Crab Drive Questions

[craigboez]

Hello All

I'm a rookie mentor on a rookie team, and we're in the process of trying to determine which drivetrain will the best while also being feasible for us to implement. We have a kitbot that we've experimented with, and after yesterday it was driving around with skid steer on Regolith with less than spectacular results. This test has prompted me to look into a swerve or crab steer setup, but I'd like to ask for some veteran leadership before going off in that direction. Some questions I have are:

1. Has anyone tested a previous crab drive bot on Regolith with the new wheels and trailer? Are there significant improvements to manueverability vs a skid/tank steer drivetrain?

2. Is it within the realm of possibilities for a rookie team to implement a crab drive robot?

3. Is there a whitepaper on how to build a crab steer? I've looked, but come up empty.

4. Does anyone sell a "crab module", or are those all custom built?

I don't want to embark on a journey that is destined for failure, but I also don't want to "settle" for a skid steer bot if a crab drive is going to be significantly better.

Any input would be greatly appreciated. Thanks.

[Sunshine]

For what it is worth........

We are a three year team and we researched crab drive for a week. We decided that the risks outweighed the rewards. I personally researched this for about 8 hours and found "conceptual" white papers only. You may have better results than me, but I found no practical designs or drawings that I could use.

We decided that it was better to leave this as a summer experiment activity for future consideration.

But this is because of our limited knowledge. Our neighboring team is in their second year and have decided to go with it because they have an engineer who feels comfortable with designing it.

[Joe G.]

1: Wildstang had a video of their 2008 crab with a dummy trailer up, but it got taken down for some reason. It was unclear, however, whether they were "crabbing" in the video. It seemed to handle relatively well, but not particularly better than some of the better looking 4wd robots posted so far. Perhaps someone from wildstang can give us more information about this test.

2. It depends on your resouces. However, crab takes a LOT of work, and nearly all teams that use crab prototype their crab design in the offseason. I cannot say that I reccomend it, if no serious thought to your crab design has been given until now.

3. Haven't seen one either. I would recommend searching the images on chiefdelphi for crab/swerve drives, and contacting the designers for more information.

Some teams to check out for crab designs:

71
111
118
1625
973
Miscellaneous CADs by CraigHickman

4: All custom machined. Precision tools are usually required to make these parts, which if you don't have access to, would throw a wrench into plans for a crab.



Finally, there have been several videos of skid steer robots posted, many of which seem to be moving quite successfully. 1629's "wide orientation" video posted today particularly stands out.

http://www.chiefdelphi.com/forums/sh...threadid=71861

The bottom line is, unless you are sure you can design and build a successful crab in the next few weeks, I would look more into variations of a skid steer.

[Ryan Dognaux]

Maybe you can post what exactly your team did with the kit frame & drive to experience the less than stellar results? Were all four wheels being driven? Did you have your maximum weight simulated with a trailer?

If you just assembled the kit frame and gearboxes and threw the electronics on it, I could see why it didn't perform so well.

Also, did you have it set up with the 28" end as the front of your robot or the 38" end? You will have an easier time turning this year with the wide chassis setup.

Crab drives are hard on any normal year with a normal playing field surface, and this year is even more difficult. I'd stay away from it as a rookie team and focus on your ball delivery mechanisms.

[GUI]

1. There would certainly be some increase in maneuverability with a crab drive, but I'm not sure how much. I believe 111 tested their 08 bot with the new wheels and playing surface, maybe one of them could comment on how useful the extra movement was.

2. It is certainly possible for a rookie team to implement any style drivebase or mechanism. That being said, I would not attempt this without building a prototype in the off season, unless you have significant resources (both time and manufacture) to dedicate to your drivebase.

3. I don't know of any particular whitepapers, I suspect that there might not be one simply because of the inherint uniqueness and diversity in crab drives.

4. As far as I know, any team that has used crab drive in FRC custom made everything. You may be able to find off the shelf parts that would significantly simplify making modules, but it all depends on how you would design your particular chassis.

I don't want to discourage you from attempting this, but realize that a crab drive is more complicated, expensive, and difficult to build than skid steer. However, if you have the resources and motivation, it's definitely possible.

[seraphim33]

our team is looking into this as a possibility we have just run into a snag at trying to find out how to build it

[Ian Curtis]

Originally Posted by craigboez

1. Has anyone tested a previous crab drive bot on Regolith with the new wheels and trailer? Are there significant improvements to manueverability vs a skid/tank steer drivetrain?

2. Is it within the realm of possibilities for a rookie team to implement a crab drive robot?

3. Is there a whitepaper on how to build a crab steer? I've looked, but come up empty.

4. Does anyone sell a "crab module", or are those all custom built?

1. Never built a crab drive, but look at 1629's drive tests with a 4wd drive base. They have just as much control on Regolith as we've had on carpet previously. So, it's definitely possible (and much much easier!) to build a 4wd system with pretty darn good control.

2. Depends on the team. If you've got a few engineers, probably. If you don't have much engineering support, it's still possible, but it will take a lot of your time, which would probably be better spent building a ball collector of some sort.

3. I've seen the CAD files of 148's 2008 crab drive on Chief Delphi. I don't remember where, but they're attached to a post somewhere, if you want a starting point.

