3 ms·

They will have drivers and steering wheels.

kgin 5y ago
They’re already running without drivers. Cruise’s purpose-built vehicle (which starts service in 2023) doesn’t even have a steering wheel. https://techcrunch.com/2021/10/07/cruise-plans-to-have-tens-of-thousands-of-origin-avs-on-roads-in-coming-years/ https://techcrunch.com/2021/10/07/cruise-plans-to-have-tens-...
- holonomically 5y ago
  > In August, Waymo began its Trusted Tester program in the city, as well, allowing San Franciscans to hail one of its autonomous, electric Jaguar I-Pace vehicles, with a safety driver on board, for free rides. There is plenty of time between now and 2023 and by then the safety record of these cars will not be good enough for regulators to allow fully driverless function. It's not good even now. Teslas constantly make all sorts of mistakes that force their drivers to take control. I doubt either Waymo or Cruise are further ahead than Tesla when it comes to the ML involved in making these things safe for the general public. Public roads are more adversarial than current ML models are capable of dealing with and this is before we even get to different kinds of low visibility weather conditions like rain and fog.
  - kgin 5y ago
    Tesla and Waymo/Cruise are playing entirely different games. Tesla is trying to make a full level 5 using only vision. Waymo and Cruise are going to level 4 in HD mapped service areas using every possible sensor they can fit onto a car. I don’t know if you can say Waymo/Cruise are necessarily ahead in some absolute technological sense, but they’re definitely closer to achieving their own goals and likewise way closer to actually driving paying customers around in autonomous vehicles.
    - holonomically 5y ago
      Good point. More sensors than vision does change things a little bit but not by much. These companies will have to have a bullet proof safety record and like Brooks says I'm pretty sure they're not going to be ready any time soon for operating on public roads. I actually think autonomous transport is a net positive but these companies don't seem to be serious about addressing safety issues or doing any research on how to mitigate the adversarial nature of public roads.

kgin 5y ago

They’re already running without drivers. Cruise’s purpose-built vehicle (which starts service in 2023) doesn’t even have a steering wheel. https://techcrunch.com/2021/10/07/cruise-plans-to-have-tens-of-thousands-of-origin-avs-on-roads-in-coming-years/ https://techcrunch.com/2021/10/07/cruise-plans-to-have-tens-...

holonomically 5y ago
> In August, Waymo began its Trusted Tester program in the city, as well, allowing San Franciscans to hail one of its autonomous, electric Jaguar I-Pace vehicles, with a safety driver on board, for free rides. There is plenty of time between now and 2023 and by then the safety record of these cars will not be good enough for regulators to allow fully driverless function. It's not good even now. Teslas constantly make all sorts of mistakes that force their drivers to take control. I doubt either Waymo or Cruise are further ahead than Tesla when it comes to the ML involved in making these things safe for the general public. Public roads are more adversarial than current ML models are capable of dealing with and this is before we even get to different kinds of low visibility weather conditions like rain and fog.
- kgin 5y ago
  Tesla and Waymo/Cruise are playing entirely different games. Tesla is trying to make a full level 5 using only vision. Waymo and Cruise are going to level 4 in HD mapped service areas using every possible sensor they can fit onto a car. I don’t know if you can say Waymo/Cruise are necessarily ahead in some absolute technological sense, but they’re definitely closer to achieving their own goals and likewise way closer to actually driving paying customers around in autonomous vehicles.
  - holonomically 5y ago
    Good point. More sensors than vision does change things a little bit but not by much. These companies will have to have a bullet proof safety record and like Brooks says I'm pretty sure they're not going to be ready any time soon for operating on public roads. I actually think autonomous transport is a net positive but these companies don't seem to be serious about addressing safety issues or doing any research on how to mitigate the adversarial nature of public roads.