3 ms·

Autonomous cars won’t be proprietary to each manufacturer or company. Once someone has perfected their system, everyone else will drop what they are doing and s

Autonomous cars won’t be proprietary to each manufacturer or company. Once someone has perfected their system, everyone else will drop what they are doing and simply buy it like an OS. It’s all in the software; hardware required (cameras, radar, GPUs) is already cheap and ubiquitous. My money is on Tesla or comma.ai for this.

nostrademons 8y ago
Most likely the business model for autonomous cars will be a rideshare - instead of selling the software, the company with the best software will buy cars and rent them out for rides via app. Waymo has already gone in this direction. Also, Tesla Autopilot is a joke compared to what Waymo or even Cruise has.
jjeaff 8y ago
Why Tesla? They seem to be near the bottom of the pack when it comes to self driving capability.
- watwat1 8y ago
  You clearly don’t own or have ever ridden in a Tesla.
- aphextron 8y ago
  >They seem to be near the bottom of the pack when it comes to self driving capability. And yet they are the only company in the world to have shipped a workable level 2 system in mass quantities. Orders of magnitude greater than Cruise or Waymo's fleet size. That puts them miles ahead in terms of data collection, which is the real challenge.
  - mrnobody_67 8y ago
    Supposedly adversarial AI networks are solving for the data problem... AI generates the data to train the AI.
notyourwork 8y ago
My understanding is Tesla and comma.ai have taken a software focused approach with limited hardware. Others like waymo have taken a hardware approach with more sophisticated sensors. What makes you feel Tesla/Comma.ai will win?
- toomuchtodo 8y ago
  Human drivers aren’t equipped with LIDAR.
- aphextron 8y ago
  Because sensors aren’t the hard part. We already have enough hardware to perform superhuman sensing but it’s meaningless without the massive amounts of mapping data and trained models to make sense of it. http://www.youtube.com/watch?v=OikZ6J1YDlI&t=18m20s http://www.youtube.com/watch?v=OikZ6J1YDlI&t=18m20s

nostrademons 8y ago

Most likely the business model for autonomous cars will be a rideshare - instead of selling the software, the company with the best software will buy cars and rent them out for rides via app. Waymo has already gone in this direction. Also, Tesla Autopilot is a joke compared to what Waymo or even Cruise has.

jjeaff 8y ago

Why Tesla? They seem to be near the bottom of the pack when it comes to self driving capability.

watwat1 8y ago
You clearly don’t own or have ever ridden in a Tesla.
aphextron 8y ago
>They seem to be near the bottom of the pack when it comes to self driving capability. And yet they are the only company in the world to have shipped a workable level 2 system in mass quantities. Orders of magnitude greater than Cruise or Waymo's fleet size. That puts them miles ahead in terms of data collection, which is the real challenge.
- mrnobody_67 8y ago
  Supposedly adversarial AI networks are solving for the data problem... AI generates the data to train the AI.

notyourwork 8y ago

My understanding is Tesla and comma.ai have taken a software focused approach with limited hardware. Others like waymo have taken a hardware approach with more sophisticated sensors. What makes you feel Tesla/Comma.ai will win?

toomuchtodo 8y ago
Human drivers aren’t equipped with LIDAR.
aphextron 8y ago
Because sensors aren’t the hard part. We already have enough hardware to perform superhuman sensing but it’s meaningless without the massive amounts of mapping data and trained models to make sense of it. http://www.youtube.com/watch?v=OikZ6J1YDlI&t=18m20s http://www.youtube.com/watch?v=OikZ6J1YDlI&t=18m20s