For the centimeterish precision needed to hover into the chopsticks, they also have the opportunity to use signals from the tower area for final alignment. I'm thinking riding a beam like aviation ILS. Just speculating but it would be easy to implement.
Optical/camera alignment is probably out of the question due to fire and smoke.
The arms themselves could have sensors on them. Inductive loops sensing the presence of the stainless steel structure?
Also, I doubt centimeter scale precision would be needed; the arms have some compliance in them, I'm sure, as well as the ability to control how far in they swing.
Optical/camera alignment is probably out of the question due to fire and smoke.