An Answer About FRC Terrain Games

Hi everyone! Considering how popular swerve has gotten in the past two years, many teams have started ordering their modules before the game has even been released, including my alumni team. However, one risk you take is that, while swerve works well in most games, off the shelf it isn’t always be best option and there’s always a chance that your expensive swerve modules will be entirely useless. One game that always comes to mind as being “bad for swerve” is an obstacle-heavy ones such as 2016, and in that case the majority of teams likely would (and definitely should) use a different drivetrain. Or, of course, you could simply ignore logic and reason and spend hundreds of dollars to use those modules anyways! Hence, taking inspiration from 4143’s iconic 2016 robot, I designed my own version of twerve!

Primary Design Constraints

  • Bolt onto SDS Mk4i and Mk4 modules
  • Use as many of the original components as possible
  • Have as few machined parts and machine ops needed to make them
  • Be “feasible” to make in-house - ie. a normal router, drill press, 3D printer, and perhaps a lathe. Not need any 3D machining or anything else that would require a CNC mill. Also heavily utilized markforge printing, because that’s what I was used to from my alum team.
  • Use as many non-VEX COTS parts as possible, while using 0 VEX parts
  • Have the lowest angle and highest clearance possible to assist with getting over obstacles
  • Be as lightweight and cheap as possible (there are at least 4 on each robot so even small increases in weight can increase the overall weight and cost of the robot significantly)
  • Be able to swap the entire twerve “module” quickly, and be able to easily change the main belt
  • Have a surface speed that is actually usable - ideally, slightly slower than the 4" wheel with the equivalent module gearing

Now with that said, here it is!


As you can see, I made two different versions: a symmetrical version, V1, and an asymmetrical one, V2. After making the first version, concerns were raised about the angle of it and the size of the lower wheel, as well as my choice to make most of the gears out of printed plastic or aluminum. So, in V2, I made it offcenter to allow for a much shallower angle and made all gears steel. I’ll be comparing these to each other, as well as the original 4" wheel.

After this, I had to make sure this design actually worked in terms of allowing swerve to traverse obstacles, so I looked through all the previous years: this design would work as-is in 2010, 2012, and 2016 (the highest vertical obstacle for each of those was ~4"), and could even be modified to go onto the 6" tall platforms in 2004 and 2019 (although it wasn’t strictly necessary either year to do the main part of the game).

V1 Design

This originally started as one large central gear, but I quickly realized that made the system WAY too fast and also limited how large I could go. To fix this, I added in idler gears between the 48T center gear, which is bolted to the original bevel gear. This leads to an additional 48:18 gear, which powers the 24T pullies on each corner. Thus, this design has the exact same surface speed as the original 4" wheel, so the speed of your DT will not change.

Issues

I had made it an equilateral triangle, but I pretty quickly realized that a 60 degree angle would seriously struggle with obstacles, and any agility gained by having swerve would be lost by how slow it drives over the obstacles themselves. Plus, the gears were mostly aluminum or plastic for ease of manufacturing and cost, but I doubted this would hold up in competition. After some consideration, I realized there was very little reason to keep it symmetrical - the only major advantage was reducing how much I would have to rotate the module to turn (maximum of 180 degrees as opposed to 90), so I decided to make an asymmetrical module. Also, this design used VEX components (I designed it back in October), which was a big no-no.

V2 design

Another change to this design - idle bottom pulley. I felt that didn’t have to be driven, and that it was more important to make it a larger diameter and aluminum. If you wanted to, there’s space to put a 36T gear on that axle and another idler, I just thought the slight weight and complexity savings were more valuable. Besides that, I chose to increase the gear size on the driving pulleys to 24T, which makes the entire drivetrain about 3/4 of the speed - you could likely change the size of this to be closer to 1:1 if you preferred. The largest change is, of course, making it asymmetrical - now, the approach angle is 44 degrees, which will be able to handle bumps much better. Additionally, this allowed me to move the center axle up, nesting the bevel gear inside the mounting block and, more importantly, allowing me to use the shaft that comes with the modules, getting rid of what was one of the most difficult parts to manufacture in house.

Conclusions

Overall? I like this design. I do think it would do a good job of it’s intended purpose: allowing me to use swerve in a terrain heavy game. Do I think it would be better overall than just doing a WCD-style drivetrain? Maybe not, but I don’t think it would be worse, and it would certainly be cool, which isn’t a bad thing. Might win us a few awards.

Let me know what y’all think - I’m aware this is a controversial design, but it was something I enjoyed making and I’m always open to suggestions on how to improve it - even if it’s “give up and don’t use swerve”. Best of luck to everyone in 2023, and hopefully it isn’t a terrain game because I bet a carolina reaper on that. Have a good day!

Links

SolidWorks Download
Twerve V1 Parasolid
Twerve V2 Parasolid
Order form

44 Likes

Programming mentor here… Would the mk4is actually work here? Won’t the triangle rotate into the motors?

Ah thanks for letting me know on the forum that from the picture it is evident you just “un-i” the mk4i

1 Like

Yes, you just need to mount the motors and tube on top of the plate, aka the “ultra high ground clearance” config. Flipping everything back up does completely defeat the point of the Mk4i but… if you already have it, this does work.

This … scares me, lol.


I was not familiar with 4143 (MarsWars) 2016, but I had recently been introduced to 16 (Bomb Squad) 2016, which also used swerve.

