Driver+

[thrill]

The loss of lidar means a loss of precision. Roughly, radar allows resolution of about 1%, and lidar about 0.01%. If your software (and hardware too, natch) is not yet capable of doing enough with the extra detail to make the expense worthwhile, then I can see the economic decision being made to leave lidar off for now, disappointing though it may be to us geeks.

However, radar has some advantages in open-road driving (not in crowded cities, etc.) in that it penetrates moisture much better.

Sponsored

 

[ElectricDan]

Very true, there's a lot of "ifs" when it comes to what may or may not be valuable. I have experience with some of the underlying tech, but not the AD use case.

Tags would imply supervised machine learning, where as a lot of AD is now using unsupervised/deep learning/neural networks.

I would hope that field of view & relative position would be easily changed variables, to allow common algorithms to work for different types of sensors and sensor locations. I really have no idea how they do it, but there are a lot of really bright minds working on this stuff.

I guess another big question would be what is rivian using for compute, and can that system handle that volume of data, in the event it is usable.

 

[barsaec]

Good article on competing tech:
https://insideevs.com/news/467357/mobileye-ceo-tesla-autopilot-will-fluster/

Sponsored