3 ms·

Hey I made that comment! I assume Waymo has thought about it. I assume lots of people at Waymo have spent plenty of time in NYC. I also don't say that this was

Hey I made that comment! I assume Waymo has thought about it. I assume lots of people at Waymo have spent plenty of time in NYC. I also don't say that this was impossible or I think it will fail. What I do believe is that there are tons of edge cases, and diminishing returns with trying to solve for all of them. And, I would guess that it will take quite a while (the decade I mentioned) before self driving technology is capable of making >95% of trips autonomously without extra issues that most human drivers would handle. Additionally, I think there are moral and comfort issues with AI that need to be addressed. This is speculation, but Waymo likely aims for the appearance of safety and tunes things to be overly cautious. That is at odds with some of the dynamics of driving in NYC, where you need to be assertive and maybe even risky, and I'm curious to see how that will play out. I can see how it sounds like I'm just shitting on this tech. My intention was more to describe that self-driving in NYC is maybe even a harder problem than it sounds, and possibly an overreaction to the hype about this technology. I think about it pretty often when I'm driving or biking and encounter situations that seem very difficult for a computer to navigate.

ra7 5y ago
You make good points. I don't disagree that there are tons of edge cases and it could take a while. I mostly disagreed with the examples of edge cases that you think makes it hard. Quite a few of them can be solved with an up-to-date HD map (missing lane marking, construction zones, 1-way streets) and there is evidence of Waymo handling pedestrians and other complex scenarios well. NYC is definitely a harder problem. But Waymo's progress so far has been impressive and they have really solid technology from what I have followed all these years, so I'm pretty confident they can reasonably crack NYC. Your comment just sounded a bit too pessimistic for me is all :)
- StanislavPetrov 5y ago
  >Quite a few of them can be solved with an up-to-date HD map (missing lane marking, construction zones, 1-way streets) How do you keep these maps up to date in NYC when lane markings, construction zones, traffic flow, construction and a variety of other things are constantly changing on a daily basis? How do you deal with the ubiquitous "traffic cops" who stand in the center of many intersections and arbitrarily wave traffic in different directions?
  - ra7 5y ago
    Waymo claims their cars can detect environment changes and share it with the rest of the fleet in real time, with most of the map update process automated. They've written a blog post on this topic: https://blog.waymo.com/2020/09/the-waymo-driver-handbook-mapping.html https://blog.waymo.com/2020/09/the-waymo-driver-handbook-map... As for traffic cops, here's a video of Waymo obeying hand signals: https://www.youtube.com/watch?v=OopTOjnD3qY https://www.youtube.com/watch?v=OopTOjnD3qY. No reason this couldn't be replicated in NYC as well with some improvements.
    - StanislavPetrov 5y ago
      >As for traffic cops, here's a video of Waymo obeying hand signals: https://www.youtube.com/watch?v=OopTOjnD3qY https://www.youtube.com/watch?v=OopTOjnD3qY. No reason this couldn't be replicated in NYC as well with some improvements. This is an incredibly idealized version of "traffic cop" that doesn't exist in NYC. Here's an extremely well disciplined example of a traffic cop in NYC. They don't stand in one place, they are constantly challenged by traffic from all direction, they sometimes stop certain lanes, and tell other ones to go, with all sorts of gestures. And this is only a 4 way intersection without much traffic, and without any construction or significant pedestrian traffic. Its one thing to program an AI car that can drive under idealized, predictable conditions and entirely another to drive in the chaos of the real world, especially NYC. https://www.youtube.com/watch?v=m2Z7jS3Yb5Q https://www.youtube.com/watch?v=m2Z7jS3Yb5Q
      - ra7 5y ago
        These self driving companies are actively trying to solve these difficult problems, so it's not impossible just because it's NYC. If they can go from zero to driving in SF, they can solve the SF to NYC complexity.
  - ximeng 5y ago
    https://m.youtube.com/watch?v=68n4RjGhvRw https://m.youtube.com/watch?v=68n4RjGhvRw Has Tesla handling some construction. Not quite there but in theory these cars should handle these situations much like people do - have an expectation of the road layout and adjust based on what they see in reality.

ra7 5y ago

You make good points. I don't disagree that there are tons of edge cases and it could take a while. I mostly disagreed with the examples of edge cases that you think makes it hard. Quite a few of them can be solved with an up-to-date HD map (missing lane marking, construction zones, 1-way streets) and there is evidence of Waymo handling pedestrians and other complex scenarios well. NYC is definitely a harder problem. But Waymo's progress so far has been impressive and they have really solid technology from what I have followed all these years, so I'm pretty confident they can reasonably crack NYC. Your comment just sounded a bit too pessimistic for me is all :)

StanislavPetrov 5y ago
>Quite a few of them can be solved with an up-to-date HD map (missing lane marking, construction zones, 1-way streets) How do you keep these maps up to date in NYC when lane markings, construction zones, traffic flow, construction and a variety of other things are constantly changing on a daily basis? How do you deal with the ubiquitous "traffic cops" who stand in the center of many intersections and arbitrarily wave traffic in different directions?
- ra7 5y ago
  Waymo claims their cars can detect environment changes and share it with the rest of the fleet in real time, with most of the map update process automated. They've written a blog post on this topic: https://blog.waymo.com/2020/09/the-waymo-driver-handbook-mapping.html https://blog.waymo.com/2020/09/the-waymo-driver-handbook-map... As for traffic cops, here's a video of Waymo obeying hand signals: https://www.youtube.com/watch?v=OopTOjnD3qY https://www.youtube.com/watch?v=OopTOjnD3qY. No reason this couldn't be replicated in NYC as well with some improvements.
  - StanislavPetrov 5y ago
    >As for traffic cops, here's a video of Waymo obeying hand signals: https://www.youtube.com/watch?v=OopTOjnD3qY https://www.youtube.com/watch?v=OopTOjnD3qY. No reason this couldn't be replicated in NYC as well with some improvements. This is an incredibly idealized version of "traffic cop" that doesn't exist in NYC. Here's an extremely well disciplined example of a traffic cop in NYC. They don't stand in one place, they are constantly challenged by traffic from all direction, they sometimes stop certain lanes, and tell other ones to go, with all sorts of gestures. And this is only a 4 way intersection without much traffic, and without any construction or significant pedestrian traffic. Its one thing to program an AI car that can drive under idealized, predictable conditions and entirely another to drive in the chaos of the real world, especially NYC. https://www.youtube.com/watch?v=m2Z7jS3Yb5Q https://www.youtube.com/watch?v=m2Z7jS3Yb5Q
    - ra7 5y ago
      These self driving companies are actively trying to solve these difficult problems, so it's not impossible just because it's NYC. If they can go from zero to driving in SF, they can solve the SF to NYC complexity.
- ximeng 5y ago
  https://m.youtube.com/watch?v=68n4RjGhvRw https://m.youtube.com/watch?v=68n4RjGhvRw Has Tesla handling some construction. Not quite there but in theory these cars should handle these situations much like people do - have an expectation of the road layout and adjust based on what they see in reality.