Hey guys so I was thinking about this years Curiosity landing and the rover itself, and got to thinking about its swerve/castor drive. I want to start designing a modular system for swivel drive for use on my teams robot this year, and was wondering if any of you have any ideas/tips/advice for me. Tell me what you think!
Have a look at 1717’s modules.
Enough said.
Also remember half of the swerve complexity is code…
A simple, well designed Tank drive beats a complex, mediocre swerve any day. I’m not saying your swerve will be mediocre, but even with years to work on it, the complexity of a serve isn’t always worth it.
However, this is a very one-sided opinion, coming from someone who has never built a swerve. I’d talk to someone who’s done swerve before (Maybe someone from Team 16, they do awesome swerves) and get their opinion.
Either way, however you do it, please keep us updated as you iterate!
Swerve drive? In your second year? Better get started now…
First, I should warn you that the mechanical part is the easy part. After all, you only have to figure out how to rotate a wheel, and its power routing (motor power via gears, or electrical power without totally twisting the wires), and make it fit on the robot within the weight limit. Do you want all the wheels to rotate together and drive together, or do you want independent rotation, or independent drive, or split with one side controlled by one group of motors and the other controlled by a different group? To make that a little bit easier, see http://www.andymark.com/Swerve-s/148.htm for a couple of designs derived from proven ones. Also take a look at 16 and 973’s swerve designs.
Then the programmers get to figure out how to make it work right. That’s the fun part. Not only does the programmer have to get everything responding (and pointing the right direction), but then they have to tailor it to any driving modes that the drivers want.
Can it be done? Yes, and it has been done very successfully. Is it easy, no. NOW is the time to prototype one and see if you can solve the issues before the build season. You might not be able to use it this coming season due to other issues (including the game), but it’s always nice to have an extra system in your bag of tricks. For a swerve drive, you don’t want to be prototyping during the 6 weeks–not enough time unless you have a massively large team with enough tools and materials for everybody. Do it during the offseason if at all possible.
Haha call me ambitious. I mightve got ahead of myself saying “this year”, but I think it would be a good use of my offseason time to design/build one of these.
Okay. I can lend you a few tips. 1126 did Swerve this past year, and I was kinda attached to them for the time being. So I’ll tell you a few things I liked, and what I would’ve done differently.
First of all, designing your Swerve drive. 1126 bought a Revolution Swerve Module from AndyMark, dissected it, and created a CAD file from it. Then they set to work designing the full drive system.
If you DO want to go with Swerve, there are a few decisions you have to make. Namely, motor allocation. You could go with each module steered by its own motor, or you could chain a few together (2 and 2). That’s how 1126 did their prototype chassis. That turned into 2 and 1 and 1 for Championships (2 chained together, 2 are independently driven). Now they are working to make it totally independent. I’m not exactly sure what the advantage is to making it all independent, but I’d imagine the modules would rotate faster and easier. It’s also heavier and a little more complicated. So keep that in mind.
Now, we’ll fast forward a bit to when the prototype chassis is built. Hand it off to Software ASAP! As said above, the biggest hurdle of Swerve is the programming. There’s a lot of things that need to be handled by Software. Not just driving the wheels like normal, but rotating the wheels and KEEPING THEM IN LINE. When steering your Swerve modules, they need to be kept in the same direction to a rather low tolerance if you want to drive effectively. Rotating the wheels itself can be a hurdle. There is just so much Software has to do with a Swerve drive. They need to be given as much time to work on it as possible.
Now, when Kick Off comes along. I would not recommend putting Swerve on the field for the season yet. That’s your choice, but 1126’s Swerve was no where near the level it should have been to compete. And it really hurt them when they got to competition. Now they know, and have been improving it for the coming year (if it offers advantages for the game). I would recommend giving Swerve another season to improve (AT LEAST). There’s no point to using Swerve instead of Tank if it’s barely mobile at all.
Just remember when undertaking Swerve; it requires time, resource and practice. And even with all those things, it may not work as well as you’d like. In which case, you need to invest more time and resource in it until it does work like you want. Don’t expect to be at the level of 1717 or 16 when you’re done with the prototype. It takes years to get Swerve to that level. Just remember; don’t get discouraged. If you’re truly motivated to do a Swerve drive, don’t give up on it. Just keep working at it until you get it!
I hope this helps. Let me know if there’s any advice you need, and I’ll help as much as I can!
-Leeland
If I recall correctly, we tested the 2-1-1 on the prototype swerve but did not find any significant advantages to try and change it on the actual robot. The only change to our swerve in St. Louis was the Colson wheels. I guess, when one of our modules came off during a match, we did have 2 chained together and one independent, but that was unintentional 
There was a team of people that worked over the summer to convert the swerve to 4 independently steered modules. Originally, that is how we wanted to build the 2012 robot, but the shooter team grabbed a window motor for the turret. Having 4 independently steered modules seems to be the better way to go in terms of programming, mobility, maneuverability etc. Last I checked, it is mechanically complete and is being reprogrammed.
Second, there are loads of presentations, threads, objective tables etc out there on these decisions. I will try to summarize them into this: build a swerve drive because you have gone through the thought process and decided it is worth the challenge and will offer you a benefit during gameplay. Do not build it (for the competition season) because you really just want to try it or you think it is cool. If you do want it, make sure you prototyped it first. 6 weeks isn’t nearly enough time to design a swerve drive from scratch and do it well (especially when the best swerve drives have been perfected over years).
Garrick
One of the problems of the drive train your talking about, is that in our appliances, we usually don’t incorporate the suspension system the Mars Rover has, so if you found a way to make a dropcenter swerve drive that could tackle bumps like the barrier in the 2012 game, i believe it would become very popular very quickly.
As long as its simple.
Look up Winnovation’s 2010 swerve, all six wheels of it.
It’s still the only 6-wheel swerve to compete in FRC. Swerve by definition isn’t simple… and adding an extra 2 wheels doesn’t help.
Most of the time a 4-wheel (or even 3-wheel) swerve is used, the field is flat. But, this year, several teams used swerve and also crossed the barrier without using the bridges (and then balanced the bridges).
If you DO want to go with Swerve, there are a few decisions you have to make. Namely, motor allocation. You could go with each module steered by its own motor, or you could chain a few together (2 and 2). That’s how 1126 did their prototype chassis. That turned into 2 and 1 and 1 for Championships (2 chained together, 2 are independently driven). Now they are working to make it totally independent. I’m not exactly sure what the advantage is to making it all independent, but I’d imagine the modules would rotate faster and easier. It’s also heavier and a little more complicated. So keep that in mind.
We changed from 2 pairs of wheels being rotated by one window motor to all four wheel being rotated separately because it would allow the robot to turn more efficiently. When the wheels are connected together and you try to spin the robot, two of the four wheels are dragged on the carpet. This both damaged the carpet and we had to change the wheels tread often in order to have enough grip to climb the bridge.
Truth. If you have some members who are stuck on the idea of swerve, tell them to design one. They’ll quickly see that it’s not as easy as they think.
They get a shot of all 6 wheels at 1:45. Its crazy cool.
Though also remind them that the design is less than half the battle. Programming a swerve is the hardest part.
Truth. If you have some members who are stuck on the idea of swerve, tell them to design one. They’ll quickly see that it’s not as easy as they think.
Designing a module is a relatively straight forward challenge with all the CAD out there. A better challenge would be to price one out on a small budget, swerve gets expensive fast.