Can someone please explain where the innovation is here? I’m not very familiar with solar generation.

Given that we know the suns position and that of the mirrors, how is computer vision able to better aim the mirrors?

What am I missing?