17-06-2016 19:38

Scott Kozutsky

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

Sorry, double posted the pictures.

17-06-2016 20:07

ctt956

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

Good concept IMO. This would probably also be advantageous if a motor/gearbox/drive module failed, as it seems like it could be pulled along by just one wheel. All of the gears should probably be protected better so debris doesn't get in them.

17-06-2016 20:35

The other Gabe

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

this is ridiculous, a complete waste of time, and probably not particularly effective. I love it!

17-06-2016 20:48

Chak

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

It would be a very interesting math problem to calculate the theoretical max speed. I don't think it's 1.414, because that either assumes the motor can get up to max speed immediately after a fast turn, or it assumes that omniwheels have no friction laterally. Instead, as the robot turns, it decelerates in one direction and accelerates in the other. With the omniwheels, the net acceleration will probably increase the speed. But to what maximum?

17-06-2016 21:03

Scott Kozutsky

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

"hypothetical" might be a better term for that. This unit is geared pretty low (free speed at 14.9 fps) so the acceleration on 6 CIMs should be pretty good. The lateral resistance of omni wheels is kind of an unknown at this point.

17-06-2016 21:07

Chris is me

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

I'm really having trouble seeing what this gains over a normal kiwi drive setup.

If you have six motors on it, you're going to have plenty of torque in any direction of travel even if all wheels aren't pointing in the same direction. You'll be able to do advanced maneuvers (simultaneous rotation + translation) with a kiwi drive, but not with this drive.

17-06-2016 21:30

Scott Kozutsky

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

You'd have one and a half times as much power in any direction ,assuming a power if 1 unit per wheel you'd get 2*cos(30)+1*cos(90)=1.73 to 1*cos(0)+2*cos(60)=2 for kiwi and 3*1=3 for omniswerve. If you have the motors mounted you might as well use them.

You are able to do advanced maneuvers with this because you can just tank turn at any time, there's no scrub from the omni wheels so even with the sub optimal platform it would still work.

17-06-2016 22:35

Chak

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

I think we can all agree that this design is not meant to be realistic anyways.

17-06-2016 23:27

Dan_Karol

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

Maybe i'm missing something, but how does it turn?

If all pods are ganged together then you have no control over what your heading is or you have no ability to strafe.

17-06-2016 23:56

GeeTwo

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

TL;DR Trying to wrap my brain around what this does and why you'd want to do it.

If all the wheels steer in unison, you have crab drive - with the inherent [crab] inability to rotate the chassis (not good for a triangular chassis in most conceivable games) and the additional [omni] weakness of being highly vulnerable to cross-drive forces. I don't see any way to get anything faster than geared speed in this scenario, either.

If only the wheel near the steering motor is steered, I can imagine a scenario where you can get up some extra speed, though I haven't done the math to convince myself it's viable. If this is the case, why are the two non-steered wheels built up so they appear to be swerve modules?

Or is something else going on (e.g. one of the wheels steers counter to the other two, or something even more arcane)?

Edit:

Maximum speed is achieved when the drag/frictional forces are equal to the drive forces, and the drive forces are applied. Once you turn the drive in another direction, you'll begin to slow in the direction you are moving. You may be able to achieve a bit better maximum velocity, but after accounting for the longer distance traveled, I don't see how you can achieve a higher net speed.

Edit2: That is, at maximum speed, saying you don't lose forward momentum, even ideally, is not valid. Maximum speed is inherently the result of significant loss of forward momentum due to friction.

17-06-2016 23:59

Scott Kozutsky

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

The way I have it is they all point in the same direction. You can tank turn in any orientation by simply driving one side forward and one side back. Only the pivots are chained together not the actual drive.

The reason you can't turn in conventional crab is because the drive is all chained together and because the wheel scrub from trying to tank turn is too significant.

here's a crude vector diagram of my 3 wheeled swerve rotating.
http://imgur.com/0HMYj0g

If they were traction wheels the steering would be awful (non existent) because the wheels would oppose the lateral forces exerted on them. With omni wheels this does not happen so it should be able to turn quite nicely.

18-06-2016 00:48

Aren Siekmeier

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

Hahahahahahahahaha.

I love the three wheels though. Also, your packaging on a non-coax module is not bad.

18-06-2016 01:17

asid61

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

How are you getting above max speed? You might get more power in a direction (like with a kiwi), but not more max speed.
The major advantage I can see to this is being able to run a crab drive without worrying about turning scrub, but at that point one might as well run mecanum or pure swerve. I think this would be usable for a team with low programming resources that wants to get into swerve (say during the build season) but wants a good backup plan for if it doesn't work out. Crab is significantly easier to program than full swerve.

18-06-2016 07:06

Scott Kozutsky

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

drive straight to max speed, turn left at 90 degrees and accelerate in that direction to max speed. Because of the omni wheels, you actually (ideally) don't lose your forward momentum after turning and, as such, move in both directions at your maximum geared speed, hypotenuse of that vector triangle is 1.414 times the side length.

There are also the standard "butterfly" advantages of this over traction wheels. See 33 in 2014 for similar reason, why would they go for omnis over traction in a standard tank. This concept is silly but IMO does merit actual consideration, even as just a thought experiment; it has some significant advantages.

18-06-2016 07:20

Gdeaver

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

It's summer. No pressure. No time constraints. Why waste your time on this project. Isn't there something your team needs to work on to improve it's skill set? Something that could actually be used in the future?

18-06-2016 07:52

Scott Kozutsky

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

This was entirely my own project and an excellent opportunity to brush up on CAD. I figured Chief Delphi posters in general would also enjoy this kind of thought exercise but apparently not.

