible for majority of crashes. I would feel much more comfortable flying in a completely automated airplane.

panglott 12y ago

No question, if robot cars will obey speed limits and not blow through red lights. The biggest safety improvements will come from convincing motorists/passengers that it's OK to travel just a bit slower.

ocfx 12y ago

This is the natural progression of driving, dunno why people are getting so offended.