Getting A Bot To Turn

[Andrew Blair]

Today I was brainstorming drive train ideas, but I was struck with a problem we encountered this year and never fixed really. The problem is, we have extremely grippy wheels on a 6-wheel layout. This works fine on a low bot, and it turns rather well, but with a tall, heavy mast like we have, we encounter a strange phenomena when we turn. The robot will begin to violently rock the more we turn. The things like a train when it's going straight, but when it turns quickly it's occasionally in danger of completely tipping. Any ideas for it or for a new 6 wheel bot with a high mast? thanks, this is an interesting problem.

[ahecht]

A) Don't use 6 wheels.

The friction equation says that on a flat surface, four wheels will have the same amount of traction as six. Each wheel will have a max static friction force equal to the coefficient of friction (which is the same for all wheels) times the downward force on the wheel (which is the robot weight divided by four). If the coefficient of friction is CoF and your robot weighs 130lb, for four wheels:
Total_Fsmax = 4[CoF * (130/4)] = 130*CoF

For 6 wheels:
Total_Fsmax = 4[CoF * (130/6)] = 130*CoF

B) Don't use grippy wheels on all six wheels.

If your robot turns about its center, there is very little sideways force on your middle wheels when turning. However, there is a great deal of side-force on outside wheels when turning, and your grippy rubber fighting that sideways force is causing your robot to rock. Put grippy wheels on the middle wheels and slick wheels [or even omni-wheels] on the outside four wheels. Even with four-wheeled robots, many teams will put grippy wheels on one set and slick wheels on the other set (with 190, if we want the robot to turn, we always put grips on the back wheels) to improve turning.

C) Don't make your wheels in a straight line.

Make your wheels slightly V-shaped, with the middle wheels lower than the outside wheels, or make the center wheels larger. Therefore, at any given time, only four wheels are on the ground, but you still have most of the advantages climbing stairs or other obstacles that six straight wheels would have (although you will rock forward and back quite a bit). Some of the Home Depot shopping carts work like this.

[Andrew Blair]

Originally Posted by ahecht

A) Don't use 6 wheels (the friction equation says that on a flat surface, four wheels will have the same amount of traction as six)

We have used the same wheels in a 4 wheel configuration, and have had even less success. Alot of the problem is that our robots are the long way(long wheelbase).

Perhaps something like this would help out a 4 wheel config. Hurricane

[David Guzman]

Definitely have your middle wheels lower a little bit. My team lowered the middle roller in our treads .25" and we could turn on our own foot print. The concept of 6 wheel is pretty much the same as treads. As far as turning goes that is.

Just my $0.02
Dave

[Holtzman]

I know exactly what your talking about. I saw number of 6 wheels that suffered from the same problems.

Question: Where you using the KIT Bot chassis in the 6 wheel configuration?

The kit bot chassis was notoriously bad for twisting under turning forces(sorry John). This twisting forces the front and rear corner wheels into the ground, giving you the center and rear wheel on the ground on one side, and the center and front wheel on the ground on the other side. This is what causes the oscillation your talking about. It is only compounded by the fact that you have a large weight up high. Reinforcing the chassis will help to minimize this twisting. A good example of this is 229's robot(Thanks John).

Its also very important that you balance your robot about the center wheel in a 6 wheel drive robot. Not balancing about the center wheel will also contribute to these oscillations.

[JVN]

The physics of drivetrain turning is all laid out quite nicely right here:

http://www.chiefdelphi.com/forums/pa...le&paperid=222
Courtesy of Chris Hibner. (I can never give this whitepaper enough props.)

Basically, you need to reduce the turning scrub.

The ways a typical 6WD system accomplishes this, is by reducing the wheel-base in the fwd-back direction at any given time.

Think about an example where your robot was half as long; it would turn much better. Since we can't permanently reduce this wheel base, (it reduces robot stability in the fwd-back direction) a 6WD solution with a lowered middle wheel rocks back and forth between 2 of these 1/2-long wheel-bases.

To understand how this shorter wheel-base increases turning, again just look to the physics of it.

I have taken one example, there are many others. All of them, relate directly back to the "turning-physics". If you understand that, you will find 10 ways to solve your problem.

