5 ms·

There is a long, long distance between what Tesla has today and a car with no driver.

by untog 11y ago

toomuchtodo 11y ago
I bet you $1000 to a charity of your choice they (Tesla Motors, either itself or through a subsidiary) have a fully autonomous (level 4) vehicle in less than 3 years, and actively provides a ridesharing service with its vehicles. https://en.wikipedia.org/wiki/Autonomous_car#Definition https://en.wikipedia.org/wiki/Autonomous_car#Definition "Level 4: The vehicle performs all safety-critical functions for the entire trip, with the driver not expected to control the vehicle at any time. As this vehicle would control all functions from start to stop, including all parking functions, it could include unoccupied cars."
- rayiner 11y ago
  I'll take that bet!
  - toomuchtodo 11y ago
    Now I'm curious :) Do you see regulatory hurdles as more pressing than technical hurdles? Because technical hurdles for self driving cars appear to be (mostly) solved. I don't even believe laser scanners Google is using are required anymore: http://gizmodo.com/mit-figured-out-how-to-make-cheap-3d-scanners-1-000-tim-1745454853 http://gizmodo.com/mit-figured-out-how-to-make-cheap-3d-scan...
    - chipotle_coyote 11y ago
      Do you see regulatory hurdles as more pressing than technical hurdles? Not yet. Based on what I know of Google's efforts, I think your three-year horizon for Tesla is, well, optimistic. As many miles as Google has logged, they're not testing at highway speeds yet -- I don't think their driverless prototypes (as opposed to the modified Lexus SUVs) are even capable of highway speeds. Also, AFAIK, they're only testing cars in Mountain View and Austin, which means there's a lot of weather conditions -- and to some degree terrain -- that they haven't had to accommodate. The most recently-given figures they've supplied that I can find is that they expect to be able to commercialize their technology between 2017 and 2020 -- and they've been working on this project since 2009. Tesla'a engineering is good, to be sure, but I would be really surprised if they beat Google to market in this area. Now, once somebody really does start offering these for sale to the general public (or just to taxi/rideshare companies or etc.), the answer to your question shifts from "not yet" to "yep."
      - toomuchtodo 11y ago
        > As many miles as Google has logged, they're not testing at highway speeds yet -- I don't think their driverless prototypes (as opposed to the modified Lexus SUVs) are even capable of highway speeds. Tesla's autopilot functions at up to 90mph. > I think your three-year horizon for Tesla is, well, optimistic. http://www.wired.com/2015/10/obviously-drivers-are-already-abusing-teslas-autopilot/ http://www.wired.com/2015/10/obviously-drivers-are-already-a... "Four days later, Reese, Mastracci, and Roy piled into a Model S and took off. They covered 2,994 miles at an average speed of 51.8 mph, a figure that includes the time spent plugged into Supercharger stations along the way. They had autopilot mode engaged 96 percent of the time, Reese says, using it at speeds around 90 mph. It eased the burden on the team, a big deal when you’re in a car for 57 hours straight." 96 percent of the time, at speeds up to 90mph, across 2,994 miles
        gozo 11y ago
        Speed is not the problem. Highways are going to be the first sections of self-driving roads. Because they are simple predictable environments, which requires less hardware, where accidents have high costs, liability issues are small, trips are long and where computers offers an advantage over the human mind in terms of endurance. This is something we will have in 3-5 years. To tackle all the other problems to truly have a self-driving cars is going to take much longer. Not at least since there will be decreasing incentive to solve these much harder problems in a cost effective way.
        greggyb 11y ago
        I take that as a very strong piece of evidence that 96% of the time on a very long road trip is spent on roads with near-ideal conditions. It sounds like the trip was mostly on highways. US interstates are wide, straight, and nearly uniformly impeccably painted and signed. 4% of the time there's inclement weather, ambiguously signed construction, heavy traffic in snow that occludes the entirety of the road, cars stopped in their lane, detours into the lane normally dedicated to oncoming traffic, and all sorts of other low-frequency events that are much harder to solve than keeping a car moving forward on an interstate. I'd argue that 96% autonomous driving time isn't even enough to say we're 50% of the way through the challenges of creating an autonomous vehicle. Think twice before accusing me of not understanding percentages.
    - skewart 11y ago
      I was about to jump on that bet too. Regardless of the technical hurdles, I would bet the regulatory hurdles alone are too big for 3 years. 5 year? Maybe not. 10 years? Hail no. But 3 years is a very short amount of time to get everything cleared up, especially with an election year coming up. My guess is that the interest groups who would oppose autonomous cars is a diverse enough coalition across the political spectrum that changing laws will not be super easy.
    - rayiner 11y ago
      Both. In the 2000's, I worked at a startup developing spectrum-sharing radio technology.[1] We ran into a range of problems similar to what I believe will hold up autonomous car development: 1) When you scale from individual autonomous devices to vast networks of them you run into emergent phenomena that you simply could not predict ahead of time. 2) You have to spend enormous engineering effort protecting incumbents. E.g. there has been a ton of work on figuring out how to protect wireless microphones used by churches and stadiums. They are intermittent and have a "dumb receiver" (that never transmits), making them very difficult to handle using a "listen before talk" approach. And that's for fricking church PAs. Imagine the push-back when your technological edge case runs over a kid. 3) You can't deploy commercially until you change the regulatory regime. That can take years to decades if you're asking for seismic shifts in policy. I would predict another 5-10 years before the technology works correctly at scale, rather than under controlled conditions. Then maybe another 5-10 years after that for limited regulatory approval, and maybe 10-25 years for widespread regulatory approval. I don't expect to be able to catch a self-driving cab in New York or Chicago before 2040. [1] The end goal was technologies that could obsolete fixed spectrum allocation by letting networks of cognitive radios sort out for themselves what frequencies to use. There are various narrower applications, such as technologies that can take advantage of TV "whitespaces."
- untog 11y ago
  I'll take that bet. Even if they produced a working prototype of this tomorrow I doubt the government would be able to pass the necessarily legislation in three years.
  - toomuchtodo 11y ago
    Deal. I'll setup a LongBet and post the info in this comment.
    - untog 11y ago
      Sorry, I'm not really going to bet a thousand dollars on an internet comment. I have bills to pay. That's not say I don't believe what I said, just that I don't have $1000 to throw around to back it up.
      - toomuchtodo 11y ago
        > I'll take that bet. Your comment: https://news.ycombinator.com/item?id=10673357 https://news.ycombinator.com/item?id=10673357 Wha?
        untog 11y ago
        "I bet you $1000" is, 98% of the time, not an actual offer to bet a thousand dollars with a person. The numbers are even higher when said person is a stranger you are talking to on an internet message board. You are lucky to be able to consider such a thing seriously!
        toomuchtodo 11y ago
        Forgive my honesty and authenticity then.
        untog 11y ago
        I don't really think it's a case of honesty, just a frequently used phrase. When my friend tells me he'll kill me if I drink the last beer I do not take it literally and call the police.

