Ball drive

As my season is over (the teams I help out) I started my “off season” project. It maybe crazy but then I am an old retired crazy dude lol

What I am attempting is to make a ball drive using 6 in balls instead of tires and arrange it in an X drive or Kiwi Drive. First attempt is going to be Kiwi as I never done one

So first you need some hubs

3d Printed of course - I selected PETG and I stepped a couple of times on it (295lb) So looks like it might be strong enough
They are glued together and use some #6 sheet metal screws to keep them close together while drying. As glue I use gorilla glue which works fine on PETG (and HIPS and ABS etc)

There are 2 TPU partial spheres that are glued to the hub (the tool boxes and bolt box is to keep things in place while it dries. Next to it the polar capse that will glued and put in place with #6 deep recessed sheet metal screws. (at least 1/2 in . At the thickest spot the polar caps are about 18mm at the thinnest the tire is about 5.6mm

pics off a prototype to cage the ball there will be a bottom part that will allow at least 2 in ground clearance max to clear obstacles potentially more . The bearing ball assy are from furniture movers and rated at 120 lb each

parts of an omni wheel and holders amongst others The omni also includes the double helical ring gear of a planetary at its center and some bearings/races to take the load of the gears.

The rollers have teeth to give them a chance with a little wiggle here and there to catch and line up with the holes on the tire which also double as a profile. The hole are small enough to provide enough torque and big enough to appear almost flat to the furniture mover balls. The rollers will be mounted in 2 sections of 6 each offset similar to the 6 in omni from andymark and will have brass bushings and steel shafts and brass eylets to protect the TPU rubber from lateral forces and cut down on rolling resistance as you only have brass on brass and brass on steel contact (5mm rods)

The sun gear which is the input of the planetary will be a bevel setup that takes a cim/neo (8mm keyed ) shaft input.

Those are the planetary holders and the holders for the bearings that function as wheels

Here is the bevel side one with the 6806 bearing inserted that will hold one end of the sun on the outside the sun also has a 688 bearing for the 8mm(or 5/16) shaft/bolt used as an axle. The bolts used for mounting are m5 machine screws the heads are counter sunk and the nuts are captive to allow a tight package. The bevel will also have a 688 gear on the 8mm center axle (most likely a 5/16 bolt as that is 7.937p mm and the bearings slide on really nice on that most 8mm shafts need some work with some sandpaper to get the bearing on. The bearings used in the planets and the ones used as wheels to take the loads off the planets are 625. The gears in the planetare have 1.3mm modulus are 20mm wide and are double helical - herring bone and should give a smooth performance and are strong enough. The helical angle is 20 deg and the pressure angle 14.5 and of course involute with a .5mm fillet as a crown.

Is it going to work - IDK I will also have to make a frame that I probably will make a composite (3dp base with Fiberglass wrap. and some 12 in square aluminum tubing.

This is a candidate - probably will do it with trusses but the shape will be about that. This particular design will have 119.5 in circumference figuring on 1/8 in fiberglass (3-4 layers of 10 oz cloth on the outside. ) Which with the 3/4 in plywood from the bumpers flush against it should be strong enough.

Well its crazy but it beets watching netflix - at least in my book.

Comments/Critique/Ideas/Questions welcome


This is extremely cool so far. Very excited to see your progress. Love these projects!


thanks - oh forgot to mention. What prompted this was they - in my opinion short comings of omnis, meccanum and I think now 3 and 4 in swerve to deal with rough terrain.

What in the James Bruton…. All I gotta say is let him cook


Just about 20 years exactly since @Andy_Baker did this sort of thing, and I believe patented it as well. Of course your version has some differences.

Next time @krf says I need to dial back my insanity I’m pointing to this thread.

Love it. Def watching this thread for updates.


I considered that ball drive but the little rollers on the poles could still hang up

Not aware of that one - you got some links?

The one, the only, the original:

Search term: “ball drive” technokats–that should turn up a few other results.

1 Like

you ever try it without the 2 omnis. Reason why I am going down that road is to be ball only. And how well does it do in a pushing match - Oh I probably got 100s of questions lol. Looks cool

1 Like

Ask @Andy_Baker.

I would say one reason they did 2 omnis was that back in 2003, the motor list was a LOT more limited. 2 drills, 2 CIMs, 2 FP motors (roughly a 775-class motor), some window motors (which at least one team did use in a drivetrain), the globe motors (cute little buggers with an integrated gearbox), and a couple other random motors. And you were quantity-limited to KOP numbers.

1 Like

I tried something like this 4042 Nonstandard Deviation - Ball Drive Reveal - YouTube. Only was not happy with the grip of the omnis to the ball. Hence the toothed rollers and that still requires a gear box on the CIM/NEO and possibly chain/belt. Plus if those are pneumatic balls like I tried first then the air pressure can mess with the whole thing. So in attempting to also do well in a pushing match and to be consistent and not having to worry about proper inflation I went solid with a rubber “tire” and to minimize the space the whole thing takes up I revived an old prototype
of a planetary in a wheel (never competed got killed by covid) and adapted it for an omni. So the space should be the 2 in of the omni, a 5mm gap and the 2.5 in of the thickness of the cim/neo - so less than 5 in plus as its a bevel gear setup (overall is 6:1 to 9ish:1 depending on gear selection - mostly bevel as that is the least work to replace). To allow as much room as possible for intakes and climbers, shooters - whatever the year demands. So you could say thats my ball drive Mk 2.something lol.

