16-01-2015 12:08

JesseK

Re: pic: McCannot Wheel

It's like a planetary transmission that doesn't need a ring gear!

16-01-2015 12:17

Mr. Lim

Re: pic: McCannot Wheel

I posted this because there is some disagreement whether Mecanum wheel rollers will roll when driving straight forward or backwards. Some people say they will roll, and some people say they will stay stationary.

In my experience, I've noticed that Mecanum wheels often spin faster than the robot is moving, even when driving straight forward and backwards, much like the McCannots would, except not nearly as badly. Many say that it's because the Mecanum wheels are losing traction with the carpet, but I didn't see that being 100% the case. My theory is that the Mecanum rollers do roll, even when going straight forward or backward.

I think it's fairly obvious that these "McCannot" wheel rollers will definitely roll when driven straight forwards or backwards, resulting in less than ideal robot movement.

My question is for those who believe that mecanum rollers WON'T spin when driving straight forward or backwards. How many degrees do you need to start angling the McCannot rollers until they stop rolling completely when driving straight forward or backward? 90? 45? 1? 0.00000001?

I don't know the answer to this, but I'd be curious to see what others think...

16-01-2015 12:34

Ether

Re: pic: McCannot Wheel

Look at my force and kinematics papers and it will answer your question.

Change theta from 45 to 35 to 25 to 15 to 5 to 1 to 0.00000001

and then from 45 to 55 to 65 to 75 to 85 to 89 to 89.999999

and observe what happens to the forces and wheel speeds.

Then think about what happens if roller friction or axial free play is non-zero.

16-01-2015 13:04

brynnanotbrenda

Re: pic: McCannot Wheel

It seems like your "McCannot" wheels are just Omniwheels, minus the axle.

16-01-2015 13:05

brynnanotbrenda

Re: pic: McCannot Wheel

Nevermind, the wheels go a different way.

16-01-2015 13:26

Culvan Van Li

Re: pic: McCannot Wheel

It is interesting that you posted this. I've been giving this design some thought recently as a way of proving that force applied through a roller can be effectively lost with no useful conversion.

I believe in a perfect setup the mecanum rollers wouldn't roll in straight forward/backwards driving. However I suspect there are artifacts of reality that probably make them spin in most cases. There are so many ideal design constraints. If the wheel centers aren't at the corners of a square you will get some slipping that will result in roller movement. If the weight isn't evenly balanced it could cause the same.

This is clearly true for a robot with only McCannot wheels. Let's take a moment and consider the case where you have 2 McCannots in front and the other wheels are standard traction wheels. Let's take the example further and say all the axles are coupled together so the wheel movement is exactly the same. Let's further assume that this is driven in such a way that there is no loss of traction or slipping. In that case, I wouldn't expect the McCannots' rollers to spin at all.

I believe this is similar to what happens in a mecanum drive going straight forward. If you only look at a diagonal pair of wheels they contribute motion in a 45 degree angle. If you watch a robot doing that, you will see the rollers spinning. When you run all the wheels, the opposite diagonal set is pushing the wheels along the axis transverse to the rollers. The net effect should be (in a perfect setup) that the rollers don't spin because the opposite diagonal sets are countering the movement lost to the rollers.


My answer to this question, I believe that any roller angle between an Omni wheel and a Mecanum wheel can create this effect because the wheels on the opposite side have enough movement to counteract the losses accross the rollers. Once you go past that from Mecanum to McCannot you are losing more than half your potential movement to the rollers, if the rollers are frictionless. There isn't enough movement left to counteract the opposite diagonal's losses. Friction changes that though. With enough friction in the rollers, even the McCannot could have non-spinning rollers.

Andy

16-01-2015 15:14

lcoreyl

Re: pic: McCannot Wheel

My experience has been one revolution of the 8" AM mecanum wheels translates the robot 8*pi inches forward. Measured precisely to the last decimal point (math joke!).
Unfortunately, with no team this year I can't test this myself, but I have a theory that part of what you're seeing might actually be due to the fact the circumference of the wheels have the visual breaks in them. If you have some under-utilized newbies, maybe you could convince them to convert a regular wheel (or construct something like an 8" wheel), by drawing on the marks to make them look mecanumish and then roll them around and see if it is a visual effect...



Here are some items that I highly recommend anyone with access to a mecanum drivetrain try for themselves to help their understanding:

1) draw some lines on the sides of the rollers so that it is much easier to see if/when they are spinning.

2) mechanically lock all wheels/axles so they cannot spin (rollers should be left to spin freely). push your drivetrain around.

16-01-2015 15:38

lcoreyl

Re: pic: McCannot Wheel

What is slipping? the roller is losing traction with ground? If so, I don't understand how a rectangular wheel arrangement would cause this?



I am confused by "wheels have enough movement to counteract the losses (of movement?)", and "losing more than half your potential movement". Could you re-state this in a static case just talking about the forces?

16-01-2015 15:51

s_forbes

Re: pic: McCannot Wheel

Are you driving on shag carpet or on a plate of glass?

16-01-2015 18:03

Monochron

Re: pic: McCannot Wheel

Spoiler alert: the rollers don't spin when the wheel is driving forward or back unless they are losing traction. Same as any independently driven wheel. If they are spinning it means that you can make modifications to the drive train to increase its performance.

You can test this your self by slowly rolling then and marking the roller's position. Just make sure not to wobble or translate the wheel while you do it.