Your next steps should be:

• Read the whitepaper.
• If you need help understanding what is in that whitepaper: ask for help understanding it.

Hopefully this will be a thread with engineering contributions to the discussion, and not another one where people simply spout garbage like "4 wheel drives dont turn ever u shud go wit a swerve".

-JV

[Rickertsen2]

I have seen this rocking before. If you search the forums, you will find many discussions about getting bots to turn.

[sanddrag]

Perfecting a robot to turn well is one of the greatest challenges in FRC robot building. We have been in it for 5 years now and still haven't quite perfected it. I'd say the best handling robot we've had so far was wider than it was long and had four wheels. It had Skyway 8x2 beadlok wheels (very short wheelbase and wide track) with the foam rubber tires which were turned down to smooth/flat on a lathe. It was very easy to keep straight but turned with authority any time you wanted it too. The wheels were pretty high traction on the carpet and it still drew a good amount of current while turning, but it was a very well handling machine. It is also important to mention that this robot had a very low CG and was only 13" tall total.

I think the one of the best configurations for turning and handling is to have a robot whose track width is significantly more than the wheelbase. However, if the robot's CG gets too high, or you have to climb a stair of something, it might not be possible to have it wider than it is long, especially with the 28" base dimension that was new this year.

So, getting a long skinny robot to turn is the real challenge. And this is where the craze over six wheels came from. We jumped on the bandwagon and it is good, but it still isn't perfect. I like the wide and short four wheel setup better. We had the weight over the back four most of the time so it tracked pretty straight but turning still wasn't superb with our super grippy McMaster tread. On the other hand, team 968 had a long 6 wheeled robot but they had it well balanced over the center wheels. It could spin in any direction at any time with minimal effort and great ease, but if you wanted to keep it in a straight line, it's not easy, it takes some skill.

One thing I saw done on a long robot was six wheels all touching the ground but the back ones were very close together and the fronts were non driven omnis. This way, it is as if your wheelbase was very short for the purposes of turning, but long for the purpose of stability. I didn't watch it too closely in competition but from what I saw it looked like a fairly effective setup.

I'm thinking 4wd tank on a wide bot can perform to meet a standard which I call perfect, but I still don't know what the "right" answer is for turning long FRC bot. 6wd with the center ones lower comes close, but it's still not perfect.

[JVN]

Originally Posted by sanddrag

It could spin in any direction at any time with minimal effort and great ease, but if you wanted to keep it in a straight line, it's not easy, it takes some skill.

.

Turning & driving in a straight line are based off the same thing; turning scrub.

This may be news to some people, but:
If it is hard for the drivetrain to turn, it will drive straight really well.
If it turns really well, it is tough to keep it driving straight.

On 229, we built a drivetrain that turned really well, then taught our drivers to control it.
The more scrub you have, the more load it places on the motors during turning, the higher the current these motors draw.

I like efficiency, and low motor-load; the kids can handle keeping it moving straight, they're sharp kids.

[Karthik]

Originally Posted by JVN

The physics of drivetrain turning is all laid out quite nicely right here:

http://www.chiefdelphi.com/forums/pa...le&paperid=222
Courtesy of Chris Hibner. (I can never give this whitepaper enough props.)

....

I have taken one example, there are many others. All of them, relate directly back to the "turning-physics". If you understand that, you will find 10 ways to solve your problem.

Your next steps should be:

• Read the whitepaper.
• If you need help understanding what is in that whitepaper: ask for help understanding it.

I can't over emphasize how much I agree with John's statement. Chris's whitepaper should be required pre-season reading for anyone who's going to be working on a FIRST drivetrain this year. All beliefs and myths about robot turning can be shown or disproved with this paper. Once you read this paper, your design process will be simplified greatly.

In fact, I recommend as an exercise that you try applying these calculations to your 2005 drivetrains. The lessons will sink in much more easily that way.

Basically, read the paper, do some calculations, and you'll learn how to turn.

And they say math isn't fun...

[sciguy125]

I couldn't come up with any pictures, but our 6wd performed much much better than expected. There was no noticable jumping as our previous drivetrain (4wd with pneumatic wheels). We also couldn't believe how smooth it drove. We had solid wheels with SBR rubber as traction material. It was also extraordinarily low. I think that the clearance between the bottom of the chassis and the ground was just over 1/4". There wasn't enough room to get your fingers under it.

