pic: Jester Drive:Mecanum Wheel Drive Train

did making those wheels take up a lot of your build season time?

The one roller vs. two roller difference is indeed significant. Two rollers allow for more controlled motion because of better release of friction. One roller wheels have a 50% higher coefficient of friction causing for more draw and no extra pushing power for the cost.

To answer another question, building the wheels was a very long process. We prototyped throughout the summer in our SRP (Summer Robotics Project)… our team doesn’t have an off-season really. Anyway, deciding to make and use these wheels during the six weeks isn’t really feasible. There’s a lot that goes into the process, from hubs to molding rollers. You cant go out and buy these wheels pre-made, so there’s a lot of hardworking and trial-and-error stuff that goes into it. But it’s a fun process, and when you finish, you feel like you’ve accomplished something awesome (cause you have).

We haven’t done any tests on the two roller model, as we decided that the machining difficulty for the hub was be far too great and supporting the rollers from only the middle would be difficult. I highly doubt that it is a 50% higher coefficient of friction - the only way it would hurt you is if one roller was going to be spinning at a significantly different speeds. I’m no ME (I’m an EE), but doesn’t the coefficient of friction deal with the actual force you apply to the ground, meaning you would actually have more pushing power? I agree that the two roller design is better, obviously Airtrax had a reason for using it, but I think you are blowing it out of proportion a bit. The single roller design is much easier to manufacture, although it is more parts. (The hub is simply a piece of square 3x3 aluminum stock with the corners cut off). We had absolutely no issues with pushing power.

Thanks Alex. :slight_smile: as the engineer from beach bots said think and have a back up plan before u get into doing this. Making these is extremely labor intensive and you have to make them from scratch I have seen prints for wheels like 357 and there a whole off season’s worth of work as Alex stated.

It’s not coefficient of friction… sorry i hadn’t had coffee yet. The 50% figure comes straight from the studies Airtrax performed. It’s that the single roller model has 50% less release of friction. The two roller model has two independent rollers, meaning that either side can move either forward or backward, so one could be spinning in a different direction than the other allowing for more release of friction.
The one roller design does not limit pushing power, but it draws more juice to get the same pushing power as the two-wheeled design.

How many parts did you need per wheel?

We only needed 30 parts per wheel total. 8 rollers, 8 shafts, 8 screws, hub, Lexan piece, 4 bushings in two sizes. This does not include the sprocket or its attachments.

I’m not an engineer, I’m a student. Thanks for the compliment.:smiley:

Making the wheel we had at nationals (for display only) took about 2 years from initial concept through prototyping to the wheel. The first year: After seeing the ones used to remove containers from airplanes, my dad decided to try to make some. By the end of the summer, we had a set for the Robovation kit. We also had the programming to operate it. The second year: two experimental wheels were made, similar in design to the current ones. When we decided to try them this year and keep six wheels as a reserve, we had a set made. However, when we tested that set, we found that it did not have enough traction, and we could not get another set of rollers designed and built in time to test them before ship, so we went with six wheel drive. We had the textured rollers made before ship day, but have not tested to find out which roller tread is best. More testing to come at a later date.

One set of wheels can be made in about 3 days, from part production to assembly to mounting on a robot.

i’d like to see a picture of what you’ve done. Have you driven with it or anything yet? no bearings? how are you’re rollers attached to the hub? is that even more parts? did you mold any parts inside the rollers? are these mecanum wheels or omni/grenade wheels?

When I first saw 357 strafe, I was driving in a match in Philly, and I don’t know if they were on our alliance or the other, but I was amazed. Especially since how it worked wasn’t immediately visible. I had a look later, and loved what I saw. Very nice job guys.

As for control’s why not go for a keyboard and a joystick? make the keyboard be WASD, for strafing and forwards/backwards, and joystick for turning. A mouse would be even cooler, but a bit harder to implement. Would be neat to see…
I mean it works for First person shooter’s, why not a robot :wink:

A keyboard limits you a lot - allowing you to only move in 4 directions rather than an infinite amount that the joystick allows you. A mouse doesn’t really give you anything that a joystick doesn’t, and you have to worry about running out of room on the mousing surface. We (190) use a joystick for the direction vector (relative to the driver) and a 2nd joystick for rotating counter clockwise or clockwise. We’re probably going to set it up with a single 3 axis joystick to make it more intuitive, however. The second joystick also has the issue with more experienced drivers, when they start driving it tank-drive style.

The three axis joystick is definately what we have been looking at, but there’s one problem with it - it’s designed to be used from vehicle persepctive. Even in talking with Airtrax, they use the joystick for their drivers, and they said that doing it without having that first-person perspective would be difficult. so if we do, indeed give it a try, i suspect we’ll be training with the joystick for a few months.
as for our current controls, we only use a single joystick.
normal: front, back and spin
trigger: direct motion in the direction the joystick is pointing
-it’s actually a lot easier than dealing with two joysticks

We use 2 joysticks on 1083 with our omni wheeld holonomic bot. Its really easy to drive, much like what 190 does.

double DITTO!

Controlling an omni wheel is much different than controlling a mecanum wheel. Since the rollers are perpendicular on omni-wheels, the direction produced has *almost no similarities * with controlling mecanum wheel motion, which depends on the rotation and speed of each individual wheel.

Omni wheels all 45 degrees apart from each other produce exactly the same effect the mecanum wheels do, and you can control the robot exactly the same way. This is how we prototyped the code for our field relative drive and found that we needed to have the split in the chassis to have equal wheel pressure.

the control system that i came up with for mechanums was to have 2 joysticks pitched on 45’s with open loop programing. you could have a noormal tank drive hangling with the abilty still to go left and right. however your driver would have to be realy $@#$@#$@#$@# good. Of course i came up with this because im complletely useless with programing things of any nature except in legos.

As BrianBSL said, its exactly the same. Mechanum wheels have rollers at 45 degree angles, for the omni holonomic drove you put the entire wheel at 45 degrees. There are slight differences but its still the same.

We shimed our wheels so that all the wheels are touchign at the same time. Also we have encoders on everywheel to track how fast each wheel goes. We had a suspension of sorts on our mini bot we prototyped on last year and found it wasnt super important since the field is relatively flat.

You’re completely right… I thought he was referring to omni wheels set in the same frame as our mecanum wheels. In fact, we were encouraged early on to try using the 45 degree omni-wheel set-up because of the complexities of the mecanum drive system. However, there is a disadvantage in the 45 degree omni-wheel drive train when it comes to driving over obstacles. Correct me if I’m wrong, but I think the set-up also has a harder time adapting to different surfaces.

Sorry for the lack of posts lately. I’ll be answering questions more quickly more often.

it looks like the wheels are set up so they are just easier to turn, ie less resistance for more maneuvarbility

Sorry I took so long to respond, but here are the answers.

Have you driven with it? I only drove the EDUbot prototype. Others have driven the real thing.

No bearings? Two bushings pressed together (one was smaller than the other) formed the bearing on each side of the wheel. Nylon is slippery enough that the rollers didn’t need bearings.

How are your rollers attached to the hub? The hub looks something like two thin Skyways with a connecting bar between the hubs. The rims have holes in them to hold steel axles, which are kept from falling out by a piece of lexan screwed to the outside.

Is that even more parts? No. The roller axles were labeled as shafts in the parts list.

Did you mold any parts inside the rollers? Nothing in the wheel was molded. It was all formed using SLS (a laser drawing patterns in powder). There are no parts inside the rollers.

Are these mecanum wheels or omni/grenade wheels? Mecanum, something like what some airlines use to unload containers from airplanes.

The only picture I have is of the side of the hub, nothing else was attached.