View Full Version : Mecanum Advice Request
We are considering having mecanum this year and were wondering what the subtleties of it are. Anything and everything would be helpful such as what gear boxes to avoid, what ratios would be best or various problems with programing.
We are considering having mecanum this year and were wondering what the subtleties of it are. Anything and everything would be helpful such as what gear boxes to avoid, what ratios would be best or various problems with programing.
This is our (461) second year in a row that we are using mecanum; i don't know a whole lot about it, but i do know that:
Much better maneuverability
Less pushing force
Harder to program/longer
Problems can occur with a direct drive setup and thoughbox nanos, because it pushes the bearings with a force and direction they're not designed to take.
But all in all, if you're programmers and drivers are up for it, it is a great choice
TubaMorg
16-01-2011, 16:16
You ask a broad question with many details that is difficult to answer succintly. Specific questions are better in the forum format. However, there have been quite a few threads that address ALL of your sub-questions in some form or another. The technology hasn't changed since those discussions were posted. Also, there are a couple of white papers that discuss the physics behind mecanum drives.
Problems can occur with a direct drive setup and thoughbox nanos, because it pushes the bearings with a force and direction they're not designed to take.
We were considering using the nanos. Have you tried using them? What went wrong if you did?
Team 2950 used a 4 wheel drive combination of two Mecanums in the front and two Plactions in the back for a mix of great maneuverability AND traction in Breakaway last year. It worked out very well for us and could be good for your team as well.
Chris is me
16-01-2011, 16:23
I've heard good things about direct drive + Toughbox Nanos in mecanum drives, as well as regular Toughboxes.
We were considering using the nanos. Have you tried using them? What went wrong if you did?
Well, we haven't tested them out yet, but the mentors noticed that the bearings might not be able to take the 45 degree angle load from the direct mecanum drive, at least for the nanos. we aren't 100% sure yet, but it's definitely something to look into
Dracon Phinex
16-01-2011, 16:45
If you want an agile bot, go ahead. The team I'm with wanted to do mecanum wheels, but decided not to because of programing issues tied up in it. The ideas are great, but are not very good for newer teams. The frame might have to modified in order to do this.:cool:
Last year 1279 used 8" mecanums, direct drive to long output shaft nanos, 12.5:1. We used a bearing on the other side of the long shaft.
My suggestion is get it done ASAP, so the driver(s) have time to train, and programers have time to program it. We actually built a test bot using a second set of Mecanums, using a 3/4" plywood base fot driver training and programming.
Last year 1279 used 8" mecanums, direct drive to long output shaft nanos, 12.5:1. We used a bearing on the other side of the long shaft.
My suggestion is get it done ASAP, so the driver(s) have time to train, and programers have time to program it. We actually built a test bot using a second set of Mecanums, using a 3/4" plywood base fot driver training and programming.
We are planning to do the same thing except with a legit second drivetrain.
Our plan is to use the 6 inch mecanum. Is the 8.45:1 ratio to slow? I know that mecanum is inefficient and will lose some of that speed but how much?
Chris is me
16-01-2011, 19:19
Our plan is to use the 6 inch mecanum. Is the 8.45:1 ratio to slow? I know that mecanum is inefficient and will lose some of that speed but how much?
Nope. For reference, 12.75:1 with 6" wheels is ~9 feet per second.
If you have an algorithm that translates in every direction the same, you will get about 71% ( sqrt(2)) of full power from your motors when going forward with no other modifications. 1675 usually includes a "beast mode" for going stright forward/back with 100% power.
Going on the topic of bearings, and this might sound like a stupid question, are spherical bearings ever used (as opposed to the traditional ball bearings)? and if so, what are the pros and cons respectively?
thanks
-duke
If you have an algorithm that translates in every direction the same, you will get about 71% ( sqrt(2)) of full power from your motors when going forward with no other modifications. 1675 usually includes a "beast mode" for going stright forward/back with 100% power.
Can you further explain this? Why would you only get 71% when going forward?
Creator Mat
16-01-2011, 20:19
If you are direct driving a mecanum drive or any drive for that manner, make sure you support the shaft in two places. If you don't you'll bend your shaft and destroy your drive train. The tough box series from andy mark does have two support points internally and you can get away with no external supports on the shaft for short distances. However if you use an extended shaft you'll just kill the gear box and the shaft if there is no support on the other end of the shaft. My team used mechanum last year and used the P80 gear boxes (http://banebots.com/c/P80K-S4) with an extended shaft. It was extremely efficient, had good mount points for the encoder, and we had no problems. Hope this helps
Eigenvector
16-01-2011, 20:49
We (1002) ran 8" mecanums last year off standard Toughboxes. They're surprisingly quick and we were very pleased with how well they worked. I'd recommend avoiding chain if you can, the added strains from the different types of motion didn't agree with it. Also, once you've got them assembled, check that all the rollers spin freely.
