Given the popularity of the spatula intake, I think that an important issue for many of us is keeping the plexiglass off the ground so as to reduce friction while driving. However, for those of us that are using a catapult, this is not as simple as sticking 2.75" omnis on the end because “firing” two omnis (moving them up and down with the catapult when shooting) in addition to the stars and intake is, at very best, gloriously inefficient. My team has come to the conclusion that fabricating smaller, lighter omni wheels out of existing materials is the way to go. We have tried an “omni ball design” similar to this video using sanded off nylon spacers, but had an issue with making the little t-joint in the center of the ball. We are also looking into a pneumatic tubing loop with spacers that can freespin on it and 1x3 sections of aluminum bars to provide an axis of rotation. I would really appreciate it if you would post any ideas that you have below, but please do not respond that good structure can prevent this issue entirely, because good structure weighs too much for a catapult, especially one that has to unfold.
Good structure can prevent this issue entirely. 
In all seriousness, I was thinking about this issue as well. You may be able to just take a roller off an omni wheel and use that, the only issue would be mounting it and that shouldn’t be too difficult. I think the pneumatic tubing idea sounds good, you just have to make sure you do it in an efficient way as far as weight and size. Whatever you do you have to make sure the rollers are very secure, because they’ll likely quite a bit of stress on them, depending on the design of your intake.
As long as there isn’t a lot of weight on the end of the scoop, I think just a nylon spacer should work. It rolls in one direction, and in the other it doesn’t have much friction if it’s not pushed into the tiles.
Although you did get me thinking about what the smallest omni wheel possible would be…
You could try making a socket out of two bent one-bys and putting a sanded spherical spacer inside of it.
Why would you have to “fire” them?
What he means is that in his example, the intake multifunctions as the platform of the catapult. Thus, any extra weight added onto the intake is extra weight that must be propelled when catapulting. Even though the intake stays attached to the robot, it still requires some force to move.
Yeah, but why lift the wheels? In Clean Sweep, the early robots had the sweeping intake mechanism mounted on the bucket, but later robots mounted the sweeper separately so it did not have to be lifted. There must be some way to support the end of the bucket with wheels that don’t come back up.
Because having a second unfolding mechanism is heavy and reduces the space available to have an intake, making it significantly harder to intake objects at oblique angles.
I have tested with three stars/ one cube and the spacer gets pushed into the tile enough to slow down our turbo four motor 3.25 inch wheel drive significantly.
Do you really need an omni wheel to keep off the ground? Try using something simpler. I think the top of screw held back by a spacer would work really well.
I did too. But I discovered that it digs into the foam tiles with three stars.
Maybe something like snow mobile rails then? That’d help distrubute the weight and not leave any hard edges, especially if you made a curve on the sides, too.
We have done this in the past with lexan. Worked fairly well.
If I had a dollar for every time a student on the VEX forum wrote something off as unworkable without trying it, I would have a lot more dollars than I do now.
I agree wholeheartedly with you. However, I was not writing off your idea, I was simply saying that making a small omni wheel seems to be a theoretically better solution, and my team wants to explore it thoroughly before moving on to other other solutions. If our team cannot come up with a workable design for one, your idea is the first place we will go.
just a note, but you are kind of writing off the omni wheel design itself without testing. 
Definitely something to test.
No worries, but my suggestion was really to try to figure out how to use <a mechanism, including the micro-omni> without having to lift it at catapult speeds. I think the micro-omnis are pretty interesting, but also agree that lifting them with the catapult mechanism is probably not the best idea. I think you can find a way to support the lip of your catapult bucket from the frame without having extra weight that you have to lift.
I’ve been around VEX for a while, and I saw some robots in Clean Sweep that had sweepers attached to the buckets (254A), some had the sweepers attached to the frame (575), and others that used scoops (44). All three of these example teams were very successful that year, but all built different mechanisms. Other teams followed all these strategies, too, obviously, but I remember these three from Worlds. That 254A team may still hold the VEX record for the most tournament wins in a season, and 44 and 575 went to the finals at Worlds.
What about linear slides along the insides of the drive? The slides (probably the old kind) could then make a quick and easy resting platform for the dumper, and you could also put normal omniwheels on the slides to shoot out.
If you want to stick with micro omniwheels, I have a potential idea that you could try. It’s sort of like a sandwich. Start with a metal gear insert, arrange 4 of the long black spacers as rollers around the gear insert, use a single ziptie as an “axle” for all the rollers, put another metal gear insert on the other end, place the whole thing on an axle, and use collars on both ends.
The rollers might be a little wonky trying to turn on the flat ziptie, so you can try shaving the ziptie to make it a better axle.
EDIT: Try some thin string from unraveling the legal rope, too. That might work better than a ziptie for a pseudo axle.
That could work! I’m going to go prototype that right now.
EDIT: It doesn’t work. In order to get the spacers to stay on the inserts, you have to put so much force on them and the resulting friction prevents them from rolling.
Aww…
What happens if you forgo the inserts entirely, and just use the collars?
You could try larger spacers or plastic inserts (possibly sanded) instead of metal inserts.
If the idea of using rollers was to minimize friction from the scoop dragging, it might be better to avoid having the scoop drag at all. My rationale is that the rollers act as a limiter which prevents your scoop from going too deep into the foam surface. Well, if you apply a mechanical limiter or a mechanical tuner which allows you to drop your scoop as low as possible, you fix your problem. Instead, now you don’t need to mount anything onto your scoop which will increase your catapult’s needed power.
If I recall correctly, teams like green eggs from clean sweep and 12 a and b from sack attack didn’t need rollers, but still had great drivetrain capabilities