Chief Delphi

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)
-   Technical Discussion (http://www.chiefdelphi.com/forums/forumdisplay.php?f=22)
-   -   Mecanum vs Treads (http://www.chiefdelphi.com/forums/showthread.php?t=110793)

MamaSpoldi 09-01-2013 10:16

Re: Mecanum vs Treads
 
Quote:

Originally Posted by Wing (Post 1211325)
It has to deal with weight. Meccanum drive relies on an even weight distribution on each wheel for accurate strafing and turning. if you don't have that, the robot will strafe in a slightly offset angle. Introduce a different height for 2 wheels, and suddenly your back wheels have more weight than your front. It only affects strafing though, if I remember correctly.

In our experience, a properly implemented gyro-stabilization in the control loop for the mecanums make them very controllable and maneuverable. In addition you can certainly take advantage of the protection of touching the pyramid for shooting without fear of being pushed/bumped.

Peter Matteson 09-01-2013 10:22

Re: Mecanum vs Treads
 
Quote:

Originally Posted by Taylor (Post 1211324)
If I remember correctly, 100% of 9-wheeled teams won Einstein. So you should do that.

Just because something has never been done, doesn't mean it's impossible. See: everything ever.

I believe it's 100% of 9 sided robots, it was 3 wheeled.

No 9-wheeled robot has made it.

Also I believe 3 wheeled robots are 2 for 3.
67 -2005 Champion
16 - 2008 Finalist
148 -2008 Champion

buchanan 09-01-2013 10:44

Re: Mecanum vs Treads
 
Mecanum drivetrains, like every other, have advantages and disadvantages, but some of the disadvantages being discussed here are overstated and/or misinformation.

Even weight distribution is optimal, but not an absolute requirement for straight driving. So long as each wheel has sufficient loading for sufficient traction, it will go the way it's supposed to. The problem only arises when you're at the limit, trying to get more acceleration/deceleration/pushing force than the wheels can transmit. This is true for every drivetrain, though holonomic ones are generally to be more sensitive to it and have more interesting behavior when it happens. The better the weight distribution, the higher the limit, but until there's slippage behavior stays correct.

Concerns about non-flat surfaces tend to be exaggerated for two reasons. One, it's not quite true that loss of traction on one (or even two) wheels guarantees incorrect behavior. The effect of traction loss varies between nothing and complete failure to move, depending on which wheels lose traction and what motion is being attempted; in practice it's usually something in between. Two, this criticism assumes an infinitely rigid chassis. Even if you could make such a thing, it's a basic of holonomic drivetrain design with >3 wheels that you don't try. In practice, even without intentional flexibility/suspension, a basic kitbot type frame has all the flex necessary to keep things like tower base plates from being problems.

Anupam Goli 09-01-2013 11:35

Re: Mecanum vs Treads
 
Quote:

Originally Posted by buchanan (Post 1211392)
Mecanum drivetrains, like every other, have advantages and disadvantages, but some of the disadvantages being discussed here are overstated and/or misinformation.

I'm not sure if experience is misinformation though. In 2012, my old team (1002) used meccanum wheels, and the center of gravity was somewhere close to the back. This affected driving in more than one way. The robot strafed at offset angles, and sometimes, especially on the key, the robot wouldn't even go straight properly! It was not at all an optimal experience. I'm not trying to say don't do meccanums, I'm just trying to say that you should keep weight distribution in mind.

buchanan 09-01-2013 12:18

Re: Mecanum vs Treads
 
Experience is important, and often trumps theory. But it is possible to go wrong tracing back from practical experience to theoretical root causes.

There are challenges in getting a holonomic drive to go straight, but they probably have more to do with how computed wheel speeds map to actual wheel speeds than with traction under low/moderate force. In particular, motors tend to be non-linear, and since holonomic drives regularly run different motors at different speeds/loads, the final wheel speeds relative to one another end up different than the ratios the drive calculations ask for.

Closed loop motor controls can prevent this problem; gyro feedback can compensate for it. Without some such mechanism, it's hard to get these things to drive straight, no matter how good the weight distribution.

Weight distribution matters (for other reasons), and driving straight is a challenge (for other reasons), but they're not linked quite so directly as it might appear.

Dumper FTW 435 09-01-2013 12:24

Re: Mecanum vs Treads
 
I didn't know there was so much hate for mecanums :confused:

I would like to say that my team used mecanums quite effectively in 2011 and we were able to get it to work near perfectly without dealing with weight issues or most of the other issues people claim mecanums chronically suffer from. Of course, I think 2011 was a better year for mecanums for this year, but after seeing both treads and mecanums work well, I don't think we should just dissmiss one or the other without seriously considering them. I'm more curious about specific things this year that would suggest treads over mecanums or vice versa, I think the general pros and cons of treads vs mecanums are already well established.

Ether 09-01-2013 12:46

Re: Mecanum vs Treads
 
Quote:

Originally Posted by buchanan (Post 1211444)
Experience is important, and often trumps theory. But it is possible to go wrong tracing back from practical experience to theoretical root causes.

There are challenges in getting a holonomic drive to go straight, but they probably have more to do with how computed wheel speeds map to actual wheel speeds than with traction under low/moderate force. In particular, motors tend to be non-linear, and since holonomic drives regularly run different motors at different speeds/loads, the final wheel speeds relative to one another end up different than the ratios the drive calculations ask for.

Closed loop motor controls can prevent this problem; gyro feedback can compensate for it. Without some such mechanism, it's hard to get these things to drive straight, no matter how good the weight distribution.

Weight distribution matters (for other reasons), and driving straight is a challenge (for other reasons), but they're not linked quite so directly as it might appear.

^^Much wisdom here^^

One other factor to mention: precision mecanum wheels are expensive. The affordable ones most teams use can have significant variation in roller friction and roller axial free-play if not very carefully assembled. This can also affect the precision of the robot's motion.