Good luck!
pfreivald
16-01-2011, 20:51
If I may ask, where did you mount your encoders on the p80?
"beast mode"
love that
However if you use an extended shaft you'll just kill the gear box and the shaft if there is no support on the other end of the shaft. My team used mechanum last year and used the P80 gear boxes (http://banebots.com/c/P80K-S4) with an extended shaft. It was extremely efficient, had good mount points for the encoder, and we had no problems. Hope this helps
I don't see why the longer shaft would destroy the gearbox unless using that long shaft meant mounting your wheel far away from the gearbox which creates problems. Also wouldnt a third mounting point create more problems because a third mounting point gives the shaft a place to bend.
Can you further explain this? Why would you only get 71% when going forward?
Not sure how much you know overall about Mecanum so I'll lay it all out.
Mecanum wheels produce a force at a 45% angle when driven. The x and y components respectively equate to roughly 71~% of the total vector. (Actually thinking about it my sqrt(2) comment makes no sense. it's actually 1/sqrt(2) for each side. 1/1.414 ~= .71)
In my head, without looking at our algorithm it isn't making a ton of sense (since the forward vectors would be 71% too), but I know that when we're running "normally", which is translating at the same speed in every direction, our motors run at only that 71% when going forwards.
EDIT2: It is because the motor outputs in our algorithm are limited by a trigonometric function and motor offsets. It just so happens that our motor outputs when going forwards or backwards are +/- sqrt(2)/2, which is the same ~71%.
We give the drivers the option to drive our motors at 100% forward/backward.
If you have an algorithm that translates in every direction the same, you will get about 71% ( sqrt(2)) of full power from your motors when going forward with no other modifications. 1675 usually includes a "beast mode" for going stright forward/back with 100% power.
Can you further explain this? Why would you only get 71% when going forward?
If you use the provided library functions (such as LabVIEW Holonomic vi), you don't have to worry about this and there is no need for a "beast mode", because the normalization is performed on the wheel speeds after the inverse kinematic calculations are performed, not on the joystick commands. You get the full range of motor command to the wheels in all directions.
Now, don't confuse this with the pushing force of a mecanum, which is a separate issue. In a traction-limited situation, mecanum has less pushing force than a standard-wheel vehicle using the same tread material, because the floor reaction force (for a given forward force) is higher on a mecanum wheel so it loses traction more easily.
Trevor1743
16-01-2011, 22:14
We give the drivers the option to drive our motors at 100% forward/backward.
When you say 100%, do you mean the equivalent of ~71% of regular wheels or do you mean 100% of regular wheels?
When you say 100%, do you mean the equivalent of ~71% of regular wheels or do you mean 100% of regular wheels?
Still 71% force of regular wheels, 100% motor output
Trevor1743
16-01-2011, 22:37
Is it possible to move in any direction with Mecanums, like you can with omni wheels?
Still 71% force of regular wheels, 100% motor output
The forward force Ff provided by a mecanum wheel is the same as the forward force that would be produced by a “standard” wheel of the same diameter being driven with the same torque, but in the case of the mecanum wheel the reaction force of the floor on the wheel (roller) is greater by a factor sqrt(2), because this reaction force is at 45 degrees to the forward direction.
It is this greater reaction force which causes the mecanum wheel to break traction with the floor and slip at lower forward-force levels than a standard wheel. This explains why a mecanum vehicle has less traction than a standard-wheel vehicle, even if they both use the same wheel material.
http://www.chiefdelphi.com/media/papers/download/2739
Is it possible to move in any direction with Mecanums, like you can with omni wheels?
Yes.
Is it possible to move in any direction with Mecanums, like you can with omni wheels?
Yes, both omni and mecanum are holonomic. They can simultaneously execute all 3 degrees of freedom: fore/aft, strafe, & rotate.
In what different ways can you control mecanum? The ideas I know of are:
using the joystick that twists.
have a tankdriive setup with one stick that controls strafing.
one stick controls your movement in every direction and the other spins the bot to the direction you point it.
pfreivald
17-01-2011, 10:41
In what different ways can you control mecanum? The ideas I know of are:
using the joystick that twists.
have a tankdriive setup with one stick that controls strafing.
one stick controls your movement in every direction and the other spins the bot to the direction you point it.
The joy of a holonomic drive system is that it handles exactly like a first person shooter... So if you set up your controls to mimic a FPS, most drivers will already have an intuitive feel for the setup.
...that's what we did last year, anyway, and it worked great!
Ben27Lacrosse
17-01-2011, 11:12
Okay, Here are a few things to look out for and suggestions that my team (1296) uses/does.
First of we choose our gearbox, toughbox nano with the long key. The reason we do this is for extra support. We use mounting plates one both sides of the wheel, for extra support.
Secondly, throw together the KOP base and put mecanum wheels on there to give your progamming team time to flush out and software problems.
