14-05-2010 06:44

Chris is me

Re: pic: A mechanum wheel that shouldn't bump around at all.

Have you ran various mecanum wheels and encountered "bumping" problems? What improvement will eliminating this "bumping" give?

14-05-2010 10:42

JesseK

Re: pic: A mechanum wheel that shouldn't bump around at all.

There are various nominal benefits, such as more reliable traction and an accelerometer with much less noise. No single benefit stands out, but it's always fun to ponder new designs that address even the small issues in order to perhaps come up with a much better design.

This is good work for creating a nice round profile. However, the hardest part is in the mounting of the rollers; a single plate that forms the middle of a "T" may or may not be enough to support the cantilevered stresses from the rollers. 357 may be able to give some insight there. Additionally, the mounting of the rollers will inevitably create some sort of gap when mounted in the center -- i.e. each roller as shown is not possible. In reality each roller is more like two roller halves, split in the center where the mount goes.

Combining the two dominant Mecanum designs into a hybrid design may be a better solution. A hybrid design has 3 rollers with 2 thin mount plates; the two rollers on the outside look like rounded cones and are cantilevered whereas the roller in the middle looks more like a bulging omni wheel roller (or a squished donut).

14-05-2010 14:55

Nadav Zingerman

Re: pic: A mechanum wheel that shouldn't bump around at all.

Do those elliptical things actually roll (how?), or are they fixed in relation to the wheel?

14-05-2010 15:09

EricH

Re: pic: A mechanum wheel that shouldn't bump around at all.

They would roll. That's the way a mecanum wheel works--the rollers (the elliptical things) roll to alter the vectors the wheel is putting out. There would be an axle down the middle to allow them to roll; I don't see it here yet.

14-05-2010 15:36

Brandon Holley

Re: pic: A mechanum wheel that shouldn't bump around at all.

Like the segway mecanum wheel??



-Brando

14-05-2010 18:57

,4lex S.

Re: pic: A mechanum wheel that shouldn't bump around at all.

I have been thinking about how you did this, and I have to say, you have me stumped. How did you design the 3D wheel with that perfectly circular side profile?

15-05-2010 02:45

Hawiian Cadder

Re: pic: A mechanum wheel that shouldn't bump around at all.

in actuality the arm will go all the way out, and match the profile of the roller and be coated in delrin, that will slide and will make it perfect, the way i did it was to assume that each roller had to have an arc that would look exactly like a circle when viewed from 45 degrees, thus the line that defines the outer edge of that roller is an elipse with one dimension being 8, the diameter of the wheel, and the other dimension being 8/sin(45) or about 11.3, this makes the profile of the roller a perfect circle from a 45 degree angle. the rollers are just revolves with that line as the edge. the arms are .375 inch thick 7075 aluminum alloy, and according to solid works, this wheel should hold at least 200 LBS safely. the total weight of the wheel is 1.8 LBS, it is 2.6 inches wide from roller tip to roller tip, and is 8 in diameter, the axle hole will have have a hub or something to attach it to the motor. the method for manufacturing the rollers is to print off a mold on our 3d printer, then use urethane, coated in rubber. overall it should be pretty sweet.

15-05-2010 12:02

KRUNCH DUDE

Re: pic: A mechanum wheel that shouldn't bump around at all.

These wheels remind me of Airtrax wheels, did you get the idea from them? I would like to see a nice set of mechanum wheels for FIRST.

15-05-2010 23:09

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

Hi Brandon,

Could you provide a link that shows the rest of this picture? I am curious to see the rest of the platform.


~

15-05-2010 23:28

Alan Anderson

Re: pic: A mechanum wheel that shouldn't bump around at all.

That's the Segway RMP 400 Omni.

15-05-2010 23:36

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

Thanks Alan.

Looks like the wheels in the picture Brandon linked to are mounted incorrectly.


~

18-05-2010 17:52

exprg:melonhead

Re: pic: A mechanum wheel that shouldn't bump around at all.

hmmm... this wheel sure does look pretty, but does it run that way too? the side wheels (or whatever the technical term for them may be) look as if they are at a very harsh angle. now, i'm just a programmer, but isn't the ideal angle 45 degrees?

18-05-2010 18:00

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

There is no "ideal" angle.

Many mecanum wheels have rollers at 45 degrees. This makes the kinematic (and dynamic) analysis easier. The forward and inverse kinematic matrix transformations are simple, with no trig functions or square roots.

Making the angle smaller than 45 degrees (angle between roller axis and plane of the wheel) improves forward/reverse traction but reduces strafing traction, and vice versa.

So there is no one ideal angle. It depends on the application.


~

18-05-2010 18:02

Rion Atkinson

Re: pic: A mechanum wheel that shouldn't bump around at all.

You can buy some from AndyMark.