There is a long, long distance between what Tesla has today and a car with no driver.

by untog 11y ago

There is a long, long distance between what Tesla has today and a car with no driver.

I bet you $1000 to a charity of your choice they (Tesla Motors, either itself or through a subsidiary) have a fully autonomous (level 4) vehicle in less than 3 years, and actively provides a ridesharing service with its vehicles. https://en.wikipedia.org/wiki/Autonomous_car#Definition https://en.wikipedia.org/wiki/Autonomous_car#Definition "Level 4: The vehicle performs all safety-critical functions for the entire trip, with the driver not expected to control the vehicle at any time. As this vehicle would control all functions from start to stop, including all parking functions, it could include unoccupied cars."

rayiner 11y ago
I'll take that bet!
- toomuchtodo 11y ago
  Now I'm curious :) Do you see regulatory hurdles as more pressing than technical hurdles? Because technical hurdles for self driving cars appear to be (mostly) solved. I don't even believe laser scanners Google is using are required anymore: http://gizmodo.com/mit-figured-out-how-to-make-cheap-3d-scanners-1-000-tim-1745454853 http://gizmodo.com/mit-figured-out-how-to-make-cheap-3d-scan...
  - chipotle_coyote 11y ago
    Do you see regulatory hurdles as more pressing than technical hurdles? Not yet. Based on what I know of Google's efforts, I think your three-year horizon for Tesla is, well, optimistic. As many miles as Google has logged, they're not testing at highway speeds yet -- I don't think their driverless prototypes (as opposed to the modified Lexus SUVs) are even capable of highway speeds. Also, AFAIK, they're only testing cars in Mountain View and Austin, which means there's a lot of weather conditions -- and to some degree terrain -- that they haven't had to accommodate. The most recently-given figures they've supplied that I can find is that they expect to be able to commercialize their technology between 2017 and 2020 -- and they've been working on this project since 2009. Tesla'a engineering is good, to be sure, but I would be really surprised if they beat Google to market in this area. Now, once somebody really does start offering these for sale to the general public (or just to taxi/rideshare companies or etc.), the answer to your question shifts from "not yet" to "yep."
    - toomuchtodo 11y ago
      > As many miles as Google has logged, they're not testing at highway speeds yet -- I don't think their driverless prototypes (as opposed to the modified Lexus SUVs) are even capable of highway speeds. Tesla's autopilot functions at up to 90mph. > I think your three-year horizon for Tesla is, well, optimistic. http://www.wired.com/2015/10/obviously-drivers-are-already-abusing-teslas-autopilot/ http://www.wired.com/2015/10/obviously-drivers-are-already-a... "Four days later, Reese, Mastracci, and Roy piled into a Model S and took off. They covered 2,994 miles at an average speed of 51.8 mph, a figure that includes the time spent plugged into Supercharger stations along the way. They had autopilot mode engaged 96 percent of the time, Reese says, using it at speeds around 90 mph. It eased the burden on the team, a big deal when you’re in a car for 57 hours straight." 96 percent of the time, at speeds up to 90mph, across 2,994 miles
      - gozo 11y ago
        Speed is not the problem. Highways are going to be the first sections of self-driving roads. Because they are simple predictable environments, which requires less hardware, where accidents have high costs, liability issues are small, trips are long and where computers offers an advantage over the human mind in terms of endurance. This is something we will have in 3-5 years. To tackle all the other problems to truly have a self-driving cars is going to take much longer. Not at least since there will be decreasing incentive to solve these much harder problems in a cost effective way.
      - greggyb 11y ago
