Team 973 presents our offseason prototype; King Krab.

-4 Wheel Crab Drive, opposite pairs of wheels driven and steered (AB and CD driven, AD and BC steered).
-Powered by 4 CIMs through AM Shifter Guts, 6/16 fps and traction limited in low.
-Steered by dual globes.
-4" diameter 1.5" Wide wheels with nitrile roughtop.
-Bevel gears and absolute encoders allowed unlimited module rotation with the same sensor reading for actual wheel angle.
-Currently would be 54 pounds with no electronics, but an electronics board. By switching to some aluminum and lightened hubbed sprockets, and turning down the bevel gears we believe we can save another 1.5 pounds.
-#10 bolt holes on a 1/2" hole pattern along the entire frame allows the gearboxes to be mounted anyway to work around manipulators. Also came to be very convenient for mounting items. We now have a jig with drill bushings to easily and accurately do it on a drill press.
-All chains are tensioned by a bearing block pulled by a bolt (similar to how some west coast drives have been tensioned), this also houses the sensor for each chain loop.
-Module itself rotates in a 1" bronze bushing and two 1/16" bronze thrust bearings, all other rotation is with ball bearings.

Thanks for any comments and feedback.

Frame Pic
Module Pic


now drive it for the rest of the offseason nonstop, and you might be ready.
The most important thing we had to figure out with ours was making sure the drive knew how the thing would react to inputs, such as which route the wheels would take.

it reminds me very much of ours…lol but two sets of steered too

:eek: AWESOME.

Being the curious critter that I am, care to answer a few questions?
-Module weight?
-Have any other closeups that I can ogle?

This is an awesome drivetrain so far. I too would love to see some detail shots of the modules and the entire thing both to admire and to be able to give you advice on improvements.

An overall isometric view and a module view have been uploaded to CD-media.

The main improvement we would make for future iterations is custom cutting sprockets, turning down the bevel gears to save weight, switching to hex shafts (they are currently 3/8" 7075 with 1/8" keyway, wheel sits on a 3/8" 7075 dead axle), adding a second steering chain “tensioner” between the two modules on the outer edge to align the two modules angularly (they are just so slightly out of alignment) and add an adjustable rubber pad to add friction to the steering chain to ease the control loops tuning.

Also, we’d probably switch to heavily pocketed aluminum gearbox plates rather than the delrin. So far the flex has been minimal and hasn’t caused any issues (although I suspect the compression from it makes the gearbox slightly less efficient), but I’m concerned of how it will change over time.

Could you give me the part number of the bevel gears and where you purchased them? Also could you post more pictures of the individual module.

It’s a great design that you guys have come up with, I’m glad to see you finally got it running! I especially like the configuration of your swerve drive, with the two driven pairs being different from the steered pairs. You should be able to make it do some pretty awesome maneuvers.

Now you just gotta figure out how to make it get up a step, and it will be perfect for this coming game! :rolleyes:

No problem :wink:

Neato Design. I’ve always been fascinated by oddball drive systems like crab… looks like a very slick implementation. :]

But… am I just crazy, or is there no way for this drivetrain to spin in place without skid-steering?


There are very few crab/swerve designs in FIRST that can spin in place without skid-steering. The only way to do it really is to chain the wheels in an X pattern, which is slightly more difficult than the traditional row orientation.

Thanks for the kind words everyone, it really means a lot to us.

Craig, The module is 1.9 pounds as is. A 1" wheel significantly reduces the weight on top of that. The module was originally 1.1 pounds, but switching to 1.5" wheels, larger bevel gears and increasing the sideplates material all boosted it up a lot. I’m not sure, but between replacing the steel sprocket and turning down the bevel gears (they have BEEFY hubs) we can save a good deal there. The wheel could also be made a bit lighter, as it is currently .43 lbs for a 1.5".

The top of the 1x1 is 7.7ish inches, the entire frame is 6" tall (4" between levels) and we have 1.7ish inches of ground clearance. The frame itself could probably be made lighter, as it is all 1x1x1/8" except for the 4 inner cross members on the top (and the 4 short pieces of 2x1) and weighs 15.3 lbs + welds. It’s currently not amazing at climbing ramps, but we drew some varied lower levels that should allow it to climb great. The nice part about this style module design is almost the entire front and rear face of the wheel is open assuming there is no frame blocking it.

I didn’t mention before, but with the standard #10 pattern and the fact that the distance between inner and outer frame members is a constant 5", not only can gearboxes be slid up and down a side like I said before, but we can swap the locations of drive and steering gearboxes.

sdcantrell56, I’ll tell you we used 12DP miter gears, but won’t tell what tooth count or part number. I’m not trying to be secretive, but everyone I talked to about them warned me how important it is to make sure they are strong enough. We did a lot of calculations for it (only possible with the help of some great people) and still aren’t sure they’re strong enough, as we haven’t started full speed, full weight, robot to robot interaction. As you know, the wheel size, tread, gearing, robot weight, etc. all effect the load on it, so you can’t really say “it worked for 973… it’ll work for YYY”.

What kind of views would you like of the module? The posted pic shows pretty much all it’s features.

The way “opposite” pairs are steered/driven it is only skid steer.

However, A difference in drive speed also causes the robot to spin unless the wheels are exactly aligned at 0 or 180 degrees (which is 90 degrees off from our skid-steer zone)

We have ideas that make this not really an issue, one is to implement “drift” buttons that slow or kill one pair of drive motors on a curve to cause the robot to spin on center while driving (credit to 1625). We imagine most rotating will be done by this, or by a 2nd stick at the point of scoring. We’re working on implementing anglular presets with a gyro as well. I imagine we may try some gyro based turning while driving based on differing drive speeds.

Eh, didn’t entirely answer your question, but that’s our current gameplan for rotation.

Finally, after my summer hibernation, I have returned back to this world… Adam, a “King Krab” is a crab in which each wheel is independently steered.

Pavan Davé

It’s just a name Pavan, keep it up and I’ll confiscate that thingie over the e.

i absolutely love that robot
(still looking for videos of it, if anybody happens to have any)

looks really solid so far…keep posting videos of your progress on the control aspect.


Ok so as a follow up question. Could you outline how you did the calculations to find out what gears would work? Also where did you purchase the miter gears?

My apologies for these questions, but I don’t have access to YouTube at work, so all I have to go from is the CAD models (excellent work as always, btw)…

  • Are the connections, including those within each wheel module, chain or belts? You probably just used simplified models, but I wanted to be sure.
  • What size chain/belt are you using for each location?

We bought them from mcmaster, and the calculations were based on the modified Lewis Bending Numbers Equations for Miter gears.

Chains, we just use a simplified model based on sprocket pitch diameter and chain height/width to check clearances. The models are actually based on the output of Dr. Joe’s Chain Path visualizer (which we slightly modified to calculate the weight of the chain based on length as well, so we added that as a custom property so the chain weight is accurate).

We use 25 for the steering loops and for within the modules and #35 for the drive. Technically we could use 25 for the main drive loops and It’d be strong enough, but a single drivechain failure renders the robot immobile. We’ve considered two parallel runs of 25, which would still be lighter than the #35.

McMaster part #: 6529K14
that bevel gear has worked flawlessly for us, we’ve never destroyed one or lost any teeth, they are pretty durable in that regard. You can also turn down the hubs on them to save some weight.

the track record this year is, chicago, milwaukee, colorado, atlanta, and IRI, and offseason demos, and there still lookin good.

Thank you very much aren. Thats what I wanted to know. It’s a shame they’re so expensive there. We need to find a cheaper source.