JamesCH95 09-01-2013 12:46

Re: Mecanum vs Treads
 
Quote:

Originally Posted by Dumper FTW 435 (Post 1211447)
I didn't know there was so much hate for mecanums :confused:

I would like to say that my team used mecanums quite effectively in 2011 and we were able to get it to work near perfectly without dealing with weight issues or most of the other issues people claim mecanums chronically suffer from. Of course, I think 2011 was a better year for mecanums for this year, but after seeing both treads and mecanums work well, I don't think we should just dissmiss one or the other without seriously considering them. I'm more curious about specific things this year that would suggest treads over mecanums or vice versa, I think the general pros and cons of treads vs mecanums are already well established.

In my opinion yhey ought to be compared to the pluses and minuses of a wheeled tank drive and various wheeled tank drive configurations with omni and traction wheels. Treads and mecanum drives aren't the only drives out there.

In 2010 we used mecanum wheels with reasonable success without using any suspension other than a flexible chassis. Driving was reasonably consistent even getting onto and over the bumps. The point many posters are trying to make is that you should be a little careful when designing a mecanum drive.

Edit: there are some really good posts out there discussing the theory, implementation, and control of a mecanum drive by Ether (among others) a thread search might yield some great information.

Bob Steele 09-01-2013 12:47

Re: Mecanum vs Treads
 
During this discussion there seems to be some disagreement about what you meant by "treads". Did you mean regular treaded wheels or a tracked system?

This will be a game that is going to have a ton of contact. Mechanums have an inherent inefficiency because they can't use all of the force directly in one direction. They, therefore, are not by any stretch of the imagination a means to push around other robots or push through other robots.

They do have the ability to be agile and avoid some of the contact.

Because of the "pinch points" on either side of the pyramids this year... this agility might be compromised to a certain extent. They are great in the open field... because other robots can't strafe. As I said this is somewhat negated by the way the field is this year.

I think you should decide what you want to do and just be aware of the limitations AND the advantages of the mechanum (or any other drive).

You might look up a cool drive called the Octocanum drive system implemented by Team 488 in 2011. It has the advantages of both systems, albeit it is complicated... very cool though...great design work done by a great team (Go XBOT!!!)

No hate here...just opinions....why may vary.... (<understatement)

wireties 09-01-2013 14:07

Re: Mecanum vs Treads
 
FIRST Team 1296 used mecanum the last couple of years. As mentioned in this thread there advantages and disadvantages. To make effective use of a mecanum setup the frame & drive train requires precision design and assembly (so all 4 wheels are on the ground at all times), the weight distribution must be close to ideal (so the same amount of weight is over each wheel), the velocity of each wheel controlled (so encoders required at each wheel) and the software well done. If you do it correctly, mecanum is nice. We rarely met a robot that could block us or push us around, we were just too fast and too manoeuvrable. Our drivers could fake out defensive bots kinda like a running back juking a linebacker, spin around them etc. But we did not make it to St Louis, it might not have gone as well against better competition.

That said 1296 is going with a 6-wheel drive this year and building a climber.

philso 09-01-2013 14:13

Re: Mecanum vs Treads
 
Quote:

Originally Posted by Kevin Sevcik (Post 1211236)
See that drawing of that bump around the entire pyramid? That bump means mecanums are going to hate driving near the pyramids.

Kevin is wrong! The mechanums are innanimate objects so they are not capable of hating. It will be the pit crew who have to fix the mechanums after the drivers bump into the lip around the pyramids who will be doing the hating.

At Alamo last year, we learned that we had to check that all the rollers on all four mechanums were turning freely after each and every match. If a roller was binding, we removed it and bent the aluminum spider back out. Bumping into the edge of the bridge (lowered) was just one way to bend the aluminum spiders. Just a few (2-3) binding rollers caused the robot to deviate from the expected path in unpredictable ways. We could see the driving performance degrade throughout a match. I am sure this messes up your force vector diagrams.

We also found that the rollers on our mechanums would bind when new because the urethane had coated the inside of the brass bushings. It took a lot of work to disassemble them and clean them out so that they would spin freely.

www.youtube.com/watch?v=1mrNH0Czq0Il

Ido_Wolf 09-01-2013 14:29

Re: Mecanum vs Treads
 
I'd definitely leave mecanum (at least only mecanum wheels or just 4 wheels, as 6 wheels seem fine and are probably the way our team's gonna roll) out this year. While the ability to maneuver is important by all means, and throwing frisbees might be slightly more simple mechanically, compared to last year's balls (depending on the way you tackle that problem), you really don't want to be pushed way off your stance when trying to shoot. When a robot is so easy to push that every single tackle messes its shots up so bad that it doesn't even seem to be remotely close, it's a very significant issue for scouters who watch it.

Regarding your pick-up issue, it's only the 4th day so it's not too late to rethink your concept, or maybe scrap the pick-up system overall and just rely on the feeded; your chosen mechanisms should be modified so that they stand in line with your ideal strategy, not vice-versa.

commodoredl 09-01-2013 14:37

Re: Mecanum vs Treads
 
Mecanums?
Treads?
Why not both? :p

(Not actually endorsing mecanums, treads, or mecanum-treads)

wireties 09-01-2013 14:39

Re: Mecanum vs Treads
 
Quote:

Originally Posted by Ido_Wolf (Post 1211556)
you really don't want to be pushed way off your stance when trying to shoot.

just gather your 4 disks and park near the pyramid where you can't be bumped/pushed - it might make picking up disks difficult though

Bryscus 09-01-2013 14:41

Re: Mecanum vs Treads
 


That. is. AWESOME!

- Bryce


All times are GMT -5. The time now is 18:10.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi