I'm pretty sure I read that Tesla uses the Bosch radar system. Seems like when I read the datasheet (which I now cannot find) it had 2x3 and were fixed in position, but adjustable by calibration. I don't think they actively move when in operation, they just classify which zone they are in.
No automotive radar has moving antennas, they are electronically swept by adjusting the relative timing of the signal emitted by each individual antenna. To aim the beam to the left, each antenna starts each individual wave (not pulse) slightly before the antenna to its left, to steer the beam right, slightly after the one to its left, for straight ahead they all start each wave at the same time.
Phased beam steering sounds a lot more mysterious than it is. Think about light as a wave, like a water surface wave in a shallow pool. The first thing to realize is that because there isn't anything holding the wave together (unlike a particle), the wave essentially restarts at every instant from whatever configuration it's in at the moment. That means that if you can get a wave train into a certain configuration, that is with all the peaks at a particular position and frequency, it doesn't matter how it got there, all wave trains with that configuration behave the same way no matter how they were created.
Now, think about a wave train with peaks running squarely left to right (angle = 0) hitting a wall which is at an angle to the waves, it will reflect off at an angle, the reflected wave's peaks at some angle to horizontal, say 30 degrees, moving away and to the left. Not surprising, but what is the difference between the the reflected an original waves? Measuring the original waves peaks distances from some wall at 0 degrees left to right, situated behind the reflecting wall, at any moment all the points on a single wave peak crest will be at the same distance from that wall, since a line along the crest is parallel to the 0 degree wall. Now, again measuring from the same 0 degree wall, but to the reflected waves, at any moment the distance to a single wave peak crest will increase as you move left to right since the angle of a line drawn along the crest is 30 degrees.
However, because we've earlier established that it makes no difference at all how a wave got into a certain configuration, it's possible to create both these waves, parallel and angled, by first removing the angle wall and using a series of water emitters on the 0 degree wall, and very accurately controlling the timing of when the water is pulsed out of each. If all the emitters pulse together, it produces a wave whose crest is at the same distance from the wall no matter where you measure it from left to right, although again, over time that distance will keep increasing as the wave moves away from the wall. That looks just like the original unreflected wave, although moving in the opposite direction.
If, on the other hand, each emitter puts out its pulse a tiny bit before the one to its left, at any moment its pulse will be a bit further away from the wall than the one from the emitter to its left. But that looks exactly like the reflected wave, the distance to the 0 degree wall increases from left to right and the line drawn along the crest is at 30 degrees. We've just steered the wave not by changing the angle of the wall with the emitters, but by carefully adjusting the timing of when the emitters pulse. If each emitter pulsed slightly before the one to its right, the wave would go off in the opposite direction and so on.
Apart from not having to physically move the wall to change the angle of the waves, it also allows the angle of the waves to be very rapidly slewed left to right because all that's needed is to change the emitters timing relative to each other which takes no additional energy to do. You do need fast valves and controllers though, and for radar that means very fast electronics, but not only are they available, they're cheap.