Pretty cool - so you’re trying to solve the traction between the drive system and the ball, have you estimated/tested the traction between the ball and the floor? It seems like turning the spikes out on the ball instead of on the drive roller would be more beneficial for traction and the same for that interface, but just trickier for the passive idler bearings you have at the top.

All that motor power is only useful if you can apply it without breaking traction, it’ll be interesting to see how this geometry plays out with stronger materials as you progress.

Have not been able to test it on a proper field yet but that might happen in an off season competition if the mechanics and frame etc work out. To put the spikes on the ball to increase grip on the floor would be nice but I fear it would R201 "no digging in the carpet and it would be a detriment on flat surfaces like ramps and teeter totters This is 90 shore TPU which is slightly softer than the tpu on the andymark tires or omni rollers. Now I can drastically improve grip on smooth surfaces by spraying on 40 hardness TPU IDP How often I would have to reapply it and how necessary it is. Also spikes would make it tough to find a way to coral the ball. The holes are 3mm the balls atm are I think 6 so they bump not too much on the holes about maybe 1mm with 4mm spikes like on the rollers they could hang up. (most likely).

And what do you consider “stronger materials”? I know the gears in PETG will need some lube but silicone spray lube will do the trick and the gearbox is designed to (hopefully) not make a mess. I could go to nylon on some parts and might. But the aim is to keep costs low. Each wheel assy weighs a little less than 400 g it holds me stepping on it and with the material I currently use that means a little less than $6 per ball tire so inexpensive enough to keep a spare on hand the omni should come in at about 7 or 19 using nylon gears plus about $5 in bearings and shafts and bushings and fastners. The frame (if it works) will be about 25 in filament 12 in aluminum and about < $30 in fiberglass and resin. So something in the range I can afford


Makes sense. The swerve tread seems almost on the edge of that rule, since it tends to interact pretty tightly with the carpet texture.

This part under your finger seems at risk for damage since it’s doing torque transfer both from the gearbox and to the ball. Depends on how aggressively you’re going to command it, but it seems like you want to drive pretty hard. What kind of reduction are you getting between this part and the ball?


that part is the roller of the omni wheel Its a 6 in omniwheel on top of the 6 in ball. I got to do some more cadding and printing to get it done - tomorrow hopefully as we have power failure here. The omni wheel setup has about an 8:1 reduction so at max - figuring a CIM at 40 A that is 48 in/lb of torque (6 in lb max torque of a CIM running at about 40A * 8 from the gearbox figuring theoretical 100% efficiency as its 3 inches out from center (6 in omni) that is a weight equivalent of 16 lb. TPU is difficult to test as it flexes a lot and I have not been able to destroy one yet except with saw or drill. So if you have enough pressure then the motor should not be able to skip it before blowing the fuse on the PDP or setting the controller on fire if not fused . The contact pressure is difficult to say but should be an average of about 50 lb as a robot with batt and bumpers weighs about 150 lb and this will work on a 3 wheel setup (kiwi) In my experience it is unlikely that the pressure will exceep 80 lb or go below 20 and the ball has been tested for 300 so maybe I should make a lighter one in the future but then sometimes robots fall off when they climb (not this year) So I expect the point of carpet/ball to be the weak link. The reason there is that omniwheel with rollers on top is to allow kiwi or x drive for nimble moving while still giving adequate pushing power. and (hopefully) sufficient grip on multiple surfaces and definitely better obstacle clearance than pretty much any drive train I know Tank cannot clear obstacles while turning and needs to go straight over, Omis, Mecanum will not go over anything sideways that is > .3 in thereabouts. And swerve will also have to have the wheel pointed straight at the obstacle and yes they can turn but any wheel should have a reasonable ability to clear .35 * diameter. A ball driven by X or kiwi should clear .35 its diameter in any direction without a problem no matter if the robot goes straight sideways or head on. A Kiwi advantage is that with 3 wheels all of them will allways be in contact with the ground. X drive should have more pushing power and you can put a skid ramp in front of ball or wheel to clear taller obstacles. But with proper software 5,6,7, 8… ball drives are possible. Theoretically - figuring the need for some space between wheels/balls the max number of wheels to fit into a 120in circumference should be close to 16 So you could make a 16 sided robot (each bumper length would be >6in) you should be able to fit 16 balls and 16 omnis with the gear boxes inside on top of it. Not that I think its the ideal solution for any season I have so far experienced and you probably run out of spots on the PDP. But as the robot would be almost round you probably could twist yourself out of a pinning situation or move down the field wildly spinning lol. (tornado bot anyone ? )

1 Like

Oh got it - I wondered how you were coupling that to the gearbox! Now you last image makes more sense.

Also agreed TPU is nearly indestructible, but it often just gets out of the way (which isn’t always ideal). Avoiding damage is still a net plus though.

Or an old favorite
The Big One Brochure.pdf (22.3 MB)


Yeah something like this only FRC sized. And we might do a ball drive ftc sized too. This years bot was xdrive