pic: Coaxial Swerve Drive Module with 2-speed Ball Drive and Nitrile Tread

Iso View of Pwnage Team 2451 Coaxial Swerve Drive
Highlights: 7 lbs, 3" Nitrile Tires, 2-speed Ball Drive, Absolute Encoder for steering angle, Incremental Encoder for wheel speed
4.9 feet per second actual low speed/6.01 fps theoretical
15.7 feet per second actual high speed/19.12 fps theoretical

Fancy. How much does each module weigh?

Great-looking design. With just 1 CIM per wheel, though, isn’t that top speed a bit high, unless you’re planning to use autoshift code?

What is the benefits in running this one over the last design? 2 speed shifting is nice but you have decreed mechanical efficacy and a higher center of gravity.

Still like the design though really smart roundabout thinking.

7lb 3" it says it in the post.

Are you planning on building a drive with these before the season?
I really like the bevel gear embedded in the drive wheel eliminating an extra stage.
What kind of module rotation speeds are you looking at (rps)?
Did you design your own ball shifter? From the cutaway it looks different then the VexPro one.

I think we would all appreciate you posting the CAD or a few more views, you can hardly see the rotation motor.
Very nice work!

Complete CAD file can be downloaded here.


After our successful 1st year using our “In wheel” swerve we learned a few things and applied those lessons to our 2nd swerve design.

-Swerves can usually out maneuver other robots when there is an open field
-A single speed swerve is not good in a scrum with multiple defending robots
-Slip rings add weight, complexity, point of failure and the legality of the best Mercotac slip rings is being debated currently
-Colson tires get good traction and have a low rolling resistance but are more traction limited than Nitrile treaded tires

Therefore we developed a 2-speed coaxial swerve so we have speed and the ability to get out of a jam. We switched from Colson wheels to custom aluminum nitrile treaded wheels to gain added traction. We reduced the weight per module by 1lb each and the area is 33% of the “In Wheel” swerve.

Looks pretty great, I would have personally repackaged it similar to 1717 and sacrificed some of the bellypan area just to get that CIM out of the air and lower the COG but I really like this evolution of swerve. When are we going to see a manufactured unit?

This module is approximately 7lbs.
A little under in Solidworks, a little over with fasteners in the real world.

My apologies, I stated the wrong speeds. The speeds listed on the description are for a 3.25" wheel, with a 3" wheel the speeds would be around 4.55/5.55 fps in low and 14.47/17.65 fps in high (actual speed/theoretical).
It was easier to chance the wheel diameter than the gearing.

I posted what we feel the benefits are on my other reply with the CAD link.
We will loose around 4% efficiency due to the 3rd gear stage but should gain efficiency by using roller bearings over the delrin rollers on the cim end caps like the “In Wheel” module. Should be close to the same efficiency.
We were just talking about the higher center of gravity last night during our Mechanical Team meeting and this is a trade off. We have another version of this unit where the CIM comes out to the side but for that version there are two more stages of gearing and the extra weight and loss of efficiency didn’t seem like a fair trade off. Ground clearance is an issue if the CIM is tucked under the robot. But if being low to the ground or having a low center of gravity is valuable than we can modify the design. The “In Wheel” modules center of gravity is 3.25" off the floor and the coaxial unit is 5.75" off the floor. This version puts 20% of the robots total weight 2.5" higher, not really enough for concern with the center of mass.

We are starting machining of the prototypes at our next off-season team meeting. All purchased components are on order.
The rotational speed is 48 RPM at peak steering motor power, “In Wheel” module was 40 RPM. They can spin faster during driving with no pushing.
The ball drive is the 217-2792 unit from VexPro with a modified output shaft.
CAD link is posted in my other reply.

Where did you get the “actual” numbers from?

Crazy! This is definitely the most advanced/ best swerve I have ever seen. Dual speed at a weight of 7bs per module rivals that of WCD; it’s like having torquey holonomic motion at the cost of 4-5 lbs.

I have a couple of application questions:

  1. Why did you not flip the cim? It look like you have the room to do so. Flipping it and adding a belt drive to the first stage of the gearbox would add half a pound or so, but would alloy you to add another cim or a minicim to the drivetrain if you wished.
  2. Why the small banebots motors over something powerful like the RS-775 18v?
  3. Why did you choost to have a seperate pair of gears for the absolute encoder instead of simply having an encoder on the versaplanetary output?
  4. Tons and tons of machined parts (the most notable to me being the miter gears). What do you expect the turnaround time to be for these?
  5. How are you planning on fixing anything if a module breaks? It seems very compact and complex and hard to repair.

Again, very nice swerve drive. It’s a bit beyond my team’s capabilities, but I hope that you can use this this year and improve this further next year.

This looks beautiful. PWNAGE has been advancing swerve to new levels of elegance.

I do not see the steering motor in the views I have seen (or am blind) - I see the steering encoder, pneumatic cylinder and CIM - what are you using for steering motors/transmissions?

Also are you planning to be field-centric steering next year (of course with game dependency disclaimers)?

The steering motor is hidden behind the main shaft. It’s an RS550 on a 100:1 versaplanetary plus the final stage with teh big orange gear.

Looks like its ~82% of the theoretical speed. Maybe based on the number of gear reductions?

Really cool design by the way. Cant wait to see this in action!

If 2451 truly gets about 82% of the theoretical free speed out of these transmissions, that’s in the efficiency ballpark of most FRC skid-steer drive trains. Some teams can hit the low 90s, but that is the exception rather than the rule.

Also, the intuition of “what’s too fast for high gear” is somewhat different for an independently steered/driven swerve than for a skid-steer drive train. Your wheels never need to fight each other, so you are asking less of your drive motors than in a 6/8WD where you need to force wheels to slip sideways in order to turn.

Beautiful model. Thank you for sharing. Marie

I really love this design overall. I’ve wanted to design a swerve with a single reduction ball-shifter for quite some time. I do have some questions and concerns though.

  1. It looks like there is no thrust bearing above the vertical miter gear. This is not a big deal by its self, (We used thrust washers last year) but it will increase the speed of deterioration on the gears, and decrease efficiency in that gear-set. It also looks like you’re planning on using the VEX Pro aluminum Miter gears. This is also not a big deal, lots of team use them, but I suspect the fact that they are aluminum will increase wear speed. With those two things combined, your drive performance may not suffer, but you will probably find yourselves replacing the miter gears pretty often. What is the process for replacing a miter gear? How long will it take?
  2. How many man-hours will you spend to machine all the parts? How about to assemble the modules? I feel like you could have designed a module with the same performance, that would have been much less resource demanding to build.
  3. What kind of bearing are you using to move the robot’s weight from the red base plate to the top of the castor? In the cross section, it looks very thin.
  4. Not sure if it is a problem or not, but I couldn’t help but notice that your pneumatic cylinder is mounted to a plate, that has standoffs to a plate, that has standoffs to a plate, that is stood off from a plate, that has standoffs to a plate, that is mounted to your frame. Not sure why, but that makes me cringe a little.
  5. What gears are you using on the ball-shifter shaft, and as the CIM pinions?

VEXpro Bevel gears are 4140 hardened steel not aluminum.


This is a fantastic design. There have been many designs of two speed swerves posted over the years, but nothing has come close to what you’ve put together.
I know from looking at your swerve design from last year that you guys shouldn’t have trouble machining the parts in this design, but I’ve got to ask, how do you guys even have time to make these parts and finish the robot on time? What sort of tools do you have, and how many people do you have working?

Thanks Aren, that makes a lot more sense. I was actually amazed that the gears seemed to be holding up.

I’m still curious how the miter gears would be replaced, and how long it would take.