On my latest calibration it started around 0.36, and climbed steadily and just kept going over 1.
It’s at 1.18 now and leveling out. I think this has happened before with older hardware/firmware too.
I just assume 1.0 would be the highest?
1 Like
No, fitness is 1/ average estimated error in mm
There is no upper limit.
David Lang
1 Like
Nice that this topic come up again!
Been meaning to ask the question, since I couldn’t find a specific/detailed explanation of fitness number:
So David do I understand you correctly if I’m saying a higher fitness number is actuality better than a lower fitness number, given that for a higher fitness number the estimated error in mm is lower?
Second question is how should I interpret fitness number? Is fitness number just something for the firmware to adjust of off to control the motors?
Just curious!
Thanks Waldo!
Waldo-cnc wrote:
So David do I understand you correctly if I¢m saying a higher fitness number is actuality better than a lower fitness number, given that for a higher fitness number the estimated error in mm is lower?
Yes, it was setup to be a ‘bigger is better’ number
Second question is how should I interpret fitness number? Is fitness number just something for the firmware to adjust of off to control the motors?
No (as I understand the details) the fitness number is just taking all the
measurements during calibration (the belt lengths of each measurement point) and
the final coordinates of the anchors.
it then takes each set of measurements and tries to ‘connect’ them to the anchor
location. Due to various factors (including the fact that we only measure the
belts down to something like 1/80th of a mm. belt stretch, frame flex, etc) the
belt lengths aren’t going to exactly meet at a point, so the difference between
where they do end is the error for that measurement point. For one point, the
belt to the top left may be a little long, for another point the belt to the top
left may be a little short. But it wasn’t able to find an anchor location that
worked better.
it does this for all four belt measurements for each point (4 x 81 for a 9x9
grid) and averages them together. 1 / this measurement is the fitness
the system doesn’t have any way to improve the anchor location calculations,
it’s just reporting it so that we can get an estimate of how accurate the
location calculations are.
David Lang
1 Like
Thank you for explaining this David!
1 Like