        I take that as a very strong piece of evidence that 96% of the time on a very long road trip is spent on roads with near-ideal conditions. It sounds like the trip was mostly on highways. US interstates are wide, straight, and nearly uniformly impeccably painted and signed. 4% of the time there's inclement weather, ambiguously signed construction, heavy traffic in snow that occludes the entirety of the road, cars stopped in their lane, detours into the lane normally dedicated to oncoming traffic, and all sorts of other low-frequency events that are much harder to solve than keeping a car moving forward on an interstate. I'd argue that 96% autonomous driving time isn't even enough to say we're 50% of the way through the challenges of creating an autonomous vehicle. Think twice before accusing me of not understanding percentages.
  - skewart 11y ago
    I was about to jump on that bet too. Regardless of the technical hurdles, I would bet the regulatory hurdles alone are too big for 3 years. 5 year? Maybe not. 10 years? Hail no. But 3 years is a very short amount of time to get everything cleared up, especially with an election year coming up. My guess is that the interest groups who would oppose autonomous cars is a diverse enough coalition across the political spectrum that changing laws will not be super easy.
  - rayiner 11y ago
    Both. In the 2000's, I worked at a startup developing spectrum-sharing radio technology.[1] We ran into a range of problems similar to what I believe will hold up autonomous car development: 1) When you scale from individual autonomous devices to vast networks of them you run into emergent phenomena that you simply could not predict ahead of time. 2) You have to spend enormous engineering effort protecting incumbents. E.g. there has been a ton of work on figuring out how to protect wireless microphones used by churches and stadiums. They are intermittent and have a "dumb receiver" (that never transmits), making them very difficult to handle using a "listen before talk" approach. And that's for fricking church PAs. Imagine the push-back when your technological edge case runs over a kid. 3) You can't deploy commercially until you change the regulatory regime. That can take years to decades if you're asking for seismic shifts in policy. I would predict another 5-10 years before the technology works correctly at scale, rather than under controlled conditions. Then maybe another 5-10 years after that for limited regulatory approval, and maybe 10-25 years for widespread regulatory approval. I don't expect to be able to catch a self-driving cab in New York or Chicago before 2040. [1] The end goal was technologies that could obsolete fixed spectrum allocation by letting networks of cognitive radios sort out for themselves what frequencies to use. There are various narrower applications, such as technologies that can take advantage of TV "whitespaces."
untog 11y ago
I'll take that bet. Even if they produced a working prototype of this tomorrow I doubt the government would be able to pass the necessarily legislation in three years.
- toomuchtodo 11y ago
  Deal. I'll setup a LongBet and post the info in this comment.
  - untog 11y ago
    Sorry, I'm not really going to bet a thousand dollars on an internet comment. I have bills to pay. That's not say I don't believe what I said, just that I don't have $1000 to throw around to back it up.
    - toomuchtodo 11y ago
      > I'll take that bet. Your comment: https://news.ycombinator.com/item?id=10673357 https://news.ycombinator.com/item?id=10673357 Wha?
      - untog 11y ago
        "I bet you $1000" is, 98% of the time, not an actual offer to bet a thousand dollars with a person. The numbers are even higher when said person is a stranger you are talking to on an internet message board. You are lucky to be able to consider such a thing seriously!
        toomuchtodo 11y ago
        Forgive my honesty and authenticity then.
        untog 11y ago
        I don't really think it's a case of honesty, just a frequently used phrase. When my friend tells me he'll kill me if I drink the last beer I do not take it literally and call the police.