Incorrect CAD data for Robotis parts

[petercohen]

Hi, as you all know, the CAD data of robotis products are available for download from their website. I have compared the weight of several motors and some frames. I found that the mass of the parts from the CAD data are very different from the corresponding real ones. Why the mass of the CAD data are incorrect? Can I trust the Center Of Gravity of the parts as well? Is there any way to obtain accurate mass and COG data of robotis products (especially the Bioloid)? For your reference, I use Autodesk Inventor 2013 to obtain the mass and COG of the STEP data files. Thank you.

[Gort]

Very interesting!

I have used almost all of the STEP files to 3-D print out parts for my Bioloid humanoid. So far I have had no problems. Is it maybe a software issue? On creation of the files or something happening in Inventor? What is mass difference?

[petercohen]

Thanks for your reply. The sizes and hole placements may be correct. However, the mass of the parts are different from the real ones. In Inventor, I used iProperties to obtain the mass and COG. I found that if I choose different materials, I also obtained different masses. How do I decide which material to choose? For the frames, it may be easier (tried various PVC, plastic options but still do not get the correct weights) but for the dynamixels, there is no single option as the motors are made of different materials.

Which CAD program and file format are recommended to obtain the mass and COG?

An Important question: Can I assume that even the mass of the virtual and the real ones are different, the COG is correct? If so, I just use my measured mass and use the COG and dimensions in my simulated model.

[Fritzoid]

petercohen,

I went through this when I created the model for use with OpenHRP. The calculated weights failed to match the actual weights and I was forced to customize the material specification to match the actual weighed parts.

I also found weight differences between two otherwise identical brackets in the same kit. Seems the plastic is not homogeneously spread within the injection mold from one forming to the next. So I weighed all the brackets of one type together and divided to get an average mass for each part.

From a CAD software perspective the parts contain uniformly distributed resin and so the COG calculation will be independent of the mass of the material and close enough for your purposes. This works fine for individual brackets but not for the servos. Also, some adjustments will be necessary when combining the parts into assemblies for the additional weight from the bushings and screws.

You can find the parts and assembly files in Alibre format in my files area at http://robosavvy.com/Builders/Fritzoid The bioloid_assemblies.zip file contains a text document (parts_specs) that lists my final calculations. It should be interesting to compare our results with mine.

[limor]

If you are interested in Bioloid parts there are also these ones that were precisely measured. the CG of the AX12 was measured too.
http://humanoids.dem.ist.utl.pt/Iden_ex ... rview.html

[Fritzoid]

While the above link is extremely useful particularly for the AX-12 COG, the data was generated using parts from the comprehensive kit which used the same molds but a different plastic resin than the premium kit. So the dimensional and COG data is useful, but the part masses are not correct for a more recent Bioloids.

I was unable to reconcile the moment of inertia numbers at all!

[Gort]

This is all very interesting.

I know they have changed both the manufacturer and plastics used for the brackets.

Sounds like I need to weight out brackets in the kit and compare them to the ones that I have 3-d printed out. I need to see weight difference between PLA, ABS and PC.

I know that the brackets are made with PC so that would be a good test to see weight differences between the injected molded plastic parts and the ones that I am 3-d printing out.