18-05-2010 18:19

Hawiian Cadder

Re: pic: A mechanum wheel that shouldn't bump around at all.

true, the method that i used to make these could be aplied to any angle of mechanum wheel. when we build them we may use closer to a 35 degree angle,when strafing speed is nessisary, and they will climb better.

19-05-2010 08:45

Ryan Dognaux

Re: pic: A mechanum wheel that shouldn't bump around at all.

Do a search for Jester Drive. 357 has been creating mecanum wheels since 2005 (actually this thread right here has some good discussion in it - http://www.chiefdelphi.com/media/photos/20664). Gives some information on the development of our wheels and our relation to Airtrax.

I've attached a few images from this year's CAD renders, feel free to check them out.

19-05-2010 10:14

exprg:melonhead

Re: pic: A mechanum wheel that shouldn't bump around at all.

so what you are saying is that in order to to drive easily both forward and back, it would be best to have a good 45 degree?

19-05-2010 10:26

Rion Atkinson

Re: pic: A mechanum wheel that shouldn't bump around at all.

What program do you use to render those?

19-05-2010 11:27

Ryan Dognaux

Re: pic: A mechanum wheel that shouldn't bump around at all.

That was done in Inventor 2010 by one of our students, Kyle Tress.

19-05-2010 11:29

Andrew Schreiber

Re: pic: A mechanum wheel that shouldn't bump around at all.

Ryan, do you machine your hubs from solid? I assume you cast your rollers?

19-05-2010 11:34

Rion Atkinson

Re: pic: A mechanum wheel that shouldn't bump around at all.

Interesting... I've never seen Inventor render anything that pretty... Looks like I'll have to play with 2011 some more.

19-05-2010 11:42

Joe Ross

Re: pic: A mechanum wheel that shouldn't bump around at all.

There's a nice description (with lots of pictures) of their 2006 process in FIRST Robots: Aim High: Behind the Design

19-05-2010 11:42

Ryan Dognaux

Re: pic: A mechanum wheel that shouldn't bump around at all.

The hex hubs are machined from hex stock and the outside fins that hold each roller are bolted onto the outside of the hub. This way, if we somehow bend a fin during a match (the bumps liked to do that this year), we can swap them out between matches with spares and you don't lose an entire wheel.

The rollers were cast with certain materials this year from silicone molds. The rollers themselves have two layers, a hard inner core and a softer outer traction layer. I'll see if I can get some of the students who made the rollers this year to post more details.

19-05-2010 16:23

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

No, that's not what I said (not even close). I'm not sure how you concluded that from what I wrote. Help me understand what part led you astray.


~

19-05-2010 18:56

Hawiian Cadder

Re: pic: A mechanum wheel that shouldn't bump around at all.

nice, that is the idea with these too, we should be able to replace the arms, the main hurdle now is making the rollers so that they will fit perfectly with the bearings, can you turn urethane on a lathe?

19-05-2010 18:59

Brandon Holley

Re: pic: A mechanum wheel that shouldn't bump around at all.

You sure can.

-Brando

19-05-2010 19:50

EricH

Re: pic: A mechanum wheel that shouldn't bump around at all.

I can understand why he said that... But to ONLY go forwards and backwards, it's best to have a traction wheel. (literally, he said forwards and backwards only...)

What Ether said, in essence, was that many mecanum wheels used 45* rollers, but that other angles could be used--less than 45* puts the advantage to front/back and greater than 45* puts the advantage side-side in terms of traction. I assume that this is a correct summary. If it is not, then the original post needs to be modified, as that is where I got this from.

From that, it is easily (and logically) concluded that, for the optimal front/back and side/side traction, 45* is a better angle than many others. If that is not the conclusion that was intended, then better words need to be chosen.

19-05-2010 21:40

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

OK so you understood. Good for you.


Justin, what part of my post led you astray? Was it the part about the forward and inverse matrix transformations? I can explain in more detail if you are interested.


~

19-05-2010 21:48

Rion Atkinson

Re: pic: A mechanum wheel that shouldn't bump around at all.

Honestly I would just tell him to read Eric's post. It made plenty of sense. (Now yes, yours made sense to. But there is no need to re-explain that which has been explained if it is now understood.)

19-05-2010 21:54

LLogan

Re: pic: A mechanum wheel that shouldn't bump around at all.

A forty five degree angle makes the vectors of the straight and perpendicular motions equal (this should be obvious if you know basic trigonometry). That is why it is chosen by many teams as the angle for their rollers. It theoretically should make your strafing and forward motions equal. However... I do not think that's the case. There is just too much inefficiency in a mecanum drive system.

Changing the angles will change the vectors of the wheels. A smaller angle (assuming the maximum angle, 90 degrees, is just a roller mounted straight lengthwise) should theoretically give more strafing motion. Vice versa for a larger angle and forward/back motion.

