pic: Omni Design Trial



A trial design I did for an Omni/Holonomic drive system/chassis. The black material is 1.25" thick UHMW. The rest is aluminum sheet metal, 0.1". Electronics would go in the middle, with the black below the orange sheet being a sheet metal holder. Wheels are 8" AndyMark Aluminum omnis, motors are 2.5" CIMs. I’m revising it to fit a planetary gearbox in with all the motors. Upper left battery design is to hold battery, lower right is where a joint for adding a manipulator / arm can be placed (placeholder currently takes that spot).

I’ve never designed a full robot myself, let alone an omni system, so I’m welcome to any comments and criticisms. Let me know what you think. Thanks!

Looks nice. One problem though, currently won’t be going anywhere, with just a cim coupled straight to a wheel. Looks nice and light. How much does it all weigh?

Do you have a waterjet or lasercutter, because all those font are going to take a nice while otherwise.

It looks really nice. How fast will this go after you get the gear boxes in and how would you account for the wheels slipping?

It looks like a neat design, I wish I could use Inventor so well!

One comment on the drive choice, it seems to me that omni drive like this has a lot of “wasted motion” since either all 4 wheels are always turning at an angle to the direction you’re going, or two are going in the direction of movement and the other are not turning at all. I suggest you look into changing to a mecanum drive setup, the only substantial change to your design is that you would rotate all 4 wheel/motors 45 degrees so that the wheels are parallel to the long dimension of the frame.

Mecanum has an advantage with this type of game, where you will probably spend a lot of time driving straight ahead, and in this situation the rollers on all 4 mecanum wheels do not rotate relative to the wheel body. With mecanum, you only have “wasted motion” when strafing (sideways) or rotating the robot.

(if you’ve already ordered the omni wheels it might be expensive to change now though!)

Also mecanum programming is something your team should be able to do, there are several code examples on the web that you can use to make it work.

Is this true? I was under the impression that a Mecanum wheel with 45* rollers was essentially the same as using 4 omni wheels located at 45* angles.

If the rollers are at 45* angles how is there any difference in what the part of the wheel contacting the floor any different in the forward direction than it is side ways.

This is hard to visualize! but easy to demonstrate if you have a mecanum wheel to play with. Unfortunately they’re hard to find…and hard to make…and expensive…

The rollers act as “tread” when the weel is going straight ahead. To be able to roll relative to the wheel, there must be sideways motion. If there is no sideways motion, the roller in contact with the floor cannot roll, it can only “go along for the ride” as the wheel turns.

(I am assuming the rollers get good traction on the floor)

I was under the impression that the rollers were moving sideways but since the forces for the left side were equal in magnitude but opposite in direction they resulted in no net force (except forward).

I guess I need to get my hands on some Mecanums.

Another issue with both omni and mecanum drive, is that the robot would work better with some type of suspension, especially since there are “lumps” in the arena floor surface near the walls and overpass supports.

Without suspension, the robot will drive erratically when the weight on any wheel drops significantly, such as when driving over any bump that would lift a wheel off the floor. Both of these drivetrains work because all the wheels work “against” each other, the result is that it goes the way you want it to. If you lose traction on a wheel, that wheel won’t be pushing the robot the direction it is supposed to, so the robot will go wonky…

There is an opposing sideways force on the rollers on both sides of the robot, but since they are constrained by the forward direction of the robot, they cannot roll sideways, so they can’t roll at all.

If this were correct a mecanum drivetrain couldn’t produce diagonal motion. It’s done by driving the corner wheels, but if what you say is correct, the robot would just drive straight anyway.

In ideal conditions a mecanum wheel will behave exactly like an omni wheel at 45*. Each system will be approximately 70% (sqrt(2)/2) efficient in forward-backward and side-to-side motion.

