Ball drive

only problem is the bumper. You have to buy store sold plywood - cant make your own and 3/4 cant be bent into a 120 in circumference circle. So you got to make a polygon where each side is at least 6 in long so off the top of my head about 15-16 balls max you could make a 20 sided poligone with each side exactly 6 in and hope the inspectors tape measure sees it that way and then somehow place the probably 15 balls as they have a smaller ball circle as they got to be completely inside the bumpers

Polygon, yes. Each side needs to be fully covered by bumper if it’s <6". Try with 3" sides, but good luck with the bumper build…

lol and what kind joint would you use? glue or pin it? That bumper alone might give you an engineering award


To Be Engineered. One of the most feared phrases an engineer can say, type, or otherwise communicate.

I’d probably use brackets bent to the proper angle. But, could be any number of things.

I asked this question once and it caused a stir

1 Like

Cars are glued, should be good enough! :grin:

breaks out the engineering adhesive catalog

1 Like

That thread seems to mention kerf cutting, which is what I would suggest assuming it’s legal.

According to me - yes - according to the way most interpret the rules - no hence my round bumper thread garnered a heated discussion with over 300 comments

1 Like

Gluing/bolting the balls together - 3 balls almost done

Planets with 625 bearings inserted

sun with square nuts inserted and some tem bolts to show how that will work

Motor side planet/partial wheel mount with nylock hex nut (m5) press fit
the hex recepticels for the M5 nuts have a 2.5 deg taper. So its possible to insert them quite easily and then jam them down

The other side. During assy a temp M5 bolt screwed into the nut makes sure you jam it down straight
Also note 2mm spacers printed for the planets and a 1.5 mm clearance to the planets of the body to help keep the lube close and also doubles as an alignment helper during assy.

On the bevel input gear the m3-30mm machine screws are recessed below the 688 bearing that will align everything on the 8mm or 5/16 center shaft the tolerance for the M3 bolts is on the tight side both here and on the sun

Assy with the spacer and the sun gear and the 6806 bearing

Bad news - the sun sits a little to low - so that will require a redesign/adjustment and reprint. I show that to demonstrate - not everything always fits the first time out


Thats a really cool idea. Interested to see how it will work on the charging station and how fast it will be able to go. Cant wait to see the results.


This is so cool! I bet this’ll be fun to drive.

hope so

couple of obstacles to overcome - so I list them here

first I tried regular playground balls - that failed as they were pneumatic and the inflation changed their size. Also if they were hard which they needed to drive then Andymark 6 and 4 in omni had insuficient grab and so did gobilda 5 in.

So next some tpu. and again the above omnis had insufficient grip on the tpu even though their rollers are TPU too I tried 3 different TPU. The rollers on those omni seem to be 95 + hardness TPU. So I have 90 TPU which I printed a roller in and that slipped too (figuring that a sim can put out about 6 in/lb max at 40 amps (without blowing the circuit breaker at a reduction of about 10 to 1 thats 60 in lb so that on a test skid pad on a 6 in wheel is 20 lb pull force simulated with a scale with a partial circle sledge loaded with 60 lb ( a little less than 1/2 robot you probably will have less than 40lb average per wheel (125 lb max plus batt and bumpers is about 150lb /4)

So not having the carpet the drive transfer from the omni to the ball would slip assuming that traction to the carpet is 100%.

So I put holes in the balls and teeth in the rollers and that does not slip


The furniture mover balls I used made for an awful ride on all those hores as they are about 9mm and the holes need to be 3.5 mm. So…

I got some 15 in balls laying around and lots of 4mm balls and some 3d printers - so

I printed some holders on the bottom right is the bouht one with the “too small ball” top left is a 15mm ball assembled and you can see the top piece and the bottom piece with the 8 4mm balls inside - the 15 mm ball will push them against the walls and keep them in place

Now the 15mm ball gives a much smoother ride. I also ordered some 20mm balls but they wont be here until first week in may. So I am starting on a prototype with the existing balls and rollers and the 15mm bearing balls to contain the 6 in ball.

Building something new is a journey.

The current material is ABS - why? Well I got some ABS that was donated and has been sitting here for a little over a year - and filament does not get better with age.

I’ve made some similar ball rollers before from 3D prints, in ABS. Of course they eventually wear out, but if you use acetone to smooth them out to a near mirror finish they last much longer and produced less debris. I think I was using 1/8 or 1/16 in balls behind the larger outer ball.

TPU has less grip than a lot of people think (equating it with rubber). Still a pretty great material though.

yeah I also used ABS and PETG and HIPS and at least according to me - not enough grip

As mentioned in that thread, I think the legal way to get truly round bumpers, (as opposed to a many-sided polygonal approximation), would be to steam bend solid 3/4" wood (oak and hackberry are said to bend best).

Only the way I read it you have to start off with a sheet of plywood and I was told (I am not a woodworker) you cant bend - even steam bend 3/4 in plywood that much. Maybe someone with the proper equipment could try it somewhen

R408. You can use plywood, OSB, OR solid wood. The oak or hackberry above are not balsa and therefore would be legal.

1 Like


I will just buy my round plywood from MarkyMarksRoundPlywoodEmporium. Then I can show the receipts. @Mark_Wasserman

Durometer and coefficent of friction are different things, this line is blurred the more you approach the texture of rough top.