19-05-2010 22:16

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

The kinematic matrix transformations were mentioned but not explained. You apparently know all about them but perhaps Justin does not and would be interested in learning.

Justin, if you are interested let me know and I'll start another thread.


~

19-05-2010 22:29

Hawiian Cadder

Re: pic: A mechanum wheel that shouldn't bump around at all.

no, if the rollers are at 45 degrees, then the strafe speed is half of what the forward reverse speed is, because going forward, the wheels act like normal wheels, however when they are strafing, they are 2 vectors at 45 degree angles, so only half of the magnitude of the vector is in the direction of the strafe.

19-05-2010 22:42

LLogan

Re: pic: A mechanum wheel that shouldn't bump around at all.

I do not think so. The resultant of the vectors is in the direction that they roll. There is no way that the magnitude of the vectors would be the forward/reverse vector because then the vectors that composed it would have to be at 45 degrees which would mean strafing motion would be impossible.

This is what the vectors of a mecanum wheel look like.

http://lib.store.yahoo.net/lib/yhst-...lSpecSheet.pdf

19-05-2010 22:45

Rion Atkinson

Re: pic: A mechanum wheel that shouldn't bump around at all.

I knew something was sounding off... Thank you for clearing that up. Definitely will keep this in mind.

19-05-2010 23:59

Hawiian Cadder

Re: pic: A mechanum wheel that shouldn't bump around at all.

hmmmm, i know that it is less speedy/ powerful strafing than side to side though.

20-05-2010 00:48

AdamHeard

Re: pic: A mechanum wheel that shouldn't bump around at all.

Can you please post your team number and location? You post a lot in a somewhat negative tone, which is easy to do when you have no fear of anyone associating you with a team.

20-05-2010 00:59

EricH

Re: pic: A mechanum wheel that shouldn't bump around at all.

Actually, there is a significant observable slowdown when side-to-side is run as compared to front and back. Part of this is due to friction and other inefficiencies (remember, two of the 4 wheels are running backwards); part of it is due to the vectors. I'm not exactly quite sure how that happens off the top of my head; I think Alan Anderson posted an explanation a while back.

Ether, I posted the part I think led Justin "astray"; namely the logical conclusion that if X< than Y, Z is more effective one way and X > Y, Z is more effective another way, therefore if X=Y, Z is optimized for both directions, and that he simply misstated what he meant. It's up to him to correct me if I'm wrong on that. (Also note that I'm in college, and I don't know/don't remember how to do a matrix transformation. I wouldn't be surprised if that reference confused a number of people. Some explanation may be in order, either in here or in another thread, as to a) what they are, b) how to do them, and c) why they apply here.)

20-05-2010 03:54

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

A request was made for a more detailed explanation of my earlier post concerning kinematic analysis of a mecanum wheeled-robot, and the derivation of the inverse and forward transformation matrices.

I decided to start a new thread for this discussion in order not to divert the focus away from this thread's OP's intent.

http://www.chiefdelphi.com/forums/sh...51&postcount=1


The analysis answers many of the questions posted here about the effect of roller angle and whether fwd/rev vs strafe should be the same speed etc.



~

20-05-2010 09:18

Joe Ross

Re: pic: A mechanum wheel that shouldn't bump around at all.

He's from 2474 in Michigan.

20-05-2010 15:41

LLogan

Re: pic: A mechanum wheel that shouldn't bump around at all.

Yes, I agree with you that there is a significant difference in the two motions. However, the bolded portion of your text is not entirely relevant to the inefficiencies of a mecanum system. Simple vector analysis shows that the strafing motion is theoretically equal to the forwards and backwards motions. The fact that two wheels are running backwards doesn't matter at all (unless you're talking about motor bias, but I don't think you are.). The forward/backwards vectors all theoretically cancel out and leave only the perpendicular vectors, which are, in magnitude, equal to the forward/backward vectors.

20-05-2010 15:59

EricH

Re: pic: A mechanum wheel that shouldn't bump around at all.

When running front-back, all 4 wheels are contributing "full power" in one direction. When running side-side, all 4 are contributing "full power" in another direction, but two of them have to work against the other two to do it. This will amplify any effects of friction and other inefficiencies that were masked by running all together.

The problem with theory is that it often leaves out reality. This is why engineers use safety factors and try to account for reality wherever they can. Theoretically, there is no friction and no inefficiency, and inefficiency (where it does exist) is uniform any way you look at it. In reality? Not only is there friction and inefficiency, but it's hard to give them a nice, easy number.

20-05-2010 16:19

Alan Anderson

Re: pic: A mechanum wheel that shouldn't bump around at all.

The left and right wheels are working "against" each other when running front-back, just as much as the front and back wheels are working "against" each other when running side-side.

