3 ms·

> couldn't see general self-driving vehicles coming within 20 years. I still say that's true. 20 years is a long time, arguably you're already wrong because se

by rmtech 7y ago

> couldn't see general self-driving vehicles coming within 20 years. I still say that's true. 20 years is a long time, arguably you're already wrong because self-driving vehicles already exist. But Self-driving with the ability to replace Uber drivers will be here by 2030 at the latest.

thebruce87m 7y ago
If you’re talking about level 5 then I disagree with this. None of the technologies I see in use today can work in the same weather conditions that humans currently drive in. Things like snow, fog, and even heavy rain. If you’re talking anything below level 5 then all we will be doing is making current human drivers worse. Some of the deaths so far have been people who knew the limitations of the technology and still got complacent. If it requires you to have constant supervision then people simply won’t do that and will die.
- rmtech 7y ago
  People obsess about weather conditions as a limiting factor for level 5 and it's bizarre. Bad weather conditions will favour robots over humans because the robot drivers have sensors that can penetrate fog and rain whereas humans have to rely on mk1 eyeballs that can't. The real limiting factors for level 5 are less legible to humans. We understand the physical world via years of experience and good priors. Teslas crash straight into concrete lane separators and Ubers hit pedestrians that they know are there. In the latter case the actual reasons for the collision would strike a human as absolutely nuts - things like the algorithm not having a proper sense of object persistence or being terrible at commonsense reasoning. Yes, the Uber system was quite primitive, but the way that it broke gives us some insight into how AI drivers face a totally different set of challenges than what we would think of as "difficult conditions". See: https://www.jefftk.com/p/uber-self-driving-crash https://www.jefftk.com/p/uber-self-driving-crash
  - thebruce87m 7y ago
    Can you link to some of these sensors? As far as I know, anyone who knows what they’re doing is using Lidar at the moment which has significant difficulties with weather.
    - rmtech 7y ago
      Radar, ultrasonic. The radar can penetrate fog, rain, etc. https://navtechradar.com/features/autonomous-vehicles/ https://navtechradar.com/features/autonomous-vehicles/ Ultrasound works well up close for detecting a potential collision.
    - rmtech 7y ago
      Tesla will not use lidar apparently. They'll use cameras, radar and ultrasonic. In theory you could have all four.
  - jefftk 7y ago
    My interpretation of the Uber situation is that they were rushing in an incredibly reckless way. I think the Uber system described in the NTSB report is probably far worse than what Uber's more conservative competitors (Waymo, GM Cruise, etc) had at the time. (Disclosure: I work for Google which is a sibling of Waymo)
- pasttense01 7y ago
  Obviously you don't have to operate the self driving vehicles in all conditions in a ride sharing service, because you have the option of operating human driving vehicles in the worst conditions. [I also expect you would use human driving vehicles to help out in peak demand periods--you don't want to spend big money on a self-driving car unless it is going to be heavily used.]

by rmtech 7y ago

thebruce87m 7y ago

If you’re talking about level 5 then I disagree with this. None of the technologies I see in use today can work in the same weather conditions that humans currently drive in. Things like snow, fog, and even heavy rain. If you’re talking anything below level 5 then all we will be doing is making current human drivers worse. Some of the deaths so far have been people who knew the limitations of the technology and still got complacent. If it requires you to have constant supervision then people simply won’t do that and will die.

rmtech 7y ago
People obsess about weather conditions as a limiting factor for level 5 and it's bizarre. Bad weather conditions will favour robots over humans because the robot drivers have sensors that can penetrate fog and rain whereas humans have to rely on mk1 eyeballs that can't. The real limiting factors for level 5 are less legible to humans. We understand the physical world via years of experience and good priors. Teslas crash straight into concrete lane separators and Ubers hit pedestrians that they know are there. In the latter case the actual reasons for the collision would strike a human as absolutely nuts - things like the algorithm not having a proper sense of object persistence or being terrible at commonsense reasoning. Yes, the Uber system was quite primitive, but the way that it broke gives us some insight into how AI drivers face a totally different set of challenges than what we would think of as "difficult conditions". See: https://www.jefftk.com/p/uber-self-driving-crash https://www.jefftk.com/p/uber-self-driving-crash
- thebruce87m 7y ago
  Can you link to some of these sensors? As far as I know, anyone who knows what they’re doing is using Lidar at the moment which has significant difficulties with weather.
  - rmtech 7y ago
    Radar, ultrasonic. The radar can penetrate fog, rain, etc. https://navtechradar.com/features/autonomous-vehicles/ https://navtechradar.com/features/autonomous-vehicles/ Ultrasound works well up close for detecting a potential collision.
  - rmtech 7y ago
    Tesla will not use lidar apparently. They'll use cameras, radar and ultrasonic. In theory you could have all four.
- jefftk 7y ago
  My interpretation of the Uber situation is that they were rushing in an incredibly reckless way. I think the Uber system described in the NTSB report is probably far worse than what Uber's more conservative competitors (Waymo, GM Cruise, etc) had at the time. (Disclosure: I work for Google which is a sibling of Waymo)
pasttense01 7y ago
Obviously you don't have to operate the self driving vehicles in all conditions in a ride sharing service, because you have the option of operating human driving vehicles in the worst conditions. [I also expect you would use human driving vehicles to help out in peak demand periods--you don't want to spend big money on a self-driving car unless it is going to be heavily used.]