16-01-2015 20:35

Culvan Van Li

Re: pic: McCannot Wheel

Sorry, that's part of my kinematic model for mecanum. The slipping happens when rotating. It actually wouldn't effect the straight movement. Sorry about that. Someday I'll actually finish writing it up and upload it so people will know what I'm talking about.


Yes, I mean force. I believe half the force coming out of the motors translated through mecanum wheels is lost. However I was trying to avoid saying that because Ether is probably going to jump all over me. I won't get into the details here. I'm going to try to create an experiment that will help clarify what is happening and will continue to discuss it with Ether in private. It is possible Ether is right, but I will need more convincing than can be done without a practical trial.

Andy

16-01-2015 21:16

Foster

Re: pic: McCannot Wheel

Ether is right, he has the math behind him. (In front of him?) That's what engineering is about. Think of an idea, do the math, if the math says build it, then build it, otherwise save your dollars.

Think of all the cool things in science / architecture / building / etc. All went up because the math worked. Things went up and fell down because the math didn't work. Even the Roman and early cathedrals worked because some math was done.

If Ether says it will fail and you build it and it fails, you owe a "Math wins again post"

And let me tell you as a mentor when you come up with the "WE WILL CRUSH THE OTHER TEAMS WITH THIS IDEA" and a rookie roboteers says "Well according to the math you fall over here." it's a great day, another idea dies without metal being involved. Nobody likes to see metal fail in the wild.

But I'm new here, feel free to experiment.

16-01-2015 22:16

MrBasse

Re: pic: McCannot Wheel

Couldn't you just put some traction wheels of the same diameter on one side of a robot and your mecanum wheels on the other and drive straight? If it turns one way you'll know you have rollers causing loss of forward motion on that side.

For that matter, use your encoders to rotate your wheels a known number of revolutions. Calculate the distance you should travel based on circumference of your wheels and check that against the actual distance traveled.

I just feel like this is a great opportunity to go and try it to figure out the answer. Unless I'm missing something and neither of those would give the answer you are looking for.

17-01-2015 00:10

Monochron

Re: pic: McCannot Wheel

To my understanding, yes, a portion of the force exerted by the mecanum wheel is not used in moving the wheel directly forward or back. Because the wheel exerts force at a non-zero angle in reference to the forward/back direction, a component of the force is going sideways. However, the component of the force exerted in the forward/back direction is identical to the force exerted by a regular wheel in the same system.

My point being that yes, some force is "wasted" by the mecanum, but it is a sideways force The mecanum puts out the same force as a regular wheel in the same direction.

17-01-2015 14:15

Ether

Re: pic: McCannot Wheel

Probably not, if you drive all four wheels with the same torque. It's a bit difficult to analyze quantitatively because wheel slipping would be involved, but my guess is the vehicle is likely to veer to the left if the mecs are on the right side (and vice-versa).

18-01-2015 23:14

PAR_WIG1350

Re: pic: McCannot Wheel

What if a set mecanum wheels was built that used 2 axles per roller? Since the roller can't rotate about 2 axes simultaneously, there would be no rotation, but there would still be axial free play. Then you could build a set of wheels that is mostly identical, but which uses only one axle per roller to allow rotation. By comparing the 2 wheel sets in situations where the robot strictly moves forward or backward should tell us if there is significant rotation of the rollers in these cases.

19-01-2015 00:37

dtengineering

Re: pic: McCannot Wheel

I think you're on to something there... depending on the style of wheel it should be possible to non-destructively lock the rotation of the mecanum rollers.

Compare the performance of the robot with locked rollers to that with free-wheeling rollers. If there is a difference in performance, I'd suspect it would be due to the rollers... rolling.

Jason

19-01-2015 02:29

Chris is me

Re: pic: McCannot Wheel

If someone really wanted to find this out, grab your favorite 240 FPS camera (smartphone, gopro, etc), put a sharpie line on the edge of each roller somewhere, and drive. You'll know exactly how much each roller spins. Seems easier than building custom wheels, locking wheels, etc.

21-01-2015 14:25

Ether

Re: pic: McCannot Wheel

Do rollers spin when going forward?

http://vimeo.com/32952134


Is net force (along roller axis) greater than forward force?

http://vimeo.com/33315586

21-01-2015 14:34

Chris is me

Re: pic: McCannot Wheel

While these are helpful videos, in the case of the first video I'm not satisfied with a "test" consisting of someone rolling the wheel along the ground with their hands. They are in complete control of the wheel and could easily induce more axial loading than is present on a robot, etc. It's more likely than not that an on-robot test would yield the same results, I'm sure, but I just wanted to point that out.

I wonder how much of the second video discrepancy is caused by the brief moments in the rotation of the mecanum wheel where the traction wheel loses contact. I'm sure not a lot, but it's something to consider.

21-01-2015 19:51

Ether

Re: pic: McCannot Wheel

http://www.chiefdelphi.com/forums/sh...d.php?t=133309

22-01-2015 19:46

lcoreyl

Re: pic: McCannot Wheel

I totally agree with you (although maybe not calling the test a "test" ). I don't have access to this stuff currently, but I'm actually surprised someone hasn't done the sharpie marking with a closeup of a robot moving. Anyone with a good mec drivetrain, sharpie, and smartphone/gopro out there??????

yes, it would be better if the axle of both wheels were actually attached via some structural members, and the traction wheel width was as wide as necessary.
Anyone with some bored underutilized students, a spare mec wheel and extra wide traction wheel of the same radius out there??