Finally, when assembling your final bot be sure to distribute the weight equally do that your wheels function properly.
More details of what BigJ is talking about can be found in this paper (http://www.chiefdelphi.com/media/papers/2434) about the control strategy used by FRC Team 1675. Basically the output is limited based on the inverse kinematic calculations to allow for maximum speed output of the robot to be the same in all directions with overhead for the yaw rate to be added allowing the same yaw rate control regardless of the heading. These calculations result in the reduction of total overall maximum speed (not using the full range when traveling in the forward/reverse direction) in favor of overall normalization of speed and yaw rate. "Beast Mode" basically allows the driver to override this control decision.
The considerations we have mistaken before include accidentally mounting the mecanum wheels in X and not O configuration and balancing and maintaining the weight/downward force on each of the wheels.
The typical 1675 mecanum setup looks like this:
Four CIM motors
Four BaneBot P80 gearboxes (http://banebots.com/c/P80K-S4)
Four AndyMark 8" Mecanum Wheels (http://www.andymark.com/ProductDetails.asp?ProductCode=am-0083)
We have also experienced issues with the AndyMark 6" Mecanum wheels (http://www.andymark.com/ProductDetails.asp?ProductCode=am-0137) (GEN 1 and GEN 2) in that the hubs did not hold up to the impulse loads incurred by 120lb robots in competition. The new AndyMark 6" Mecanum HD wheels (http://www.andymark.com/ProductDetails.asp?ProductCode=am-0732) should be more than adequate based on the testing document (http://d1pytrrjwm20z9.cloudfront.net/6inch-HD-Mecanum-evaluation.pdf) they have published.
We have used two different encoder mounting strategies as sometimes we have used chain and sprockets off of the P80s.
Vex Quadrature Encoder on an idler sprocket
Vex Quadrature Encoder (Modified output shaft) or USDigital Encoder
Considered but did not use the austriamicrosystems magnetic rotary encoder which just requires affixing a magnet to the end of the shaft
More details of what BigJ is talking about can be found in this paper (http://www.chiefdelphi.com/media/papers/2434) about the control strategy used by FRC Team 1675. Basically the output is limited based on the inverse kinematic calculations to allow for maximum speed output of the robot to be the same in all directions with overhead for the yaw rate to be added allowing the same yaw rate control regardless of the heading. These calculations result in the reduction of total overall maximum speed (not using the full range when traveling in the forward/reverse direction) in favor of overall normalization of speed and yaw rate. "Beast Mode" basically allows the driver to override this control decision.
As Ether noted this year the math in the provided holonomic drive VI's corrects for this already and no optimizations are required to achieve optimal drive output, though this was an issue last year.
The considerations we have mistaken before include accidentally mounting the mecanum wheels in X and not O configuration and balancing and maintaining the weight/downward force on each of the wheels.
This is something that we had trouble with the first time we used holonomic wheels also. Be careful that viewed from below your robot the rollers on your wheels form an O and not an X.
Let me just try to clear this up for me and others...the rollers should form an "X" from top-view and an "O" when looking at it from below?
pfreivald
17-01-2011, 14:49
No, the rollers should form an O either way.
No, the rollers should form an O either way.
I'm not looking at a holonomic robot right now, but I'm pretty sure that the rollers should form an O from below and an X from above.
As described here:
http://www.chiefdelphi.com/forums/showthread.php?t=80962
EDIT: With mecanum wheels at least, mounted parallel with the drive chassis. Omniwheels are another story.
GearsOfFury
24-01-2011, 22:31
If I may ask, where did you mount your encoders on the p80?
Peter - Last year, BaneBots sold a kit, EN-G0561-KT, that consisted of an optical digital encoder that could be mounted directly (glued) to the output shaft of the gearbox, with a neat plastic cover and mounting kit. It doesn't appear that they offer these any longer on their website, though. I believe they were US Digital encoders, though, so the BaneBots folks may still be able to help you locate some sets.
pfreivald
24-01-2011, 22:55
Peter - Last year, BaneBots sold a kit, EN-G0561-KT, that consisted of an optical digital encoder that could be mounted directly (glued) to the output shaft of the gearbox, with a neat plastic cover and mounting kit. It doesn't appear that they offer these any longer on their website, though. I believe they were US Digital encoders, though, so the BaneBots folks may still be able to help you locate some sets.
Thanks for that... Given BB's inability to ship anything even vaguely on time (including non-P60 stuff that wasn't notified as having any kind of delay), I think we'll figure out something for ourselves!
I'm not looking at a holonomic robot right now, but I'm pretty sure that the rollers should form an O from below and an X from above.
I am looking at one, and you are correct.
Bird eye view = X , Worm eye view= O
vBulletin® v3.6.4, Copyright ©2000-2017, Jelsoft Enterprises Ltd.