4WD Turning Difficulties

[ZakuAce]

Originally Posted by SWIM

I believe the FIRST colloquial term would be "swerve drive". edit: apparently not

The major obstacle in your way is probably going to be replacing the chain-drive with shafts and gears. Also, CV joints would likely be involved. Making the steering rack it's self wouldn't be too bad, but the whole system would add a couple layers of complexity to what needs to be the most reliable part of your robot.

While turning, you'd be able to make more efficient use of your traction, and it would be far more controllable at speed. Whether that's worth the extra complexity, weight, and motors depends on what your team finds important in a drivetrain.

edit: come to think of it, you could use a single CIM to drive all four wheels, and a CIM to power the steering rack. that could potentially be a pretty good solution, but you'd need two or three limited-slip differentials for it to handle properly, and I think andymark is fresh out of those...

Would it be possible to do the car steering with half a swerve drive (the front wheels), then have the rear wheels powered normally?

Also if you wanted to put more weight in the back to reduce the moment the trailer creates, would it not then be a good idea to use rear wheel steering?

[SWIM]

Originally Posted by ZakuAce

Would it be possible to do the car steering with half a swerve drive (the front wheels), then have the rear wheels powered normally?

Also if you wanted to put more weight in the back to reduce the moment the trailer creates, would it not then be a good idea to use rear wheel steering?

If you're referring to what I thought was "swerve drive", but is appearantly known as "ackermann", then

http://en.wikipedia.org/wiki/Ackerma...ering_geometry

The image on that page should give you a good idea of how steering in a car works. The yellow bar moves back and forth using a rack and pinion (hence, steering rack)

http://en.wikipedia.org/wiki/Rack_and_pinion

If you want to have a 4WD drivetrain with car-style steering, you're going to have to have a way for the shafts to the front transmit power while steering, probably through a simple CV joint

http://en.wikipedia.org/wiki/CV_joint

Alternately, you package a wheel and motor assembly together, and steer the whole thing. You'd need 3 motors for that type of drive, though.

Whatever drivetrain you pick, if you put the steering where more weight is, you'll have more available traction to steer with, and have less of a tendancy to understeer. Be careful if you put a steering rack on the rear wheels though, I believe the steering won't try and center it's self like it does in a front wheel steering system, although if it's electronicly controlled, it shouldn't matter.

An interesting way to power a robot with wheels that can steer would be to have the motor in the center of the car, with shafts going to the front and rear which then split to the left and right sides. To distribute the torque at 90* angles, you could have simple locked ring and pinion setups and live with a little wheelspin on cornering, or go with a differential setup, either electrically or mechanically controlled.

edit: seriously consider a traditional 4WD skid steer setup first, though. if your COG is going to be shifted significantly to the rear, then a conventional drivetrain will likely be perfectly adequate. run the numbers through that spreadsheet with coefficents of friction being .12 in every direction, and you'll find that a square wheelbase with a rear-based COG is going to turn easily, without resorting to complex steering systems

[Racer26]

I'm of the opinion that any ackermann based steering system will experience major understeer and oversteer. I don't really know what the answer is as I think there isnt any traditional FRC drive system (that can be done with the rover wheels) that is particularly good, and I think thats the point.

[SWIM]

Originally Posted by 1075guy

I'm of the opinion that any ackermann based steering system will experience major understeer and oversteer. I don't really know what the answer is as I think there isnt any traditional FRC drive system (that can be done with the rover wheels) that is particularly good, and I think thats the point.

With this little traction, any drive system is going to have issues with under- or over-steer, right? Skid drive is going to slide around as well.

[Racer26]

Absolutely it will. I don't think anything will be particularly good, but I know that cars on ice dont steer worth crap, and tanks steer better than cars, so, by extension I would think similar would apply to FRC bots.

[SWIM]

Originally Posted by 1075guy

Absolutely it will. I don't think anything will be particularly good, but I know that cars on ice dont steer worth crap, and tanks steer better than cars, so, by extension I would think similar would apply to FRC bots.

But what happens when you put the same tires on the car and the tank?

[Greg Peshek]

Originally Posted by SWIM

But what happens when you put the same tires on the car and the tank?

Being on a team that built an Ackerman steering robot for last year's game (see: http://www.chiefdelphi.com/media/photos/29780) and as a drivetrain guy myself, I can say that the steering robot was a very, very high traction system. In the picture you can see that we used dualie 10" rubber wheels on the differential and 8" rubber wheels in the front. It needed a lot of traction to work.

We used those same 8" wheels on a tank drive in the Aim High game, and the robot had entirely too much traction and essentially "danced" while trying to turn (we have since learned..) So if you extrapolate you can conclude that the high traction environment was needed on the car steering, while completely overkill on the tank drive.

