Sacha Arnoud, Director of Engineering, Waymo - MIT Self-Driving Cars

I believe this talk will specifically and completely answer your questions: ▶ this video Short answer: Self driving vehicles do use deep learning, but they avoid some types not because it is unreliable, but it is processing power intense. Traffic lights, road signs, pedestrians etc.. That's a whole ball of wax to unravel. Check the lecture, its fantastic imo.

The second article I linked to also mentioned work on unmapped roads: link Waymo has solved problem after problem you think they haven't solved yet, for some reason. They've intentionally gone after unpredictable situations and dealt with them. Look at any video where they explain how it works, and you'll notice how many situations they can manage: ▶ this video And even if Waymo only works on well-mapped cities? That's still far from "can only work if they repurpose existing transportation infrastructure to only belong to self-driving cars" which is the comment I was replying to.

It is both an over and underhyped field depending on the area you are looking at. HUGE difference between consumer self driving car [Everywhere, at all times] and the Machines as a Service [Geo, time, weather-fenced operations][1]. Waymo appears to be at the head of the pack; Sacha Arnoud, Director of Engineering for Waymo gave a talk a few weeks ago at MIT[2] and gives a good idea. We are about 5 years out until these start rolling out as MaaS (machines as a service). Probably 10+ years for level 4 highway operations for consumer models according to Frazzoli. [1]Emilio Frazzoli, CTO of nuTonomy link [2] ▶ this video

Google/Waymo demo'd the AI driving in a snow storm and the method they used to see through the flakes and maintain object detection. Having trouble finding the video I'm thinking of. Here are some videos related to self-driving in snow and progress on weird edge cases. - link (snow, like the google demo) - ▶ this video | ▶ this video (snow, but not the one I was thinking of)