You could ask around if teams have older versions of current models they’d be willing to sell for cheap. This can be an inexpensive way to try it for the first time. You might have to replace some individual parts or find a few teams with extras to make a complete set. They might not have the latest and greatest packaging, weight, gearing, or wheel width, but they still work great and I’m sure you can do it cheaper than the cost of new modules.
If any teams are looking for some MK3 parts let us know. We upgraded our MK3s to MK4s and still have the leftover parts if those are useful to someone else.
We tried to do this with out first swerve drive. We spent two years trying to make our own, based of another design. One of those years (Stronghold) we decided it wasn’t a good game fit. Then we broke down and went with MK2.
We ran MK2 and then MK3 and then a hybrid version of MK3/MK4. This years we are making a full transition to MK4i. I run a good sized machine shop and sponsor/mentor the team. I looked at making the parts in house this year and decided it was cheaper for me to buy the SDS kits and donate them to the team instead.
We sent the design published by a “world-class” team to a local machine shop (which may or may not have been a sponsor). It was hard to get the tolerances we needed before we even assembled the parts. Finally modules almost worked smoothly in the air (on blocks). The robot on the floor bound up somewhere between seconds and minutes of use. There wasn’t enough time or grease in the world to keep it working.
I’m not sure cost effective is the right word here.
The SDS and other offerings are incredibly cost effective. We used to pay the same price nearly 20 years ago for the first andymark shifters! The swerve module at least has a wheel in it too
Can you make them cheaper? Marginally yes, as others have pointed out… But with far more labor invested by your team which is likely better spent on other efforts.
What’s the right answer? Depends on the goals of your team.
I think it’s also worth pointing out that if you are just messing around with swerve, you can get away with 2 modules only - Then use casters in the other corners. Don’t necessarily recommend that for competition play (I know of teams that have), but all the principles apply.
You can also get away with not including an absolute encoder, and manually aligning the wheels, if that’s helpful. Then just use an incremental encoder, like the one on PG motors (or inside brush-less motors).
And if you are just driving it around for fun, you don’t need an encoder on the drive motor.
These options should reduce the price tag dramatically for you if you just want a demo chassis.
I’ll also throw the “Mucho Cheapo Swerve” into the mix from this time last year. If the goal is a test bed for code, or just proof of concept that you can make a swerve drive, it doesn’t get much cheaper than this (especially if you already have access to a 3d printer).
Remember, “world class” in FRC != world class manufacturing practices or documentation. FRC is a race to cut corners the most efficiently. Some of the top teams put together technical binders and extensive internal documentation, but many more don’t.
This is interesting. I’m not a top team (I’m just a person) but here’s a quick ad for Extensive Internal Documentation:
Are you tired of all your work being ephemeral?
Do you want to do stuff that makes a positive impact on more than one cohort of participants?
Does information entropy make you sad?
It doesn’t have to.
I appreciate the references to CheapoSwerve, I never intended that to be the official name, but I guess it’s stuck :P.
I’ve been iterating the design for the past 4 months or so for a separate project, and have since made it way more viable for FRC-related testing. It’s been assembled, iterated, tested, and it works really well.
Here’s a link to the updated design, and here are a few teaser shots:
If you want to build a swerve for the sake of having one, I highly suggest looking into FTC Team 11115 Swerve. It will be cheaper and easier to drive safely wherever you want.
If memory serves me right, most of the parts can be 3d printed and will be much more cost effective. Also, if you are looking into control systems as well, take a look at this thread and this thread I posted to group building a mecanum bot for a school project. It is significantly cheaper to not got through FRC/FTC legal hardware and do Arduino instead.
Eh, You’ve got about 6 weeks pre season. The last time I did a clean room swerve implementation it was sit down and code for about 4 straight hours to get robot centric working and then field centric was about another 10 minutes to implement (and then about a week to debug because I didn’t have hardware and the gyro used one direction for positive and the robot for the other which led to weird results)
(For the record, my field centric code is in its entirety below, it’s really quite simple matrix math)
public static Vector2d TransformFieldCentric(double x, double y, double headingDegrees){
Vector2d control = new Vector2d(x,y);
control.rotate(headingDegrees);
return control;
}
Out of 4 swerve implementations that I’ve personally witnessed, none of them were able to get absolute encoder offsets to behave within 2 weeks. There’s a lot of little things that make swerve more difficult to implement. For example, few teams know a good way to properly align the wheels to the coordinate system before setting absolute encoder offsets.
And the last one I’ve done, from scratch, took way less time than that.
There’s some gotchas but it’s not insurmountable. With the explosion in software support and example code the last few years it’s only gotten easier and easier.
I said the same thing until I started. Took me about a month overall–and that was with essentially full documentation of the 3-month struggle my brother went through last year to do swerve.
