Maslow 4 - Fitness Results Comparison

So far the only benchmark I have been able to find on the Fitness Score to understand if the Maslow 4 is well calibrated is Bar mentioning he gets ~ 0.7 for Fitness score.

Wanted to create a thread that as people calibrate their machines they can share the Fitness scores they have been getting to see on average where people land and any precision observations at those Fitness Scores so we can over time come to a range of reasonable Fitness Score (i.g. 0.5-0.8 is great and lower then that is a concern as an example)

I got ~0.5 for after my first successful Calibration. Wanted to run it again to see if I get a better Fitness score but ran into an issue when running calibration again.

1 Like

I’d suggest that people provide a little information about their setup along with their score. The most relevant things I can think of are approximate dimensions of the anchors and whether they are vertical or horizontal. I’m just curious if there’s any correlation between those things and the fitness results.

Edit: based on another thread, it sounds like the score is affected by calibration pattern size, so if you are using a pattern other than the default dimensions, note that too. Also Ken32960 suggested that we include firmware version as well, which is a great idea.

3 Likes

Resolved some motor issues I was having and got a fitness score of 0.7658! Now to start cutting and seeing how this translate to precision results.

Stats:

Frame Orientation: Vertical
Frame Height: 5.5 ft
Frame Width: 10 ft
Fitness Score: 0.7658

2 Likes

Excellent work!

I wonder about fitness vs accuracy myself.

I recently ran two calibrations. On the first I got .78, downloaded the latest and input the values, ran it again and got .66.

No hardware changes whatsoever.

Maslow values do not match my tape measure when I measure the frame myself. I measure vertical as 2426 at both locations. Last Maslow was 2435 and 2427. Prior 2433 and 2429.

12x8
Vertical

I have run calibration several times, never with repeatable results.

Is the calibration process a close process that only gets you in the infield and that’s good enough?

Is what I am seeing normal or keep digging?

How unrepeatable are the results? Mine tend to move around by about 2mm. Ideally we want to get that down to zero, but that’s the baseline for where I think things are right now

+or- 2mm for me as well.

The repeatability of the measurement is plus minus 2 mm. The accuracy vs manual is within 8mm. My gut tells me it should be closer.

I am more concerned that it measures 2435 when I know that should be closer to 2426. If I do the math on the members I cut and joined I get 2427.1 so I feel 2426 as measured by me is fairly close.

I am also wondering about the range in fitness with no hardware changes and the relevance of fitness to accuracy in that scenario.

I think that could be related to this:

I think that the calibration process ends up giving better results when cutting than entering the numbers manually would because the errors are consistent across the system…IE we’re defining the spacing of the teeth on the belts to be 2.0000mm and then using that number everywhere so then when we go to cut a straight line it’s still straight. If we measure by hand and enter the values we end up with two different systems of measurement.

It’s kinda like if if I built a house and my tape measure was off by 10% it would be OK, the house would just be 10% smaller than intended, but if you built some parts of the house and I built some parts of the house and our tape measures were different and then we tried to join the parts it wouldn’t work well.

Ultimately investigating that is top of my todo list once I get the basics of getting everyone up and running is taken care of.

+or- 2mm for me as well.

The repeatability of the measurement is plus minus 2 mm. The accuracy vs manual is within 8mm. My gut tells me it should be closer.

I am more concerned that it measures 2435 when I know that should be closer to 2426. If I do the math on the members I cut and joined I get 2427.1 so I feel 2426 as measured by me is fairly close.

I suspect the difference is either belt stretch or the belt not being exactly
2mm per tooth

try doing the calibration with different max current values.

I am also wondering about the range in fitness with no hardware changes and the relevance of fitness to accuracy in that scenario.

not enough testing.

David Lang

Frame orientation: Horizontal
Frame Height: 2481mm
Frame Width: 3985mm
Fitness score: 0.62236
Software Version: 0.66.5
X & Y offset: 800, 800

To be fair I was caught off guard with UPS arriving a few days earlier than expected so I nailed some short bits of 2x4 to my frame, drilled 7mm holes and hammered in some m8 bolts I had laying around to be able to begun the calibration :stuck_out_tongue:

I’ll go shopping today and repeat with the correct mounting system to see if it improves.

1 Like

Good example.

Can we add…
Software Version: 0.66.5 (IMHO very important)
Maybe even…
X & Y offset: 600,600 (Not as important)

1 Like

I’m using a starret and bars using a belt.
Something about everything‘s relative Goes here

1 Like