I can't conclusively say that the Ackermann steering requires significantly more traction to work, but from my experience it seems as if it does - and I would have a hard time seeing it working with a lack of traction on the front steering wheels. Take it as you may.

-Greg

[spectre107]

That sounds simillar to an idea I had about traction control. In cars its called ASR(Anti-Slip Reduction), the controller would have a set of known speeds that would equal to a set of rpms. For Example:

5fps=50rpm
10fps=100rpm
15fps=150rpm

Now while the robot is driving on the slick surface, the wheels start to slip and spin out(the rpms increase above the set known speeds)

5fps is not equal to 300rpms acorrding to the controller.

Therefore, the controller would reduce power to the motors so that the speed(fps) and revolutions of the wheel (rpm) meet the rules set by the knowns. This system in effect doesn't increase traction but keeps the wheels from spining out.

[Collin Fultz]

Originally Posted by Joe Johnson

In the past you saw robots jumping as they turned corners, drive motors that cut out from the circuit breakers tripping and sometimes smoke pouring out of the motors and you knew, ah, something is less than optimal there. This year there will be none of that. Teams will think, my robot is not hopping as it turns, my motors are not tripping the breakers or even getting hot... ...everything is right with the world.

No such luck.

I can attest to this in an empirical setting. Last night we ran four types of robots: Light (100 lbs or so) 6WD, Heavy (150 lbs) 6WD, Light 4WD (long & skinny), and Heavy 4WD (long & skinny)

The 6WD turned with relative ease (relative being the key word). It would make zero-turning-radius turns (just spinning around) and arcing turns. Of course, it slid around a bit, but not as much as we thought it would.

The 4WD also turned (which kind of suprised me, based on the math), but looking closer at the wheels as it turned, it was actually skipping across the surface, not sliding like the 6WD had. Our driver (3rd year, I believe) also mentioned it was harder for him to control the turn with the 4WD set up and it was turning slower, which confirmed that it was a skip-turn and not a true skid-turn like the 6WD.

We (234) have a lot of experience with the "dancing robot". Our 2003 robot looked like it was doing the jitterbug while trying to turn, which was scary the first time we saw it, but we learned to drive with it...we just had to take our time.

[Josh Drake]

Did you have the trailer hooked up too?

[MrForbes]

Originally Posted by Collin Fultz

I can attest to this in an empirical setting. Last night we ran four types of robots: Light (100 lbs or so) 6WD, Heavy (150 lbs) 6WD, Light 4WD (long & skinny), and Heavy 4WD (long & skinny)

Awww...you didn't do the one we're interested in, the 4 WD short & wide

[=Martin=Taylor=]

Originally Posted by squirrel

Awww...you didn't do the one we're interested in, the 4 WD short & wide

Yeah I'm disappointed too...

It would seem that the 4 WD "Wide Configuration" would be the closest thing to the "Ideal Configuration" which would be to have the trailer at the center of your turning radius, so that it would produce no torque on the system.

[Collin Fultz]

Originally Posted by squirrel

Awww...you didn't do the one we're interested in, the 4 WD short & wide

I know, sorry. That's the one I'm really interested in, too. But, we had a 6WD chassis available for testing, so that's what we used. Hopefully, we can use the kit-bot chassis to make a wide robot and see how it handles.

We did not have a trailer available. We are making that today. Hopefully we'll have a video available sometime soon.

[ZakuAce]

Originally Posted by spectre107

That sounds simillar to an idea I had about traction control. In cars its called ASR(Anti-Slip Reduction), the controller would have a set of known speeds that would equal to a set of rpms. For Example:

5fps=50rpm
10fps=100rpm
15fps=150rpm

Now while the robot is driving on the slick surface, the wheels start to slip and spin out(the rpms increase above the set known speeds)

5fps is not equal to 300rpms acorrding to the controller.

Therefore, the controller would reduce power to the motors so that the speed(fps) and revolutions of the wheel (rpm) meet the rules set by the knowns. This system in effect doesn't increase traction but keeps the wheels from spining out.

On this surface, I think you will spin out no matter what you do. If you don't spin out, you will still slide around.

Originally Posted by Greg Peshek

I can't conclusively say that the Ackermann steering requires significantly more traction to work, but from my experience it seems as if it does - and I would have a hard time seeing it working with a lack of traction on the front steering wheels. Take it as you may.

-Greg

Think you can put this year's wheels on your robot and give us a demo?

[Dick Linn]

Could you use a 2 WD hooked to a sulky and rigidly couple the trailer to that?