4WD Turning Difficulties

[Doug Leppard]

Originally Posted by ZakuAce

Has anyone seriously considered a robot with two wheels in the middle as the drive wheels and having it pivot? We're thinking it would get good traction, as all the weight (well most of it) would be on the two drive wheels, and the trailer would help balance out the wheels to keep the center of gravity in the middle. Also, the simplicity factor would allow us to spend more time making a way to score.

We considred it for awhile but figured we could not use the hitch to keep us balanced and it was too risky. But we are trying to keep the CG over the rear wheels as much as possible and trying to achieve about the same thing.

[ZakuAce]

we were thinking of using more wheels to support our robot at the front and back when it pivots. The mentors were thinking that even though the robot may be resting on an un-powered wheel at some point, the amount of force that is placed on those wheels will be negligible, as most of the force will still be on the center drive wheels. I guess it is more like the 6 wheel drive with the lowered center drive wheels, only we'll use 5 (2 drive, 2 in front on each side of our ball gatherer, and 1 in the back)

EDIT: So... the trailer hitch can never change from that height? Well that could be a problem. I suppose we could have the rear supporting wheel powered and at the same height as the main drive wheels, and have it programmed to run only when moving foreword and backword. I really want to keep the number of drive motors as low as possible.

I'll bring up this idea that was really good, but ultimately shot down because of complexity.

One of the mentors brought up an idea of a three wheeled system. it would work similar to a crab drive, but has a huge difference. The base that the wheels are attached to would be able to rotate, or rather, the robot would rotate around it. The wheels move, then the entire body moves around the base to face the direction of movement. It seemed really stable and a great way to keep the same weight distribution on all wheels, but it was rather complicated and could interfere with a ball gathering mechanism.

[Erin Rapacki]

To this general topic about 4WD: I designed the wheel base to be wider than it is long, the closest being... make sure the insides of the wheels make a square. This way the wheels don't have to slip along an arc across a surface while it's turning.

[writchie]

Originally Posted by Doug Leppard

We considred it for awhile but figured we could not use the hitch to keep us balanced and it was too risky. But we are trying to keep the CG over the rear wheels as much as possible and trying to achieve about the same thing.

From the drawings of the swivel and other parts it appears that the design intent is to create a rigid attachment without freedom to rotate about the trailer wheels.

The hitch pin is a close fit. The close fit requires that the hole in the swivel have a vertical axis. The same is true for the holes in the c-channel. The 0.385 of gap is, IMHO, intended to allow for the tolerances in the height of the hitch swivel (when level) and the robot c channel (when level). I would not expect the swivel to ride much up and down on the pin. Of course I could be wrong and only testing will reveal how rigid the attachment is in practice.

[ZakuAce]

Originally Posted by writchie

From the drawings of the swivel and other parts it appears that the design intent is to create a rigid attachment without freedom to rotate about the trailer wheels.

The hitch pin is a close fit. The close fit requires that the hole in the swivel have a vertical axis. The same is true for the holes in the c-channel. The 0.385 of gap is, IMHO, intended to allow for the tolerances in the height of the hitch swivel (when level) and the robot c channel (when level). I would not expect the swivel to ride much up and down on the pin. Of course I could be wrong and only testing will reveal how rigid the attachment is in practice.

I was thinking the same thing.

Also, I talked with the mentors today, and one of them brought up a really good point. If this was illegal, in theory, then suspension could be called illegal, since it changes the height of the robot, and could potentially make the hitch higher or lower than 2-13/16 inches from the ground. So I think we're sticking with our current plan. We're only going to have the height of the drive wheels and the wheels it will pivot onto on the front and back a 1/16 of an inch or less difference.

[Steve_Alaniz]

You are correct. This will work, however, the math proves you are not getting the full force of straight in line wheels and mechanums have a tendency to be difficult to steer straight plus a tendency for the operator to over correct. That's why so many teams experiment with a gyro feedback to help drive straight and side to side. The low friction of the surface means that the gyro would try to stop a skid but only manage to spin the wheels really fast... the exact opposite of what you need. Slower speeds seem to give more control but they're... slower. speed will be a factor in this game almost as much as friction.

Just my 2 cents

Steve

Originally Posted by Teamcap.2395

Okay, so here is a crazy thought I had...
If you did a 4WD in the same configuration as a omni-wheel drive...
...(like this) (In the same fashion as "The Pre" )
|= Chassis Perimeter
// = Wheels
. = space filler
___________
|//...........\\|
|................|
|................|
|................|
|\\...........//|

Would the low amount of friction on the wheels allow
the robot to move like an omni-drive?

