This originally was done with a blue LED source in the lab, the same as the ones at the scopes, but I recall Judit felt that the green laser was close enough in wavelength for this to work. Here is the process as I remember it. You will need an engineering version of the LABAOGTK.
Do a full alignment check of the beam in question. Make sure it is hitting the east end target, centered on the DM, and also hitting the BRT target near the secondary and going straight down the delay line rails. This alignment should also ensure the visible and IR beams are going to the right places in the lab.
Place a flat mirror where the corner cube normally goes near the BRT primary. There used to be one ready for this, including picomotors that can plug into the spare cables for this in the lab.
Autocollimate the flat as best as you can. Ultimately it should send the light through the pin hole in the laser source. I used to use the theodolite to at least make sure it put the spot in the same place as the corner cube.
Make a copy of the old /ctrscrut/chara/etc/LABAO_S2_WFS.dat file for the scope.
You can then use the green laser on the wfs to zero the boxes and save the positions. Make sure you are looking at the correct set of boxes, you want the ones for on-sky, not the reference ones.
Block the light from the auto-collimated mirror and open the S2WFS shutter for the scope and change over the the reference box positions. You might want to move the reference mirror to center the spots but I hope not. You can then ZERO the boxes for the reference position.
Double check that both box positions are OK, perhaps by doing a dichroic auto alignment and checking it still works, and then save the box positions.
The simple answer is to use the blue beacon after having carefully checked the alignment. This assumes that the focus, as determined by the lab WFS, is correct. This focus includes both the OPLE and the BRT alignments, but since we assume that anyway while observing it's probably OK. Other than that follow the same procedure Matt describes for the scope AO system.
If you don't believe the BRT and/or OPLE are properly aligned (and they probably aren't) here is what I think should be done.
Check the alignment as above. Use whatever the current default flat is.
Make a copy of the current default flat.
Use the engineering labaogtk to see where each actuator is.
You can then use the DM page to set all actuators to the same value, something close to the mean of the current flat should work.
Find a value for this so things are close to in focus on the beacon. Putting the same number on all actuators should make the DM spherical and is what Judit always advocated.
Turn on the lab laser and send it down the rails.
Put the cart at the back and use the on-rail two hole mask to check if the outgoing beam is in focus. If not, change the number on the DM until it is.
Now check that the beam size is correct (5 inches).
Here's the iterative part. Moving the BRT secondary to get the beam size right, and changing the DM focus (all actuators on the same number) to get the beam in focus going down the rail. I'm not sure it's worth being too picky about the beam size so long as it's close. You can easily get lost in parameter space here. You should also in the process try to keep the DM actuators away from their limits.
Once you're done, check that the OPLE is in focus by looking at the return beam with the on-rail two hole mask.
Turn off the lab laser and turn on the beacon.
Focus the beacon to this new setup.
Close the loop on LABAO and get a new flat.
Check that this new flat doesn't push any actuators to the limits or too close to them. If it does you will need to go back to the iterative step above.
In the past Judit used the two hole mask at the scope and an optical setup in the lab to make sure the beacon was well focused and did this process in reverse. This is indeed ultimately the right thing to do but is more diffult and time consuming, and requires things being calm enough at the scope for the beacon not to move around too much. Also I believe the 2 hole mask at the scopes has been removed in some cases.
As a first step I would suggest trying to get a new reconstructor using the beacon and not messing with the BRT and so on. Then using that reconstructor to get a new flat on the beacon.