I have not been happy with the logic of our existing calibration code, so while waiting for other things to finish at about 2am I decided to present the problem of calibration to grok. it wrote a program for me that takes the data collected during calibration and attempts to find the anchor points. it then looks at the points and sees if any of them appear to be outliers, and if so, reports them and does the calculations again without those points.
this is REALLY fast (pretty much instantaneous)
the results look plausible for my frame, but there has not been any additional testing. so this may not be producing good results
but I think it’s interesting enough to share
