Modern cars can operate high-beam headlights automatically:
They automatically switch from high-beam headlights to low-beam ones (less intense) when you enter a town or there is a car in front of you either going in the same or opposite direction so you don't dazzle other drivers or people in the street.
Oppositely, when you are in almost complete darkness and there aren't any other drivers at sight, the system automatically sets the high-beam headlights.
I am aware that in the front part of these cars there is a camera or sensor and also imagine that the automatic switching when entering or exiting a town is achieved by just having a threshold ambient illumination.
But I am unable to imagine how the recognition of other cars works. It might be that an image recognition program is used to detect pairs of front car lights (white) and rear lights (red). However, how do you deal with:
- pairs of street lamps (far from the road and not illuminating it) that could be identified as a car coming in the opposite direction,
- pairs of lights coming from the reflectors of the crash barriers,
- many other random pairs of lights that could be interpreted as cars.
Is this technology based on AI software that after intense training is able to deal with these points? Or is it a less complex image analysis program that takes into account that moving lights outside the car (with respect to the road) move differently than static lights (with respect to the road) when seen from the car?
Edit: I've seen this technology working on Audi A4 and A5 cars.