Mecanum Drivetrains

[Gabe1511]

Hello, my team is planning on developing a mecanum drive train to test with, and I was wondering what designs have been successful for other teams. So, what designs have yielded the most success, and where have there been failures and limitations with your designs?

[Pendulum^-1]

Limitation number one: Mecanum wheels are very easy to push. A problem on a field with well-defined chokepoints, such as the 2013 arena.

[Jay O'Donnell]

The Mecanum drive train that 1058 has been successful with in the past is a simple one-CIM gearbox for each wheel and a c-channel frame. When I'm not on a school computer I will edit this post and add a video my team made a couple of years ago explaining our mecanum drive and how it was field oriented (meaning that if you pushed the joystick away from you the robot always goes in that direction, no matter what orientation it is in).

[JeremyLansing]

For the mecanum drives Team 967 built in 2010 and 2011, we used the kit frame with four AndyMark Toughboxes (2010) or AM Toughbox Nanos (2011). It worked out pretty well for us, and although it isn't entirely required, we did use a gyro both years. I put a link to a video of our 2010 robot at the bottom of my post. You will hear a lot of opinions about whether mecanums are worth the effort, and I'm sure someone will bring up the mecanums on Einstein statistic, but if you implement them right, you can still be effective. Just don't expect mecanums to be a magic bullet that makes your robots better. Mecanums are not necessarily better or worse, they play to different abilities that need to be balanced in your decision making process. But since this is the offseason, prototype away!

Edit, added the link
http://youtu.be/fqeAReWKA6s

[Brandon Zalinsky]

To add to Jay's post, we used 4x 30:1/11:1 supershifters with one CIM each. In the past, we have run each of a Jaguar, but we have blown out so many Jags that we have switched back to Victors.

If you want to get creative, you can do all kinds of crazy things with mecanums. For example, in 2011, 1058 built this simple 4-cim mecanum drive with a twist. Each wheel had a third mecanum plate with cut-up pieces of truck mud flap on it. This plate could be actuated into the rollers, freezing the rollers in place, but not the wheel. With these activated, the robot would be a high-traction 4WD that could push other robots around. At the press of a button this would switch to the normal mecanum drive, giving it unmatched speed and maneuverability.

Coming from one of the biggest mecanum fanboys in FIRST- have fun, drive fast, and forget the haters.

[Gabe1511]

Thanks for offering to post the video! My team has been wanting to rig up a mecanum drive train for the past few years, but we've never done it, and I've decided to spearhead the effort this year. Through scouting I've learned the advantages and disadvantages that accompany mecanum drive trains, but I've just wanted to be able to play with one.

[Koko Ed]

Originally Posted by Gabe1511

Hello, my team is planning on developing a mecanum drive train to test with, and I was wondering what designs have been successful for other teams. So, what designs have yielded the most success, and where have there been failures and limitations with your designs?

Well you guys got to see our drive base in action plenty of times this year and it worked great-until someone focused defense on us.

[Akash Rastogi]

Originally Posted by Flak-Bait

If you want to get creative, you can do all kinds of crazy things with mecanums. For example, in 2011, 1058 built this simple 4-cim mecanum drive with a twist. Each wheel had a third mecanum plate with cut-up pieces of truck mud flap on it. This plate could be actuated into the rollers, freezing the rollers in place, but not the wheel. With these activated, the robot would be a high-traction 4WD that could push other robots around. At the press of a button this would switch to the normal mecanum drive, giving it unmatched speed and maneuverability.

This sounds pretty creative! Do you have pictures or videos of it in action?

[MechEng83]

There are technical and strategic challenges to implement a mecanum drive effectively. I will attempt to briefly address both.

Technical - We've done mecanum for the last 2 years in 1 form or another. Our gearbox of choice is the AndyMark Toughbox Nano - 1 per wheel. I definitely recommend driving the wheel directly out of the gearbox rather than through a chain. We use quadrature encoders on each wheel and have PID tuning to make certain we control each wheel's speed, rather than just input voltage. We use a gyro on the robot to allow field oriented drive, while having an alternate robot oriented drive. Make sure that your frame is not too rigid. The robot can drive eratically if all 4 wheels are not on the ground. Our 2012 robot had what is sometimes called octocanum drive, where we switched between mecanum and traditional wheels. We designed the shifting pistons to support the robot weight on more than 3 wheels. This level of active suspension is not required, but was a side benefit in our 2012 design. Test with the robot weighed down to final competition weight - drives function differently with varying amounts of load.

Stategic - Mecanum is not better or worse than a more traditional drive. It is different. Teams who do not make strategic changes will be ineffective in using a more versatile, albiet, less pushy drive. Almost everyone understands that in a pure pushing match, mecanums will lose. What many people don't realize is that you can maneuver around a robot much more easily to avoid them or get out of a sticky situation. Driver practice is key, as the extra degree of freedom means the driver has an additonal drive component to control. Practice avoidance maneuvers around pushing robots. Practice lining up in the positions you need on the field. It pained me to see teams with mecanum drives this year driving them as a tank robot to get into a specific location, when they just needed to move sideways 6 inches.

[BBray_T1296]

Several Very VERY critical points for a successful mecanum drive:

-SQUARE FRAME. Not as in length (though a square shape IS the most efficient for mecanum), but as in straightness and lack of vertical warping. That is, all 4 wheels need to be perfectly level to get ideal force distribution. If you are worried about having a lifted wheel, put some sort of rubber or spring mount to allow the robot to settle in a planar configuration. When you design the robot, try to create close to a square shape too. We did rectangles both years, but the only way that works is with a "narrow 'bot" configuration. ("wide 'bots" will not work well!)

