PDA

View Full Version : [Question(s)] Omni wheels - coefficient of friction



Baylink
11-22-2013, 02:38 PM
One factor in "how much torque do you need/can you use" is "what is the coefficient of friction of your tires?"

Anyone know how you figure this in the case of wheels like this VEX Omni (https://secure.robotshop.com/en/vex-robotics-large-omni-directional-wheel-kit.html?SID=hsq1e3q4dgm1lculeb5ohen777)?

I assume those wheels make skid-steer work better, and I assume the rollers are rubber-ish, but how well do they pull, straight ahead?

tician
11-22-2013, 03:31 PM
One factor in "how much torque do you need/can you use" is "what is the coefficient of friction of your tires?"

Anyone know how you figure this in the case of wheels like this VEX Omni (https://secure.robotshop.com/en/vex-robotics-large-omni-directional-wheel-kit.html?SID=hsq1e3q4dgm1lculeb5ohen777)?

I assume those wheels make skid-steer work better, and I assume the rollers are rubber-ish, but how well do they pull, straight ahead?
Effective coefficient of friction depends on lots of conditions, so most of the time it is finding a worse-case scenario with lots of conservative estimates/assumptions and designing everything else for that. Most wheels do not give much other than diameter, thickness, and maybe a load rating, so lots of 'ex rectum' estimates. There will always be some slippage with omni-wheels, but not quite as much as required by mecanum wheels (can make mecanum wheels a not great choice when requiring really good odometry).

Omni-wheels for skid steer, not so much. If you have all four omni-wheels parallel to each other, then you cannot guarantee the bot will not roll sideways from the slightest nudge. This is partly why one or more parallel omni-wheels is often used as the 'caster' wheel of two wheel rovers as it can permit movement in any direction without having to wait for the wheel to spin in place (and mark/shred the floor surface) like a normal caster. If you use a 'cross' arrangement of omni-wheels, you can achieve movement in any direction and the bot should not roll in any direction without the wheels being driven. Does not mean the movement is perfect; while the arrangement minimizes unpowered rolling, the motor control and roller shape can introduce lots of error to the intended movement (Darsha's omni-wheels suck something fierce because of their constant diameter cylindrical rollers that should have used a non-constant diameter to make the wheel's diameter an actual circle).

Baylink
11-22-2013, 04:28 PM
I don't know why that wasn't obvious to me. Would it be worthwhile to put those on my front tri-star bogie, and use normal tires on the back?

tician
11-22-2013, 05:49 PM
Since you are using a passive tri-star arrangement, there will always be a bit of squealing/slipping from the pair of wheels of each cluster/bogie that are contacting the floor. The important question is: will the distance between the clusters/bogies be sufficiently long to cause problems that require the omni-wheels? It might make turning a little bit easier, but are they worth the cost?

Another possibility is to not power the omni-wheels and only put motors on either the forward or backward clusters. Would not be able to turn in place about the center of the body, but somewhat less complicated and differential bases are already widely used with ros (Maxwell, TurtleBot 1 & 2, UBR-1, etc.), not that it really matters as there are also four wheel powered caster (PR2), front steering 4WD (Robotnik Summit), and omni-wheel 3WD (Robotino) bots working reliably with ros.

Baylink
11-22-2013, 06:04 PM
Since you are using a passive tri-star arrangement, there will always be a bit of squealing/slipping from the pair of wheels of each cluster/bogie that are contacting the floor. The important question is: will the distance between the clusters/bogies be sufficiently long to cause problems that require the omni-wheels? It might make turning a little bit easier, but are they worth the cost?

By "passive" did you mean "the star itself's rotation is not independently powered"? Cause my design was 4 bogies, one in each corner, all powered.

The outside dimensions of the chassis base are 18" wide by 29" long, as designed, though I can bump those by 2 or 3 inches if I have to, so the interior dimensions at the bogie tracks will probably be 2-3" less than that.

Mwalkeriii
11-22-2013, 07:41 PM
I found a site that listed a number of coefficients. I'll see if I can find them. (Some time passes).
Found them!

http://www.chiefdelphi.com/forums/showthread.php?t=50284

-Milt

tician
11-22-2013, 07:48 PM
By "passive" did you mean "the star itself's rotation is not independently powered"? Cause my design was 4 bogies, one in each corner, all powered.
Yep. And I finally remembered which real robot used an actively controlled bogie for balancing: the iBOT (http://en.wikipedia.org/wiki/IBOT).


The outside dimensions of the chassis base are 18" wide by 29" long, as designed, though I can bump those by 2 or 3 inches if I have to, so the interior dimensions at the bogie tracks will probably be 2-3" less than that.
Don't use skid-steer outside of the tiny bioloid carbots, so no real idea if that will be too long and narrow for easy skid-steer or not so easy skid-steer. Others might know some general rule for length to width ratio limits of skid-steer.

Baylink
11-23-2013, 10:21 AM
Thanks, Milt.


Yep. And I finally remembered which real robot used an actively controlled bogie for balancing: the iBOT (http://en.wikipedia.org/wiki/IBOT).

Wow. I'd forgotten about that. *Damn* those motors must be torquey. Wonder how you calculate *that*. :-)


Don't use skid-steer outside of the tiny bioloid carbots, so no real idea if that will be too long and narrow for easy skid-steer or not so easy skid-steer. Others might know some general rule for length to width ratio limits of skid-steer.

[ looks ] Ah; "Bioloid" is someone's robotics kit.

I could, I suppose, go to completely passive front bogies, with the omni wheels, but that will double the required torque and friction on the rears, and I'd rather avoid that if I practically can.