18-06-2016 12:37

Aren Siekmeier

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

The sqrt(2) speed increase is certainly a different way to think about things. I imagine it wouldn't coast terribly far on the rollers, and it's not powered through any obstacle in that direction. And sounds like you've considered how it would mess with the driver to be in a constant uncontrollable powerslide.

Put in a traction wheel and you've got a decent crab drive. Two more turning motors for unicorn.

18-06-2016 12:44

nuclearnerd

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

On the contrary, I think this is really cool! The way you have thought past everyone's initial "why" to discover some really novel performance details shows that you have the creativity, and deep thinking a robotics team needs.

I spent some time last year thinking about a wheeled version of a cyclorotor for similar ridiculous reasons

18-06-2016 15:25

GeeTwo

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

I also enjoyed the thought exercise of looking at the drive and working out the probable advantages/disadvantages of its distinctive constraints, even if no one could come up with enough advantages to want to build it.

I wish we had a few team members "waste time" this summer developing CAD skills.

18-06-2016 19:20

Cothron Theiss

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

There's never any reason to not mess around with CAD for practice, so it's awesome that you've done this. I'm hoping to post several oddball/silly designs over the next month, as soon as I get off my butt and get to CAD-ing. But great job on getting the ball rolling with off-season designs.

As for the design itself, it might be really interesting to lock the back two wheels in the forward direction, and then replace the forward omni wheel with some sort of traction wheel with VERY high amounts of scrub (or even a small tread module). Theoretically, if the forward steering module has enough friction against the carpet, powering the motor on the front steering module would actually turn the entire chassis in relation to the front wheel. With the front wheel pointed in a different direction from the back wheels, it might steer similar to a front wheel drive car, but without the need for a differential since the back Omni's can be powered at different speeds.

19-06-2016 21:42

Scott Kozutsky

pic: OmniSwerve chassis Bottom View

19-06-2016 21:44

Moskowapplepi

Re: pic: OmniSwerve chassis Bottom View

I am curious how well this would work in practice as well as how the triangle design came to be.

19-06-2016 22:03

EmileH

Re: pic: OmniSwerve chassis Bottom View

Relevant thread/threads.

19-06-2016 22:41

Scott Kozutsky

Re: pic: OmniSwerve chassis Bottom View

The triangle design is a byproduct of maximizing motors and minimizing other complexity. Minimum amount of wheels to be fully supported is 3 and maximum reasonable amount of CIMs is 6. It also ensures all the wheels are in contact with the ground. 4 CIM 4WD would end up weighing about the same, be more mechanically complex and less powerful.

22-06-2016 23:53

Hitchhiker 42

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

With the max speed thing, here was the way I though you could possibly do it:

Have one wheel face one way, another face 90* from it, and the third point right in between. If you power the two at 90* to each other, you'd get sqrt(2) the speed.

23-06-2016 00:48

asid61

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

The velocity vectors don't add like that I think. You would get sqrt(2) + 1 times the torque of the single cim pointing in the right direction, but the max speed is still only the speed of that cim.

23-06-2016 08:46

GeeTwo

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

The velocity vectors can add up so that the speed of the robot is greater than the speed that any individual wheel is driven. The issue is that with the number of rollers providing resistance to this motion, I find it unlikely that those two motors could achieve a significantly greater speed than one in which all three motors are providing drive force in the desired direction. The only way to know for sure is to build it.

23-06-2016 15:02

asid61

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

Well, if you take the two non-pointing-in-the-right-direction motors and give them each an angle theta away from the main wheel...
If you assume velocity vectors add, then the output speed is (cos(theta)+1)*v0, right? If that's the case, as theta goes to 0* (AKA point all whee;s in the same direction) one would end up with just 2 times the target speed- but all the wheels are just pointing in the same direction like a tank drive. That doesn't make sense to me, is there another way to add it?

23-06-2016 16:02

Aren Siekmeier

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

Force vectors add. Velocity vectors do not. Rather they are constrained by slip conditions and the relative motion of bodies.

It's sort of a continuously variable transmission.

Say you are constrained to go in a direction 0 (by a wall or omnis opposing each other, etc.), but an omni wheel is pointed in a direction theta relative to your heading, and spinning at speed w about its axis. The overall translation velocity vector v is identically the velocity of the center of the wheel relative to the ground, which can be broken into the rotation of the wheel (w), and the rotation of the rollers (s), assuming no slippage. Then v must project onto the wheel speed w, as w=v*cos(theta). The roller speed is s=v*sin(theta) and has no limit. The overall speed is v = w/cos(theta), resulting in a higher speed.

To do this, you could have a four wheel drive with steerable omnis in each corner. Start with them directed forward, (theta=0), and spin the wheels at speed w. Then v = w / cos(0) = w. Then turn the left and right wheels toward the middle by an angle theta. Now v = w / cos(theta) will give an increase in top speed.

However, the force available in the forward direction drops. Each wheel supplies F in its own direction. When theta = 0 then the total force is 4*F, all straight forward with no sideways component. When theta is increased, the sideways components cancel, but at each wheel, the sideways component must add with the forward component to get F in the direction theta. So the forward component from each wheel is F*cos(theta), and the overall force available is 4F*cos(theta).

In this way, steering the wheels toward the middle by an angle theta acts as another stage of reduction, increasing the speed and decreasing the force by a factor of cos(theta), assuming no wheel slippage.

Vector diagram to help

23-06-2016 18:28

bkahl

Re: pic: OmniSwerve chassis Bottom View and Concept Discussion

#fizzikz