I don't know how slick the flooring is because I was not able to attend a Kick-off.
But from what I have heard, it sounds pretty slippery.
Any guesses on how well this would work?

Edit: Hachiban beat me to it!

[Mr_I]

Purely anecdotal, but tonight we put our bot on regolith, and managed to steer without too much difficulty. It is a 4WD in standard ("portrait") orientation, with the difference being we incorporate a shorter wheelbase (to increase the space for the gullet). To use the now-standard way of drawing:

| . . . . . . . . . . . .|
| . . . . . . . . . . . .|
|_______________|
| [[. . . . . . . . .[[ |
| . . . . . . . . . . . .|
| . . . . . . . . . . . .|
| . . . . . . . . . . . .|
| . . . . . . . . . . . .|
| [[ . . . . . . . . [[ |
________________

While I can't say I've read every word of this thread (and especially not all the links off of it), I have to wonder what we did "wrong" to get it to steer. Could it be the shortened wheelbase?
(Yes, pictures coming)

[MrForbes]

The shortened wheelbase will make a big difference.

We got our new wood chassis rolling tonight

http://www.youtube.com/watch?v=Yb8IPQYp_M0

The concrete floor is pretty smooth, and it probably has a bit more friction than the proper material (which we need to buy soon!). Handling is pretty good...all things considered...but having it light, no trailer, wrong floor, everything will be different in the game.

[Paul Copioli]

Mr. I,

Did you have the trailer attached during your driving? We noticed a considerable difference with and without the trailer. Driving without the trailer was quite fun, but driving with the trailer was just painful.

Paul

[Mr_I]

Originally Posted by Paul Copioli

Mr. I,

Did you have the trailer attached during your driving? We noticed a considerable difference with and without the trailer. Driving without the trailer was quite fun, but driving with the trailer was just painful.

Paul

Good point. Tonight, maybe?

[mototom]

Originally Posted by dlavery

As one benchmark, after a bit of practice I was able to take a standard kit-bot system with trailer attached and run it from one end of the Crater (starting with the Trailer touching the Alliance Station Wall) to the far end, execute a 180-degree turn and make contact with the far wall, and run back to the starting wall in about 12 seconds. This was repeatable several times.

-dave

.

Wow Dave! That is some accelleration! Were you taking advantage of the carpet? What drive configuration were you using? Got a video?

[Brandon Holley]

If you have not hooked up a trailer to your robot, do it.

It may seem optimistic when you are driving just the robot, but as Paul said, it is just PAINFUL to drive with the trailer on.

[RobotDevil1985]

Originally Posted by Teamcap.2395

Okay, so here is a crazy thought I had...
If you did a 4WD in the same configuration as a omni-wheel drive...
...(like this) (In the same fashion as "The Pre" )
|= Chassis Perimeter
// = Wheels
. = space filler
___________
|//...........\\|
|................|
|................|
|................|
|\\...........//|

Would the low amount of friction on the wheels allow
the robot to move like an omni-drive?

I don't know how slick the flooring is because I was not able to attend a Kick-off.
But from what I have heard, it sounds pretty slippery.
Any guesses on how well this would work?

Edit: Hachiban beat me to it!

We tested that theory, granted we were on a smooth tile instead of the flooring, and found that without the ability to guarantee that all the wheels are on the floor with the same pressure. Otherwise the robot will veer uncontrolably depending on which wheels are in contact.

[jreuter]

"I have to wonder what we did "wrong" to get it to steer. Could it be the shortened wheelbase?"

The spreadsheet model (http://www.chiefdelphi.com/media/papers/1917) predicts that you should NOT be able to turn a "portrait" tank drive robot, if you use the published coefficients of friction. However, I've seen several posts which claim that the .14 lateral coefficient of friction is overstated. If you put lateral coefficients more like the in-line coefficients, the model computes that you will indeed be able to turn.

[Brandon Holley]

Originally Posted by jreuter

"I have to wonder what we did "wrong" to get it to steer. Could it be the shortened wheelbase?"

The spreadsheet model (http://www.chiefdelphi.com/media/papers/1917) predicts that you should NOT be able to turn a "portrait" tank drive robot, if you use the published coefficients of friction. However, I've seen several posts which claim that the .14 lateral coefficient of friction is overstated. If you put lateral coefficients more like the in-line coefficients, the model computes that you will indeed be able to turn.

The fact that other people have not produced the lateral coefficient of friction in their tests, and indeed have produced a smaller number...combined with the fact that your robot DOES turn would lead me to believe .14 is wrong.