I built the first version of my omni-sphere treadmill a while ago and went over my opinion of it in this video.
Since then I've spent some time thinking about how the design could be improved. The idea I have is mainly around changing the rod that the omni-spheres are mounted on. The changes would make the rod easier to assemble and eliminate the extra space the support columns take up.
With the design I built, the support columns used to make sure the rods do not flex under the user's weight take up extra space.
You can see in the image that I had to space the spheres out more to make room for the columns. I want to push the spheres as close together as possible, so this is not ideal.
What if instead, I could put the support columns at the locations shown in the image below?
These are all the points where the X and Y rods cross each other (on different Z planes). Support columns in these places may not take up much extra room, and there would be more columns to better support the user's weight.
I figured I could place support columns in these places if I could imbed a bearing into the rod itself, like shown below.
To do this, instead of using a 1/4in round rod, I'd use a tube. Specifically a tube with a 1/4in OD and an ID as close to 3mm as possible, 1/8in should be okay. I could try to find tubes that are in metric units with an ID closer to 3mm, but being in the US, and because I will need a lot of tubing if I am going to build this, I'd like to use tubes in imperial units since they're cheaper. Something like this tubing which is cheap, and shipping is free for orders over $100.
Assembling the rod would look like the image below. The 3mm rods would go through the 3mm bearing and the tubes.
To make sure the 3mm rods do not spin freely inside of the tubes, I'm thinking I could use epoxy.
With the 3mm bearings being imbedded, the support columns could look like this. They take up less space than the columns I had previously.
With this design, the spheres can be spaced closely together while the rods themselves are very well supported, as shown below.
Additionally, I think a treadmill where the rods are actually metal will decrease the bumpiness. In the build I already made, I made the rods out of plastic because, to be fair, I thought it would be fine to be lazy and let the 3D printer make all the parts. However, it's possible some of the bumpiness was due to the rods being made out of plastic, which means they could have flexed a bit and allowed the spheres to bounce up and down a little.
The last build was also much bumpier with my shoe than with the piece of wood I tried. Perhaps the user can just wear an overshoe that has a rigid, hard surface on the bottom like the piece of wood to reduce bumpiness and noise.
So in summary, I believe that if I use this new design to make the rods with embedded 3mm bearings, I can make smaller support columns that will enable me to push the spheres closer together, I can make overshoes that are rigid on the bottom, and using metal rods, will all reduce the bumpiness and noise I saw with my previous build.
I plan on making a new build that utilizes this design some point in the future, probably within a few months as of the time I am writing this.