Bomb Squad achieved all 2016 obstacles (demonstrated near end of video) with angled ‘bash plates’:
image

Your swerve design demonstrates a greater theoretical vertical obstacle than what Bomb Squad demonstrates and Bomb Squad could only travel straight over obstacles while yours is ‘true’ swerve:

(The wheels move but not the plates.)

However, I was wondering if you ever considered a system like this and what the decision to not pursue it was. Thanks!

EDIT: oh yeah it was the below thread that triggered my thoughts

4 Likes

May as well link this in, from a week or so ago:

1 Like

With the second iteration now you have caster angle to worry about which CD is opinionated about to say the least. The SDS modules are not designed for caster angle and that additional robot weight being thrown around (let alone impacting an obstacle with that huge lever arm).

1 Like

Live video please.

2 Likes

That seems to have worked well, but I wanted to try this design. I think this would be able to handle obstacles better, on top of being able to clear them in any direction (which could allow it to clear some obstacles more easily, such as Mars Wars “spinning” over some of them in 2016). Also, this felt like it would work better as a generalized design, wheras I felt that would be more game specific. Finally, while it was less of a concern in 2016, breakover angle mattered in previous games, with robots potentially getting “beached” on the obstacle. 2010 comes to mind. I don’t think those sleds would do much to help with that, wheras this could accomplish this fairly easily.

Can you send me a picture of the cad on the other side it dose not seem to work out that simply

Here is my current try at flipping it

1 Like

I think this requires a post in this thread to properly document this bet for 2023 :grinning:
My bet for 2023

1 Like

What would you be using for the belt/track/tread?

You sir are a madman, and I salute you.

So I know the x-contact bearings used in SDS mk4 modules are boasted to have survived over 100 drops from the level 2 hab platform in 2019, but have you done any calculations or testing to validate their ability to survive crossing something like the rock wall in 2016 using this design? In addition to the shock load from the fall on the far side of the wall (akin to a 2019 hab drop), you will also have a large combined load created by the act of climbing over the rock wall.

16’s skid plate design avoided this issue by transferring the forces generated as a result of climbing over the obstacles to these skid plates. Is anyone familiar enough with 4143’s 2016 drive to comment on what type of steering bearings they used on their swerve modules?

I would be very tempted to use a custom bushing in this application. Lotta weird loads going on that xcontact bearing.

For the record I am a big fan of getting rid of any weird loads on swerve modules. The thrust loads in 2019 were bad enough and that was before half of FRC decided to try swerve.

1 Like

“Give up and don’t use swerve” is still probably my answer, but if you really want to go this route I would probably spend more time looking at suspension and less time looking at tread/wheel configurations for the modules (you can always just use big wheels, right?).

Still some really cool work, and I’d love to see more!

3 Likes

I love the twerve concept (even though I am still partial to SWANK).

I have a couple of questions/comments about this design and your design choices:

  1. The 60 degree approach angle of the V1 seems a bit steep to me. I can see that you tried to fix that with the V2, but ended up with some compromises in terms of symmetry (which is probably a bigger con relative to the drive kinematics - i.e. you can’t rotate the module about its own center while the robot is stationary). If you only drove the bottom wheel on the V1 (i.e. removed the idler gears connecting the central gear to the upper corner wheels) it seems like you could lower the approach angle to 45 degrees, or even lower while keeping the symmetry of the V1. Obviously, you are losing approach height as you reduce the approach angle, so this could be optimized to the height of the obstacles (whatever they are). Since 6" wheels are pretty standard for the KOP drivebase, it is likely that any real obstacles would not need more than a 3" approach height.

  2. What belt are you using for these wheel treads? Is there a COTS treaded belt that you have found, or are you planning to try to create your own by attaching the two ends of a strip of belt together. I would think that you would have some difficulty attaching the two ends of a belt together in a robust way (I have seen the caterpillar style clamps that are used in industrial conveyor belts and while that could work, I see that creating lots of problems for you). In addition, I would think that a cogged belt like the rhino belt that AM sells would allow you to transmit the drive power into the belt without the belts slipping on your pulleys.

  3. It looks like you were able to come up with a module design that was inside the frame perimeter as long as the modules were rotated 45 degrees to the frame axes so that you would have at least one starting configuration that respected the frame perimeter. You don’t really have a good picture confirming that, but the pictures that you did post look like it is pretty close. Can you confirm?

1 Like

The question I’ve been pondering is at what point is a game non-swerve. When does a team conclude that skid plates are not acceptable. In 2016 we ditch swerve and built a tread tank. Loved that robot and it handled the field obstacles well. I dug up and revisited the cad. This will be my first concern Saturday when I get the first view of the field. To swerve or not to swerve. A decision that will drive the whole build.

Swerve provides some level of benefit in all games (in terms of maneuverability and ability to maximize scoring). It also presents limitations (cost, complexity and, in this case the inability to handle terrain). So the real question is when do the limitations outweigh the benefits. That point is going to be different for each team. It is going to be based on you own team’s abilities, your own strategy and your own tradeoffs.

I doubt that FIRST will ever give us a game where swerve is not an option. For some games, like 2016, it may not be a particularly good or easy option, but even in 2016 several teams still felt that swerve offered enough advantage to find a way to make it work.

When we developed Swank for 2016, it was all about keeping the advantages of swerve while overcoming the limitations:

2 Likes

Even if this is a terrain game, I don’t see FRC making it to where swerve is completely useless. My guess is there would be a path that could be taken that would allow swerve to still work, though it may be out if the way.

1 Like