4. To my knowledge all crab drives are custom.

Have you considered something less than a full crab drive? It would be far easier to build 1 or 2 rotating modules, which would give you some of the more precise control a full crab drive could provide.

Hope that helps!

[craigboez]

Thanks for the replies. They're not so much discouraging of a crab drive as they are encouraging for a skid steer bot. The video for 1629 is definitely better than the results we were getting, so we will need to continue development on our kit bot.

Another added benefit I can see to a skid steer is that there aren't any required chains or cables traversing across the bot, which would potentially interfere with a ball collection effort. We'll put crab drive on the shelf for now, and revisit in the offseason if we're bored.

Thanks.

[gorrilla]

its really not that hard to build a crab system. all you need is a few hours and some dedicated people....

Originally Posted by rocketperson44

Some teams to check out for crab designs:

71
111
118
1625
973
Miscellaneous CADs by CraigHickman

4: All custom machined. Precision tools are usually required to make these parts, which if you don't have access to, would throw a wrench into plans for a crab.

Hey cool, I got mentioned! Here's the deal with crab: This year is a wonderful year for it. if you drive each wheel independently, you can do some pretty awesome things with traction control. In addition, your module can be pretty minimalistic due to the traction. Your robot will slide before most modules will bend, which allows for some pretty simple designs. If you've got access to a Manual Mill, I can give you some tips on how to build a pretty simple drab system.

Also, someone mentioned just doing 2 wheel crab. This is a decent idea, but consider this instead: If 4 modules are out of your reach, 2 will be also. The machining difference between 2 and 4 is minimal; most of the work is in the setup. Instead, why not go with an Ackerman style steering on the front, which a Differential in the back? With some nifty programming (which we have more than enough power to do), it should be pretty simple to get some Yaw control going, much akin to the current Subaru STI and the Mitsubishi Evo. Power your rear wheels at differing levels while steering your front, and trust me, you'll be impressed at the performance you'll be able to achieve.

[gorrilla]

ive been thinking about slip differentials too, i think its a good idea,

anyone ever thought about having ackerman steering in the front and back?
only problem is finding a way to power them and have steering(i forget what the parts called that is used on front wheel drive cars?)



(EDIT) a CV joint, is what i meant, dont know how you would go about making one

[Chris Fultz]

If you talk to some of the teams that have done good crab drives, most will tell you it takes a while to get it to work well. There are the mechanical aspects of the system, the software aspects, and the driver aspects.

In my view, this would fall into the category of a summer / fall development program, learning, working out the details, and continuing to improve, and then incorporating the design ideas into a competition robot.

With so many things to focus on as a new team, i would carefully do a risk / benefit analysis of undertaking this type of drive. Only you and your team can determine if you have the ability and resources to do this.

Whatever way you go, best of luck in your FIRST season!

[joeweber]

Instead of going to complete crab drive how about just steering left 90 degrees and right 90 degrees. You are limited with your movement any way with the trailer you are dragging around. It would be easy to build the kit gear box with the drive attached directly to the wheel and rotating the assembly. You can use last years gear boxes to add to the two you have so that you will have three or four drives to turn.

Edit: sorry your a rookie team, you will have to purchase the extra gear boxes.

I made this post in another thread; it's relevant here too.

I'm seeing this response a lot in relation to swerves. Say we go with a 118 style crab: All wheels steered together. Even with the type, the wheels don't suddenly make a 90 degree turn, or however much people assume. The robot will (assuming that it's driving) turn in an arc. The cleanness of this arc is a factor, but not enough to disqualify the design. Unless you're using some odd turning system, the wheels won't change direction faster than the robot can adjust.

For the sake of the point, now let's move to sets of 2. The front wheels are now steered separate of the rear. With a bit of programming, and a big red button, you now have the option of an AWD 4 wheel steering system or a standard crab. Say we drive each wheel independently, this opens up even MORE options. We can drive normal crab, normal crab with traction control (I mean, seriously, the Jaguars have built in current sensors! It's not THAT hard to do!), 4 wheel steering, 4 wheel steering with slip control, 4 wheel steering with Yaw control (see the Modern Subaru STI and Mitsubishi EVO for this), and more. Plus, at the hit of a button, we can move (in a short amount of time) from moving forward and steering to side strafing. We've got traction control already, so what's to stop you from using a gyro to even out the direction of the bot? Sure the trailer's going to weigh you down, but a bit of time spent in testing, and a decent programmer makes that problem nonexistent.

Personally, I don't see any other drive system that has more available to it than a 4 wheel crab with independent power. You can do almost ANY form of driving you need.

[MrForbes]

Originally Posted by joeweber

Instead of going to complete crab drive how about just steering left 90 degrees and right 90 degrees.

For an example of something like this, see the pictures in the CD-Media photo gallery of team 842s robot Carmen. The front wheels steer and have motors and transmissions that steer with the wheels, the rear wheels are driven by another pair of motors and transmissions and do not steer (but will have software differential speed control to help the robot steer).

http://www.chiefdelphi.com/media/photos/32275 for example...this was built in a few days in a small, not very well equipped shop, by a bunch of clever, dedicated people with no mechanical engineering mentors.