-WEIGHT DISTRIBUTION. When you construct your 'bot, try to keep your CoG in the center. Not forward, not backward, not left/right, but dead center (ideally). Having a shifted CoG will cause one wheel to have more traction than others (f=uN) and will cause the robot to skew, particularly when strafing. In all drivetrains, low CoG is ideal, so that too, but less importantly.

-DRIVER EXPERIENCE. your drivers need TONS of practice, even if that means playing around with just a base for a while. Know the ins-and-outs of mecanum, and it is advantages. As previously mentioned, mecanum has poor pushing power, but with a skillful driver and well made mecanum (coding is a major factor), you can use the agility to dodge even the toughest defensive 'bot. In 2011, we burned a jag mid competition, and had to compete a match with just 3 operating wheels. From a spectator's perspective, you would not have even known the difference. Of course, holding "up" on the joystick did some funky things, but it was easily correctable for a guy who knew what was going on.

-CODING. Make sure you use encoders, as wheel speed is the single most critical factor. I'm not a coding expert (hardly a grasshopper), but perfectly calculating wheel speeds in all situations (try driving forward, while strafing left, while turning clockwise. Imagine what the wheels have to do!) will really give you an advantage. This is probably the most daunting problem teams have when building mecanums, and where most teams who have tried it and hated it went wrong.

If you don't already, depending on your driver's preference (Xbox or Playstation) get either a wired Xbox controller, or a Logitech controller. (wireless Xbox 360 and PS3 controllers are illegal, and I dont think there is a wired PS3 controller, hence the basically identical Logitech). Driving a mecanum robot is a lot like playing a first-person-shooter. Configure the joysticks to replicate how you would play Halo or Call of Duty, except without the "look up/look down" axis. This will make driving feel far more natural to drivers, and will let almost anyone have the basic ability to drive the robot instantly

Both years we did mecanum (2011, 2012) we used toughbox nanos (I dont know the ratio) with mecanum wheels directly on the output shafts. We built custom assemblies to house each wheel including our deflector plates for hopping the bump in 2012.

[Brandon Zalinsky]

Originally Posted by Akash Rastogi

This sounds pretty creative! Do you have pictures or videos of it in action?

Yep! They're around somewhere, i'll find them and post them tonight.

[Gabe1511]

Originally Posted by Flak-Bait

If you want to get creative, you can do all kinds of crazy things with mecanums. For example, in 2011, 1058 built this simple 4-cim mecanum drive with a twist. Each wheel had a third mecanum plate with cut-up pieces of truck mud flap on it. This plate could be actuated into the rollers, freezing the rollers in place, but not the wheel. With these activated, the robot would be a high-traction 4WD that could push other robots around. At the press of a button this would switch to the normal mecanum drive, giving it unmatched speed and maneuverability.

How effective was that strategy? It definitely sounds like it would appeal to my team, giving the best of both worlds.

[Taylor]

Originally Posted by BBray_T1296

Several Very VERY critical points for a successful mecanum drive:

-SQUARE FRAME. Not as in length (though a square shape IS the most efficient for mecanum), but as in straightness and lack of vertical warping. That is, all 4 wheels need to be perfectly level to get ideal force distribution. If you are worried about having a lifted wheel, put some sort of rubber or spring mount to allow the robot to settle in a planar configuration. When you design the robot, try to create close to a square shape too. We did rectangles both years, but the only way that works is with a "narrow 'bot" configuration. ("wide 'bots" will not work well!)

I disagree with this. Many teams encountered difficulties when driving over uneven field. When field components are placed under the carpet, as they often can be, this takes your carefully tuned 'square' base and lifts up a corner or two. I'd recommend taking Nathan's advice and leaving the chassis 'loose' or having some sort of suspension to negate 'imperfections' in the field.

-DRIVER EXPERIENCE. your drivers need TONS of practice, even if that means playing around with just a base for a while. Know the ins-and-outs of mecanum, and it is advantages. As previously mentioned, mecanum has poor pushing power, but with a skillful driver and well made mecanum (coding is a major factor), you can use the agility to dodge even the toughest defensive 'bot. In 2011, we burned a jag mid competition, and had to compete a match with just 3 operating wheels. From a spectator's perspective, you would not have even known the difference. Of course, holding "up" on the joystick did some funky things, but it was easily correctable for a guy who knew what was going on.

We found quite the opposite during our offseason event - rookie drivers with very little practice driving a robot had no preconceived notions about how a tank-style drive handles, so they bobbed and weaved and spun around traffic with ease using our sensor-less and simply-programmed mecanum drive.

[Andrew Schreiber]

Originally Posted by Taylor

I disagree with this. Many teams encountered difficulties when driving over uneven field. When field components are placed under the carpet, as they often can be, this takes your carefully tuned 'square' base and lifts up a corner or two. I'd recommend taking Nathan's advice and leaving the chassis 'loose' or having some sort of suspension to negate 'imperfections' in the field.

This. 1000 times this.

Rigid frames are the death of holonomic systems like this (omni or mecanum drive, swerve is a whole different ball game I don't have enough experience with to talk about).

You want flex because you want each wheel on the ground at all times. Not doing this will result in unpredictable behavior... well, ok, it's completely predictable given that you know that you're driving over uneven terrain and what the normal force on the wheels is and... blah blah blah.

[Jay O'Donnell]

Originally Posted by Gabe1511

How effective was that strategy? It definitely sounds like it would appeal to my team, giving the best of both worlds.

The locking mecanums were both effective and cool to show off and demonstrate. It's also fun being able to tell people we pushed a traction drive with mecanums.