Just a quick mock-up of a idea for a omni-directional drive wheel. Four wheels with 15 cones for rollers, mounted perpendicular to the wheels rotational access. The cones are 2"-1" diameter, mounted so that the entire cone is touch the floor. Idea is that if you have two wheels with the larger diameter cones to one side, and two to the other side, you can drive in any direction by mixing the speeds. Completely logic right now, no more then just thinking it up and sketching it out done so far.
I may be reading your post wrong, but you say the rollers go from 2 inches to 1 inch in diameter, so how big is that wheel itself? I can hardly imagine its anything less than 16 inches given the size of your rollers.
The design seems kind of cool, but you might want to scale the size down a little. That wheel is probably a too big for a practical application in a FIRST robot.
You also may want consider adding some supports inside the wheel, I know a lot of teams that have trouble with plates bending on their mecanum wheels.
Idea is that if you have two wheels with the larger diameter cones to one side, and two to the other side, you can drive in any direction by mixing the speeds.
Can you support this assertion with kinematics equations? I’m not seeing it (at least in a way that is not terribly inefficient).
How does the wheel get traction? It appears that as the wheel is driven forward or backward the rollers will spin and not generate any tractive force.
Those rollers aren’t mounted perpendicularly, so they will just free spin. If you want a sort of omni-directional drive, I suggest investigating mecanums. There’s no need to re-invent the wheel.
They aren’t perpendicular but they won’t free spin. They aren’t mounted on an axis parallel to the floor. If the whole thing is in contact with the floor there is no way it will spin since the two points touching would try to go different speeds. Picture it as 2 wheels that are different diameters fixed to a common axis but both touching the ground, if you move it they will need to be moving at different rates and since this isn’t possible they SHOULDN’T free spin.
Are there reasons to do something like this? Possibly. Will this give the desired motion? Not likely. (I haven’t sat down and really thought about it enough to say definitively)
I see your point, though it won’t slip ONLY when 2 are in contact with the ground, which will be a fraction of the time that the wheel is spinning.
I do not, however, see how this achieves omni-drive because the wheels cannot roll with a component of motion parallel to their drive axle like an omni wheel or Mecanum wheel.
No, only a single roller needs to be in contact with the ground for this to work since the two wheels Andrew is referring to would be the opposite ends of a single roller. However, I can’t possibly imagine how this would result in omni directional movement. Both styles of omni drive that FIRST sees (holonomic, and mechanum) are based upon a wheel that can only produce a force vector in a single direction and will spin freely in the other direction. This wheel will not freely move in any direction when driven, if fact it should produce a torque vector perpendicular to the floor upon which it rests.
Although that could make an interesting Swerve system. Use the torque vector produced by the wheels to turn the modules, basically moving the wheels similarly to a mechanum when it strafes?
My brain hurts,
I get that they won’t spin freely, but then wouldn’t the individual rollers lock causing the wheel as a whole to act as a normal wheel?
Yup…
Aha, I understand, thanks. I agree that this could work in an ideal world.
My intuition says there might be a problem in this situation: the narrow end of the roller will have less grip, and will need to generate more tractive force than the wider end of the roller. This is because of the larger contact area on the wider end of the roller, and the lower tractive force needed to balance the torque internal to the roller. Does that make sense?
Consider a simple static analysis.
A standard wheel of radius r being driven with torque tau would produce a forward force of tau/r, due to the tau/r reaction force of the floor on the wheel.
The cone-roller wheel of same radius being driven by the same torque would produce the same forward force, but it would lose traction and slip at lower torque levels than the standard wheel, so it would lose a “pushing contest”.
Up until the point where the roller starts to slip, the net torque on the roller must be zero. The reaction force of the floor on the roller at the 1" end will be pushing forward (call this force F1) and the reaction force of the floor on the roller at the 2" end will be pushing backward (call this force F2). F1 must be greater than F2 in order to balance the torques on the roller. F1 will be greater than tau/r, so traction will be lost and the roller will start to slip at lower levels of tau than with a standard wheel.
So for pure forward motion, the cone roller is inferior to a simple standard wheel.
And as others have stated, the cone rollers with axes in the same plane as the wheel axis provide no omni capability.
From a kinematic standpoint, any motion of the vehicle in any direction will involve relative motion (scrubbing) at the roller-to-floor contact interface. This will absorb energy and result in lower speeds.
~
Thankfully Bengt Erland Ilon disagreed with this statement back in 1973 when he dreamed up the Mecanum wheel.
I’ve enjoyed contemplating this idea and am delighted to see that people are finding inspiration, through FIRST and this forum, to contemplate designs. This one might not work exactly as planned, but keep coming up with the ideas. It takes courage and determination as much as it does brilliance and inspiration to be a successful innovator.
Just google… quote “can’t be done” …to see what Henry Ford and others had to say on the topic.
Jason
The flip side is that Ilon used his knowledge of kinematics to make a mecanum wheel, rather than drawing something in Inventor and seeing if it’d work. 
I still don’t understand what you were going for with this, because I don’t see how this could be used to achieve omni-directional movement.
-Nick