My initial idea here is using the accelerometer to test delta_radius. Since the concave or convex Z motion while attempting to move in a horizontal path is a function of incorrect delta radius(in most cases), we should be able to adjust based on accelerometer readings. It should be something as simple as making a set of moves from tower to inverse side of the bed, for each tower, averaging z motion and adjusting the delta radius accordingly. I’m sure there is a smarter and precise mathematical function for this, but I’m unaware.

This would remove the need for a perfectly flat bed to get a really well tuned delta radius.

This should also be able to detect tower skew by directing linear motion and checking for off axis motion.

Tower tilt could be a function of tilt checked over various heights?

I don’t know if rod lengths could be calculated by accelerometer, if someone does, please chime in with how!

This would also be very helpful for finding tower defects at higher levels since you could test all these variables at almost any point in your build volume.