GRT 192’s 2017 fully modular swerve drive. Belt driven drive and steering, all-plastic structural components, including an acetal rotation bearing. A combination of off the shelf and in-house manufactured pulleys. 6.9 pounds, all included weight. Adjusted maximum speed of 13.5 ft/sec
Seriously dope. Can’t wait to see it at SVR!
Thats really clean I like that alot Great work!
You say “all-plastic structural components”, is that referring to the main beige part as well? What exactly is that made from and how? Looks gorgeous.
Yes. The main beige part is printed on a Stratasys uPrint that we have at our high school. It is 80% infill ABS, which is the highest infill we can print at without warping.
I enjoyed seeing this thing in action at Hub City!
I always look forward to the innovative gearboxes from GRT. Well done!
Cool.
Is your turning pulley a series of laser cut pieces of delrin? I assume the bottom bearing is just another layer of that?
I have to ask, what do you use for zeroing the module?
And what do the screws across the module do? Are the modules printed as a single piece or multiple pieces? With support?
What made you decide to focus on using plastics for all parts of this design? What lessons have you learned from doing this and what would you do differently in the future?
I TOTALLY don’t ask because this has been on my “tinker” list for a while.
The turning pulley is two pieces of laser cut acrylic. The bearing surface attached to the wheel fork is delrin, but it runs against an acrylic plate.
As for zeroing, there is a cam located near the top of the wheel fork that hits a limit switch.
The modules are printed in two halves, and the bolts hold them together. We used dissolvable support material to minimize the amount of cleanup we had to do for each print.
We decided to use plastics because our CNC machine broke earlier in the year, and we had access to a laser cutter and a 3D printer. The modules were originally designed for metal, and later adapted for printing. FEA was extensively used to ensure the structural integrity of the modules.
As for lessons learned, GRT has a long history of overbuilding both drivetrains and mechanisms. FEA allowed us to trim down on excess material and quell the fears that the rest of our team had about plastics. In the future, we would like to improve our zeroing system, because our accuracy is limited by having a imperfect cam profile and limit switch.
We would love to see what you come up with in the future!
Have you had any issues with the acrylic shattering?
Also, having some experience with 3d printed timing belt pulleys - they hold up pretty well. 
Very nice!
How does the CIM transmit power to the wheel? A picture of that side would be great.
Good luck!
We’ve only had one incident where any acrylic shattered. The pulley on one of our prototypes was not retained and walked up the shaft, and was driven when partially engaged on the shaft, shattering the pulley. No plates have broken on our competition bot. Because we used FEA to simulate the entire module, we were able to make the module work with acrylic by choosing the right thickness, and the supporting the module on the robot in a way that reduced the stress on the plates.
The CIM transmits power to the wheel through a simple belt drive, much like the first stage of our previous gearboxes. We’ll work on getting a picture of that side uploaded soon.
This is very exciting. The idea of a functional 100% 3D printed swerve module makes me smile. I bet we see one inside two years.
With the MarkForged style printers and more advanced fiber based printing (continuous strand, not just shredded bits), it’s definitely doable. Can’t wait!
We have one of those printers and with the fiber strands, those things are very strong. I usually bring a sample with me on robotics trips to show other teams. We paid about $10k for the pro version but definitely worth it.
I believe 987 also bought one last season.
What FEA package(s) did you use? What were your design targets/what loadings did you consider?
We used the built in FEA suite in Autodesk Inventor. We based most of our simulations off of the weight of the robot and the coefficient of friction of the wheel, then aimed for a safety factor of somewhere between 4-5. We ended up with a minimum safety factor of around 3.5 in the final revision of swerve. It’s been through a regional already, and it has held up fine.
How do you cut material that thick without having trouble with the kerf taper? I’ve never seen a cutter that doesn’t have a significant amount of taper on cuts.
The timing pulley is two thinner laser cut pulleys stacked on top of each other. This way each cut goes through less material, reducing the effect of kerf. While there is still some visible taper in the final product, we have not observed any ill effects stemming from this. The belts track well on the pulley and we have not seen any excessive wear on the teeth of the belt.
After two regionals with playoffs the swerve is still going strong.
1 Like