The issue making the two directions act differently in the real world is the friction of the rollers. It helps in the forward direction, and hurts in the sideways direction. If you use sensors to maintain the same wheel speed, your front-back speed will be the same as your left-right speed. With non-frictionless rollers, though, you have to apply more power to maintain that speed when traveling sideways.

20-05-2010 18:14

LLogan

Re: pic: A mechanum wheel that shouldn't bump around at all.

I think there has been a misunderstanding. I have never said that the front/back and strafing motions were equal realistically. I actually said the opposite, that the strafing motion is much less efficient than the forwards/backwards motion.

In the backwards/forwards motion, the wheels are still acting against each other. Simple vector analysis tells us that perpendicular components of each force vector cancel each other out. Vector analysis also tells us that the same exact thing happens in the strafing motion, however, this time, the backwards/forwards vectors are the ones that cancel each other out. The two motions have the exact same oppositions, just in different directions.

The only thing that I have been trying to prove is that each component of the force vectors on the wheels are theoretically exactly equal in magnitude. I have only been trying to prove that beliefs like this:

are not entirely correct.

Believe me, I understand the fine line between reality theory/inefficiencies. However, I'm just trying to say that the inefficiencies aren't caused by things you described. The only factor at work is the friction of the rollers. If you tighten each roller so that it no longer rolls, you will essentially have four "normal" traction, but you will no longer be able to strafe. This is because the omnidirectionality of mecanum wheels is dependent on the rollers "slipping". However, because there will never be a frictionless roller, the strafing motion will never be equal in force to the forward/backwards motion.

20-05-2010 18:18

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

In the real world of aerospace engineering, systems engineers try to model friction and inefficiency as closely as possible when they are building dynamic models to allow them to build accurate procurement specs for suppliers. So theory does include friction and inefficiency.

Roller bearing friction plays a large and asymmetric role for mecanum wheels. In the fore/aft direction, roller bearing friction is a good thing. It moves the reaction force of the floor closer to the plane of the wheel, so it takes less reaction force for a given forward force. The result is better traction. For a frictionless roller bearing, the reaction force is aligned with the roller axis. As roller bearing friction increases from zero to locked roller, the reaction force moves from being aligned with the roller axis to lying in the plane of the wheel. When the reaction force lies in the plane of the wheel, that's essentially the same as a standard wheel. So, the reason a mecanum wheel doesn't have as much pushing force (in the fore/aft direction) as a standard wheel is not because there's less forward force available, but because the reaction force is larger than it would be for a standard wheel, and therefore the mecanum wheel starts to slip before a standard wheel would.

In the sideways direction, roller bearing friction moves the reaction force toward the plane of the wheel, which reduces the force vector component in the sideways direction. To compensate, the motor must output more torque. This increases the reaction force, which reduces available traction. So in the sideways direction, roller bearing friction is a bad thing. It causes increased motor power consumption (for a given desired force) and it reduces available traction.

~

20-05-2010 23:15

exprg:melonhead

Re: pic: A mechanum wheel that shouldn't bump around at all.

ok. here's what confuses me. if i want a robot that drives just as easily sideways as it does front and back, would it be best to have a 45. because the way i see it, the more you get away from that angle, the more energy it will take to move it sideways. correct me if i'm wrong because i feel like i'm missing a huge part of the puzzle here. but i think i understand the most part. it's basic physics really.

sorry for the mis-wording on my last post. i was pressed for time. i meant to say "easily move forwards and sideways". it's not that i don't understand it, because i do. i just need it explained how it would be advantageous to set it at any other angle.

20-05-2010 23:22

Ether

Re: pic: A mechanum wheel that shouldn't bump around at all.

Yes, that's probably a pretty good rule of thumb. You're on the right track. The important thing is to try to make the roller bearing friction as low as possible. For "zero" roller bearing friction, 45 degrees will make fore/aft and sideways motions equally easy. But since there is always some roller bearing friction, you might need to increase the roller angle slightly (make the roller axis closer to the wheel axis) if you want to make both motions equally easy.

You've got the right idea Justin. The more you get away from 45 degrees in one direction (i.e. smaller angle - roller axis becoming parallel to the plane of the wheel), the less efficient the sideways motion becomes. Moving the roller axis the other direction makes sideways motion more efficient - up to a point, depending on the roller bearing friction.

Take a careful look at posts 39, 41, 43, 44, and 45 in this thread - there's lots of good detailed explanation there. Ask questions about anything that's unclear.


~

20-05-2010 23:31

exprg:melonhead

Re: pic: A mechanum wheel that shouldn't bump around at all.

thank you so much for helping explain this. i know it's hard to explain this through writing, and i know i don't usually learn well this way. you helped so much on clearing this up for me. i appreciate it.