We are getting about 1 day for every 7 when the skies clear enough for me to do final keypad testing. The keypad has two methods of modeling that allows unguided imaging. The first one does a limited pointing model along the path that the image will take during the night. The second method simply measures the drift rate of a star at or near the object to be imaged. This second method is what I did last night. For most applications one needs only to take a short drift measurement of any star in a frame at maybe 2 to 4 points along the path that the image will follow. One point every 1 to 2 hours RA along the path will produce a very accurate model for drift.
Below is a 15 minute UNGUIDED exposure I took last night with the Mach2 mount, purposely NOT polar aligned, off to the point where RA and Dec were drifting at 16 arc sec/hr and 88arcsec/hr respectively. I took a single 5 min drift measurement at the beginning of the exposure and used that to set the custom tracking rates for the axes (all done automatically in the keypad). It was good for the next hour of unguided imaging. A second and third drift measurement was added for 3 more hours, but clouds moved in.
With a good model the accuracy is better, in my opinion, than guiding with PHD2 or MaximDL. There is no jumping around, back and forth pushing and pulling of the axes according to momentary guide star positions and sky scintillation. It's a smooth motion, very accurate tracking under precise encoder control. This is with a 160EDF refractor of 1200mm focal length at .9 arc sec per pixel image scale.
The drift model resides in the mount controller, so actually doesn't need a laptop to store the data. The drift measurement can be done with a consumer digital camera, and then that camera can be used all night to gather images without ever having any guider attached.