What kinds of techniques do autopilots of autonomous cars (e.g. the ones of Tesla) use? Do they use reinforcement learning? Which types of neural network architecture do they use?
I will focus on Tesla's autopilots in this answer (because that's the only specific autopilot you mention).
In their website, they mention the basic technologies underlying the current autopilots, which includes deep neural networks.
To make use of a camera suite this powerful, the new hardware introduces an entirely new and powerful set of vision processing tools developed by Tesla. Built on a deep neural network, Tesla Vision deconstructs the car's environment at greater levels of reliability than those achievable with classical vision processing techniques.
So, they are probably using convolutional neural networks for the computer vision tasks.
There's also a Wikipedia article completely dedicated to Tesla's autopilot.