Hanging out with a self driving car and its designer.
He says: “It’s not our thing, but V2V (Vehicle to Vehicle) communications are cool, because then you can see around corners.”
I asked him about the sensors, which prioritize LIDAR (top) and radar (front), and not cameras. I’m paraphrasing here, but conversation was more or less this: The car doesn’t use road signs to get optical input.
The road signs are for humans… the signs just reflect the information on maps that humans are bad at remembering. The car goes off the real-time maps, which are up-to-the-second in a way those signs can’t be.
Sometimes the maps are more accurate than the road.