Mecanum Drive Problem..

[May_Walter]

Hello.
We wrote a mecanum drive code of our own. The outputs have been checked and they're correct, but for some unknown reason sometimes the robot changes directions while we are driving (mostly while we are driving on the side "mecanum drive").
In addition, certain wheels spin slower than others, and when we spin in place the robot jumps instead of driving smoothly.
Are these problems common? do you have any ideas how to solve them?
Thanks
Team Artemis#3083

[Joe Johnson]

Hello.
We wrote a mecanum drive code of our own. The outputs have been checked and they're correct, but for some unknown reason sometimes the robot changes directions while we are driving (mostly while we are driving on the side "mecanum drive").
In addition, certain wheels spin slower than others, and when we spin in place the robot jumps instead of driving smoothly.
Are these problems common? do you have any ideas how to solve them?
Thanks
Team Artemis#3083

Seems almost like an intermittent hard spot in the drive system that is changing the speed of one or more wheels.

Try putting the robot up on blocks and running the wheels. Listen for changes in speed. It is possible that it is some sort of phasing thing between the drive sprocket and the wheel sprocket (e.g. every (N-M)/(N+M) turns the eccentricities line up in such a way that the chain gets tight or too loose and jumps or whatever...)

Just some suggestions.

Joe J.

[Rob]

Do the rollers in your Mechanum wheels turn smoothly? If one or more rollers are "sticky" you could see some odd driving behavior.

Wheels spinning slower than others could be caused by a host of problems. Mechanical binding, faulty sensor feedback, program outputting for a different state than you think...

Good luck, this does not sound like a trivial issue to troubleshoot.

[May_Walter]

Hi again
1. we already tried checking the wheels on a jack, and we are sure that the program's outputs are correct, besides - the jaguars lights change properly.
2. we also tried fixing the rollers before.. they were too tight, but that problem was fixed about two weeks ago, but the problems persist.
We really tried everything we could think of. Just don't know what's wrong.

[May_Walter]

OK, a small update.. the side turning was a minor bug in the code that didn't give the correct orders to the engines.. so we solved that one..
now it's only the turning in place...

[Joe Johnson]

OK, a small update.. the side turning was a minor bug in the code that didn't give the correct orders to the engines.. so we solved that one..
now it's only the turning in place...

Since you are able to spin at all, I assume you have the X vs. O thing figured out, but then the down side comes into play, the correct orientation of the Mecanum wheels means that you are going to be scrubbing wheels (i.e. moving them in the direction of the axles rather than rolling them) when you spin in place. If your little wheels are grippy then they are not going to like that much.

The question is how much are you jumping? If you are just seeing "normal" 4WD with wheels in the corners jumping then there isn't much to talk about. If it is something more serious then perhaps there are things to do to improve things.

Can you provide more details? How about a movie of the behavior?

Joe J.

[Alan Anderson]

...when we spin in place the robot jumps instead of driving smoothly.

That's a strong clue that you have your wheels swapped. A robot with mecanum wheels in a perfect square wouldn't be able to spin at all. With a more typical rectangular configuration it will manage to spin, but with obvious reluctance. With the robot turned off, can you push it by hand and make it spin easily? If so, that confirms the problem as swapped wheels.

The pattern of rollers touching the floor should be in an "O" or "diamond" configuration. They will look like an "X" from above.

[May_Walter]

The pattern of rollers touching the floor should be in an "O" or "diamond" configuration. They will look like an "X" from above.

From below, our rollers look like:
\ . /
\ . /
\ . /


/ . \
/ . \
/ . \

(note that the lines are along the rollers spin axis, not their rolling direction)
Is this the correct way?

[Jared Russell]

That is incorrect - you want the opposite arrangement.

[May_Walter]

Are you sure?
Any way, can this be the reason for the jumps?

[Jared Russell]

If I understand your diagram (and I think I do), then yes, I am sure.

Here's a picture of what it should look like:

http://robotics.ee.uwa.edu.au/eyebot/doc/robots/omni2-diag.jpg

[Cl4p-tp]

The jumps might be cause by un-even chain tension. I myself had to fix that problem...

[May_Walter]

We dont have chains on our robot, the connection uses sporckets only.

[Hadi379]

This is what the rollers should look like from an overhead view:

\ /

/ \

This is what the rollers should look like from an underneath view:

/ \

\ /