Flying Ducks!

I 3D print my articulated propeller design, and 3D print a scanned ICRA duck.

Updated: October 6, 2015

Personal Project Goals

For this semester, I am tentatively planning on creating a multirotor with special features for a final projecet (see week 1). I designed articulated propellers that week. This week, I got to print them out and test them.



I'm cheating a little bit here. I already have experience with the Objet, since I use it in my own research. I already know that a good separation between walls in order to prevent the object from fusing is 0.4mm. So, I already had that factored into my design. The propeller only took a single print to get working. From this, I was able to print out a working articulated propeller, as seen below. The prop .stl file can be downloaded here


Read More / Hide

Scanning is slightly newer to me. I have experience with structured light projection scanners as well as hand-held Kinect sensors, so I wanted to try the turntable scanner.

I'm ready for my closeup, Mr. Duckville

This scan went mostly smooth. I set it to high definition with eight subdivisions. I got out a duck point cloud scan that looked like this:

Dirty Duck

I used the software to manually clean up most of the outlier points, giving me this slightly cleaner duck.

Cleaned Duck

Holy ducks, Batman! The built-in software for the scanner decided to patch this duck by making it toplogically equivalent to a torus rather than a sphere - not what I wanted. So, I took it back and played with it in Meshmixer. Meshmixer has lots of tools for manually sculpting meshes, but these proved unwieldly for the task of creating a smooth, nice, watertight mesh, which we need to print.

For some reason, I didn't think that Meshlab's Poisson reconstruction would work well, because I thought the holes would be too large. But lo-and-behold, I dropped the holy duck in, ran the Poisson reconstruction, and voila!


It's our desired duck! He lost some of his features - his eyes and part of his bill got very smoothed over. This is due to the Poisson reconstruction settings; for the purpose of creating a print, the resolution in the print probably won't lead to this difference being visable. I could have increased the OctTree depth in order to get better resolution, for future reference.

Current Issues

There really aren't many issues right now. I need to make the propellers less brittle, or print them with Endur - they snap easily. They'll also need to be made longer for flight, and I'll have to hollow out the inside so I can place a motor shaft. But beyond that, this week went very smoothly.