The confusion comes in that real life is never “ideal.” 70% efficiency assumes that the rollers will have zero friction with their shafts. In an omni bot the inefficiency is universal and generally cancels itself out. In a mecanum bot however, the inefficiency causes the mecanum wheel to behave more like a regular wheel, so you end up with a little more than 70% power in forward-backward motion and a little less than 70% in side-to-side motion.

If you completely eliminate the roller-shaft friction, you get a perfect 70% efficiency, and likewise if you completely lock the rollers to their shafts you get a regular wheel. Reality lies somewhere in between.

well I for one like omnis over mecanums!! Really solid looking design, except for the lack of gearboxes, we are using omnis too!! We find the mecanum bots we have seen in person drifting side to side too much, as they are always balancing forces. good luck!! nice colours too!

I was not talking about the case where the robot is moving sideways as well as forward (diagonal motion). In this case, the rollers do roll, as you say.

In ideal conditions a mecanum wheel will behave exactly like an omni wheel at 45*. Each system will be approximately 70% (sqrt(2)/2) efficient in forward-backward and side-to-side motion.

I disagree with this. I have a mecanum wheel in my hand, and it has loose rollers, they turn very easily. When I hold the wheel so that one roller is contacting the desk, and then move the wheel (without turning it) so that the roller rolls without slipping on the desk, the wheel body moves in a diagonal direction.

Now I’ll turn the wheel body, pressing down towards the desk, but keeping the wheel moving only in a forward direction, in line with the direction the wheel is pointing, and perpendicular to the axle. This would be the same as just rolling a normal wheel across the desk. When I do this, the rollers on the wheel do NOT roll relative to the wheel, they act as a (segmented) wheel tread.

Any time that there is good traction and the rollers are rolling relative to the wheel body, then the wheel is moving sideways. The geometry of the wheel makes this a fact.

I guess I need to make a movie…

Anyways, the way mecanum wheels work, they are NOT like omnis, and they do NOT have the same efficiency going forward and sideways/diagonal.

This is an advantage of mecanum wheels.

There are no “cases” involved. The wheels behave the same always.

Rolling a wheel and powering a wheel are totally different. If you placed a keyed or hex shaft through the wheel and tried to turn it while pressing into the ground, you’d see the wheel would want to walk down the shaft. This is because the wheel produces a vector that isn’t straight ahead, but at approximately 45*.

This is correct, and I said as much in my previous post.

Ok, let’s try changing tack. Assume the mecanum wheels do put 100% of the power forward. We know the gearboxes can’t output more than 100%, so there can’t be any force sideways. Therefore, the robot could not strafe sideways.

However, we know that the robot can strafe sideways. Therefore our assumption is false and the wheels must be producing a sideways force all the time.

The mecanum wheels do not put 100% of the power forward. They behave just like regular omnis at 45* angles. You get about 70% power by moving forward, reverse, left, or right.

We do have access to a water jet, luckily. Classically all the designs coming from our team have consisted of large amounts of sheet metal. Which is pretty cool, because although it makes them all a bit homogeneous, they give a certain “1293” style to our robots with the artwork we cut into them.

Mostly what I’m worried about with this design is whether the motor mountings are strong enough. Like I said, this is the first full design I’ve done myself (Actually, first design I’ve really ever done except changing someone else’s) and I’m not an engineer… I’m actually the team programmer.

My point in having Omni’s vs. Mecanum is twofold. First, the one my team will probably be most happy with, is that they’re a bit cheaper, and it makes every single motor assembly interchangeable for repair/replacement. But from a design standpoint, it’s because it then becomes just as easy to herd a ball on the long end (therefore less change of it shifting left/right outside of the robot’s pushing front) as it is to drive around robots on the short end.

As for the gearboxes, those will most likely be the same ones Banebot supplied from last year (We had ordered a couple extra, but they didn’t arrive until after the season, so we can play around with those… We’re thinking about doing what we did in 2006 and building a duplicate to keep at the shop).