We were considering lowering the middle wheels as most everyone else does, but we didn't want constant rocking. We decided to raise the front wheels. If your weight is distributed correctly, this makes it so that the robot stays level under normal conditions. If your weight happens to shift forward (arm sticking out or picking something up), you start resting on the front set of wheels. The problem with lowering the middle wheels is that you are never level (unless you're balanced on the center wheels).

I realize that you wanted input for a tall robot, but I think this would still work. As long as the chassis itself is stable, I have a feeling that it'll keep the rest of the robot stable. (Or at least more stable than it otherwise would be)

[JVN]

Originally Posted by sciguy125

I realize that you wanted input for a tall robot, but I think this would still work. As long as the chassis itself is stable, I have a feeling that it'll keep the rest of the robot stable. (Or at least more stable than it otherwise would be)

The tall robot just amplifies things.
Think of it as an inverted pendulum. Any small oscillation caused by "scrub hops" will be greatly amplified by the high-mass.

The solution to this, is to:
-minimize the drivetrain scrub
-keep the CG as low as possible

John

[Andrew Blair]

Originally Posted by Holtzman

I know exactly what your talking about. I saw number of 6 wheels that suffered from the same problems.

Question: Where you using the KIT Bot chassis in the 6 wheel configuration?

The kit bot chassis was notoriously bad for twisting under turning forces(sorry John). This twisting causes the front and rear corner wheels into the ground, giving you the center and rear wheel on the ground on one side, and the center and front wheel on the ground on the other side. This is what causes the oscillation your talking about. It is only compounded by the fact that you have a large weight up high. Reinforcing the chassis will help to minimize this twisting. A good example of this is 229's robot(Thanks John). It is only compounded by the fact that you have a large weight up high.

Its also very important that you balance your robot about the center wheel in a 6 wheel drive robot. Not balancing about the center wheel will also contribute to these oscillations.

This is probably a big cause of our problem. Our kit frame is modified by removing the front brace. We reinforced it, but the front drive wheel is still beyond the brace, and the twist is noticeable. I had never really thought of this as the cause of the problem. Thanks for the whitepaper John! I had been searching, but I guess for a six wheel, not a general drivetrain, solution.

[Evan Austin]

Here is another question:
I know that some teams with 4wd robots will use 2 standard wheels on one end and use 2 omni wheels on the other end to enable them to steer easily. I wonder if anybody is using 6wd has done something similar with the 2 standard center wheels and omni wheels on all 4 corners? I know of one team which used a 2wd center drive and had small omni wheels on all corners, and they were very maneuverable and stable. However, they were not powering these wheels on the corners. If this would work, you would not have to lower the center wheel, thus keeping a flat footprint.

[Tristan Lall]

Originally Posted by Evan Austin

Here is another question:
I know that some teams with 4wd robots will use 2 standard wheels on one end and use 2 omni wheels on the other end to enable them to steer easily. I wonder if anybody is using 6wd has done something similar with the 2 standard center wheels and omni wheels on all 4 corners? I know of one team which used a 2wd center drive and had small omni wheels on all corners, and they were very maneuverable and stable. However, they were not powering these wheels on the corners. If this would work, you would not have to lower the center wheel, thus keeping a flat footprint.

This could work, however, you might as well power all of the wheels in contact with the ground. You're wasting normal force if you don't (recall that static friction is a function of normal force, and that the pushing ability of a robot is based on the friction of the driving wheels on the playing surface).

If you go this route, also be aware that your robot will rotate more easily when it experiences a sideways force; like John pointed out, that can be a good thing or a bad thing, depending on your driving style, the needs of the autonomous mode, and how often you expect to be pushed from the side. Also, consider that 3 points on a surface define a plane; 4 is overconstrained, 6 is very overconstrained. Consider how you will provide for keeping many drive wheels in contact with the ground at any given time—if you don't, you're back to the problem of wasting normal force. (A little frame flexibility, and/or the shape of the omniwheels will often conceal this in a static state, but when turning, beware of the moments acting on the frame, which can twist it, and perhaps raise a corner off of the ground.)