Rocker pod for slide drive- a brief analysis

Quick little analysis of how rocker pods (often used in slide drives) work since designing them for proper operation doesn’t seem trivial. They have a self-traction-generating characteristic but it can be overdone (or underdone).

Corresponding calculator: https://thaddeus-maximus.github.io/swissarmyengineer/rockerpod

6 Likes

Cool!

Small error corrected later in your paper. Equation 3 should have the wheel radius not the gear c-c distance.

1 Like

Fixed, thanks! :wink:

Cool stuff, Thad.
I think slide drives in FRC are generally underdeveloped, as people who want omnidirectional movement generally opt for swerve or mecanum instead, but slide has some distinct advantages.

2 Likes

Ohai Kevin!

Slide to me seems vastly superior to mecanum, yes. It’s harder to implement but a lot of that has to do with people designing overly complex center suspensions, or improperly designed rocker pods like these.

Maybe not so obvious is that rocker pods are also useful in other places beyond drivetrains whether or not people realize they’re doing it. We were trying to use one to drive the wheel of fortune (for packaging reasons… stupid 8" shooter wheel on a turret). A lot of intake systems with pivoting wheels work on (or fight against) the same physical phenomena.

1 Like

Nice. Seems pretty similar to @GeeTwo’s paper

1 Like

(post withdrawn by author, will be automatically deleted in 24 hours unless flagged)

13 Likes

Does everyone that writes papers about swerve credit 47 with using it first (afaik)? If you look at the video, his team obviously designed and made it too, so this comment rubs me the wrong way.

That’s a bit of a false equivalency imo. Patents for what’s basically the modern bevel besides wheel swerve existed before 47 ever fielded their swerve, but JVN’s rocker articulation system is unique as far as I know. Using a torque reaction like that isn’t new, but the application appears to be entirely original.

10 Likes

I think we should understand that sometimes, lack of attribution can honestly happen by accident. It’s not an excuse to not credit the original creators whenever you can, but it’s genuinely possible that someone may not know where the original idea came from.

For example: I personally have talked to my team about the 148 rocker slide pod, referring to it as exactly that (attributing it to 148 when I bring it up), because that’s what I refer to it as in my own head. But it’s quite possible that a few years from now, if I end up leaving this team, there will be someone who remembers the rocker pod design, but refers to it in passing conversation as just “that rocker pod slide drive” or something. Team members who only hear about the design from that person will never even realize that it was initially a 148 design, and therefore won’t attribute it to them.

I think, especially in the FIRST community, most do not want to take credit for something they did not create. Rather, I would tend to give people the benefit of the doubt and assume that they genuinely did not know that the design was something special that needed to be attributed, or at least did not know who to attribute it to. In this case, a simple “Hey, not sure if you knew this, but we initially came up with this design and are quite proud of it, and we’d appreciate it if you gave us credit for that” would do.

4 Likes

Thank you, that’s exactly the case for my paper a few years ago. I saw about four versions of this in a short period of time.

Hanlon’s razor: " Never attribute to malice that which is adequately explained by stupidity."

This is quite a heavy statement, how would you back this up?

I believed this until 2015, when I had direct experience with both. It turns out, a lot of bad mecanum implementations make it look worse than it actually is. If you’re of the mindset that it is faster to turn and drive forward with a mecanum than it is to strafe with one, then you haven’t implemented a properly-controlled mecanum drive yet. Once traction and wheel slip are accounted for they are actually quite agile.

(Full disclosure: I still wouldn’t build a mecanum over a tank drive most years, I haven’t gone completely mad)

1 Like

This really gets at me; there are so few elegantly controlled mecanum drive robots it’s kind of sad. Most teams simply use the strafe to line up a little bit then proceed to drive like it’s normal skid steering. I was going to post a video of a good example but literally can’t find any.

Best mecanum examples I can think of are 2012-2013 2052. Honestly, they’re still the only team I can remember doing mecanum well in FRC.

1 Like

Team 1986, Darwin division winner in 2017, is a really good example:

3 Likes

Fair enough.

3478 did well with them last year as well, FWIW.

1 Like

Not to derail thread that’s already careened (sorry JVN, I can edit, don’t smite me)…

Probably right. Hence my use of the word “seems”. Been a while since I’ve seriously played with omnidrives. But here’s my thoughts:
“Once traction and wheel slip are accounted for”… implies they need to be accounted for. I’m not intimately familiar with what this entails beyond making sure your frame is roughly flat and isn’t too stiff (which are easy enough mechanical things). But if this means sensoring and software (e.g. drag wheels), that requires tuning, and for some teams that isn’t something they have the exact skillset for. This seems to be much less controls intensive for omnis- and the construction of omnis seems to be a bit more robust than mecanums. I like things that ‘just work’.

If you can just slap a pod on with the correct effective friction, you’d be more or less set. Plus if you don’t like what you built and want to go back to a tank drive, take the module off, swap your wheels, and you’re off to the races (you could do this with mecanum, but the wheels would not be chained together).

1 Like

I am aware of 1986’s 2017 machine, but I was unimpressed in all the matches I’ve watched. The strafing ability is used almost exclusively in auto and there are several “tank moves” they make throughout matches. e.g. backing away, turning slightly, then driving forwards again to reposition where a translate would have been the same speed or faster. In theory, a well driven mecanum robot should look pretty close to a swerve.

1 Like