The main idea behind this design was to make the entire thing modular. Any part can be removed from the base without more than, worst case, 6 bolts. And those on opposite sides of the robot are interchangeable, so if something gets bent up, we only need one extra for 4 shapes and we’ll have a spare for anything we’ll need.

Also, any idea on whether the strength of UHMW is up to the challenge of a full chassis? I’d rather not have a $2K piece of UHMW turn into two cracked-in-half robots. (Which is partially why I added some of the extra sheet metal, for reinforcement).

As for the current weigh in, it’s at 40 pounds based on Inventor. But I believe the estimate of weight in Inventor for the motors is a bit low (I think it says 2 pounds), so it’s probably much more realistically around 65 including bolts and bearings. I’ll also need to add in the gearboxes, so my best guess is that it will end up around 85 pounds. Forty pounds should be plenty for the electronics and arm, I think. Although we do have a nasty habit of always weighing in at 120.1 pounds… or worse. (I think one year we dropped 8 pounds by cheese-holing everything possible for our robot, with a quite humorous effect on its looks).

Thanks!

All other things being equal – wheelbase, track width, coefficient of friction, etc. – mecanum wheels and omniwheels function in exactly the same manner. There are no differences. They are mathematically identical.

George is entirely correct in his explanations. End of story.

Since you brought up cost of UHMW, I checked McMaster. A 48" square sheet of 1.25" UHMW (item 8619K97, the smallest size that would let you get 28"x38") is $348.38. A 36"x24" piece (item 8619K95) is $164.58.

Just something to marinate on.

Obama got Oprah, Huckabee got Chuck Norris, and George1902 got M. Krass.

No, it’s not the end of the story…because I have a mecanum wheel sitting here, and you don’t :slight_smile:

No, they don’t. Mecanum wheels behave differently when they move only in the direction of the wheel, than they do when the move any other direction. In the straight ahead case, the rollers do NOT roll relative to the body of the wheel. This is because in this case they CAN NOT roll relative to the body of the wheel, as long as they have full traction. This is a fact, try it and see!

Rolling a wheel and powering a wheel are totally different.

Yes they are…as far as force. No, they are not…as far as velocity.

If you placed a keyed or hex shaft through the wheel and tried to turn it while pressing into the ground, you’d see the wheel would want to walk down the shaft. This is because the wheel produces a vector that isn’t straight ahead, but at approximately 45*.

Exactly!!!

But if you constrain the wheel so it cannot move sideways, it still produces the sideways force vector, but the sideways velocity vector is zero.

Ok, let’s try changing tack. Assume the mecanum wheels do put 100% of the power forward. We know the gearboxes can’t output more than 100%, so there can’t be any force sideways. Therefore, the robot could not strafe sideways.

…while the wheels are only rolling straight ahead. There are two different cases, which you seem to be ignoring. The straight ahead movement case is the only one where the motors can transmit all of their motion to the robot. In the strafing case, the motors are not transmitting all of their motion to moving the robot, and the robot goes slower when strafing than it does when going forward, assuming the same average motor speed among the 4 motors in both cases.

However, we know that the robot can strafe sideways. Therefore our assumption is false and the wheels must be producing a sideways force all the time.

No, the wheels are producing a sideways force all the time, but are only producing a sideways motion when strafing! Please try to understand that force does not equal motion.

The mecanum wheels do not put 100% of the power forward. They behave just like regular omnis at 45* angles. You get about 70% power by moving forward, reverse, left, or right.

Nope, with mecanum you get 100% going forward, and less when going sideways or at an angle.

Also, consider the case of omnis vs. mecanum, with all motors turning in such a way that none of the rollers are rolling relative to the wheel bodies. In the case of the mecanum, the robot goes straight along the axis of the wheels. The omni sits and spins in a circle. So, the omni is most efficient when it is not going anywhere!

So it is official, George gets the 08 nomination, congrats :wink:

Wait, so when you are rolling forward and there is a sideways force, where is that force coming from? It is the motors. You don’t get the applied force for free.