Swerve advice

[Mpengelly]

Our team built a swerve drive robot this year and although it works well on the practice field, under the stress of competition, our driver seems to struggle significantly. We have our second regional this weekend and we are in the midst of trying to figure out what we can best do to solve this problem. So to those teams that have lots of experience with swerve, I have a few questions:

Firstly, what joystick or controller is most commonly used to control swerve. Our team has an X-52 (http://www.saitek.com/uk/prod/x52pro.html) which has 3 axises and seems like it would be the perfect for swerve or Mecanum. For some reason our main driver elected to use a gamepad having throttle on one joystick and direction on the other. It seems to me that using X-52 with throttle being magnitude and direction being angle on x,y would make the most sense and then use z to rotate. I'm not a programmer or primary driver but I'm trying to figure out what might help simplify and improve our controls/driving. The reason I would hold back suggesting it to the team is that changing the drive system going into a regional doesn't seem like a recipe for success either. Thoughts?

Secondly, it's been proposed on the team that we could change the front two traction wheels to double omni's and drive tank because our driver has a ton of experience with it. While it seems criminal to not use swerve, it might serve us better and I was wondering what people thought. Another question would be if it would maintain any of its swerve capabilities which I doubt because of the radically different coefficients of friction and we don't have encoders on every wheel. That said, I would love to be wrong.

Mainly, I'm just trying to figure out what the best course of action is moving forward. Changing things seems like a bad idea, but it may be the only way to have more success at our next regional. Any input would be greatly appreciated.
Thanks

[Nuttyman54]

Team 190 had a swerve drive in 2009, and had the option for a setup which was primarily tank steering, but switched to swerve mode when the trigger on the joystick was held down. We also used the 3 axis joystick for swerve functionality and it worked well. It can take some practice to get used when to switching back and forth (and remembering to do so), but it might be more natural for your driver if he's used to tank drive. You might do best to change the front wheels to omnis for this setup, so that it will turn better.

You should still be able to swerve with this setup, although the system might not react the same way as it has been. Omnis would still provide the necessary force in the necessary direction (CoF shouldn't affect it unless they have too little weight on them). You might see a difference in rapid direction changes, as the omni characteristic could cause the drivebase to fishtail or otherwise swing in weird/unexpected ways, but it's hard to say for sure. It's also prudent to note that omnis do have some vibration associated with them, and while this might not be an issue, it could also cause problems in the swerve system if the vibration is severe.

If your robot turns ok tank style without the omnis, this will likely give you a more predictable system, but chances are that four traction wheels will not be happy turning like a tank drive.

If it is a fairly simple process to switch the current traction wheels for omnis, I would recommend you prepare that for the regional, and give your driver the option for tank and swerve. He might need some reminders of when to use the swerve functionality to his advantage, but working to add that in will probably produce better results than having him struggle with the current system, if he's used to tank.

$0.02

[Mpengelly]

Thanks for the quick reply!
I was wondering if you would see any significant benefit to using double omnis vs. single in this situation?

[Mk.32]

We used single omnisin a four wheel drive config, though not swerve, and they worked perfectly. We choose them cause they were cheaper. I don't see any reason for going double. Just my 2 cents.

[Al Skierkiewicz]

I am torn between telling you one thing and suggesting something else. Crab is very hard to learn and takes lot's of practice to get it right. It is easy to blame joysticks, gamepads and even programming but it is practice that is the real key. We have used crab several times in the past, when the game required it. Our drivers spend a great deal of time with coaching to get proficient. They need to be seeing the moves in their sleep. Practice requires driving with other objects on the field, other robots and moving devices that interfere with driving. It takes lots of team work to make these simulations real.
That being said, I usually recommend against major changes to robot design once a team has significant practice with the robot they currently have. If you change to tank steering, you need to be able to give your drivers sufficient practice to learn the new methods. If this is your second regional, you might want to think about how your drivers will react in pressure situations when you change the drive system. If they were drivers for you last year, and tank is what they trained for last year, it might work to your advantage to make the change.

[Mpengelly]

I would totally agree with you that practice is the problem, and I think given a ton of time our driver would be completely capable of learning the system and the controls. That said, our driver has driven tank with the exact configuration we would be implementing for 3 years running in our pre-season scrimmages with great success. I too am against changing anything major, but in this case, I think it might actually be to our advantage. Like you though, I'm very torn

[Al Skierkiewicz]

If you think you can change two wheels to omnis in a short period of time so that you can practice, I might agree that it worth a shot. The drawback is of course, being able to be pushed from the side but you know how to drive for that.

[Mpengelly]

I'm thinking that if we do chose that route we could have them swapped in under an 2 hours which would still give us time for practice matches Thursday. and yes the loss of traction, but our experiences at our last regional showed that maneuverability using a system our driver knows would me more important than the few times we have to push. Thanks for the great advice!

[Al Skierkiewicz]

Originally Posted by Mpengelly

Thanks for the great advice!

Anytime!

[SprocketGuy]

FIRST Team 1075 has been driving swerve for the past 3 seasons with varying amounts of success. During Lunacy we found the Swerve to be very handy with the lack of traction, and used it to great effect. Last year in Breakaway we had the crazy notion to build a Dual Sided swerve drive, which was a complete flop, but we really learned a lot about our drive system and were able to really perfect it for this season.
Here's some thoughts that come to mind:

Tank style only really worked in a wider configuration vs narrow. Found in a narrow config that it wasn't as fast and lots of friction.

This year we have joysticks (we use the Logitech Attack 3) set to the following:
Both Forward = Forward, Both Reverse = Reverse
One full back and full forward is actually what we call Monster Mode (ie Monster Truck, where the back and forward wheels turn at opposite 45s.) This was the biggest change for us. Still great turning without the issues of Tank friction. Doesn't turn on a dime exactly like a tank though.
Switchable Strafe and Lockable Strafe. These are two different buttons as one allows us to quickly switch to a strafing mode a la Mecanum and one locks the wheels full 90 for lining up the poles.
We find that we don't need to strafe that often until we get to the poles and then strafing is fantastic.
Our default driving mode is "Car" for blazing up the field and grabbing tubes.

I totally agree that practice is absolutely key, and I would say it's our biggest downfall. It takes us the full 6 weeks to build so we are lucky if we have any time to practice. It takes us at least one competition to get up to the level we should be. I have seen some horribly driven Swerve drive robots where there is almost too many options and the driver just can't control the machine properly. They will be going full speed down the field and then suddenly turn their front wheels to full 90 and flip right over.... you need to find what your driver wants and uses a lot and then program the wheels to do that specific task.

Driver/Programmer talks are critical with Swerve so that the driver can get the robot to do what they want as logically as possible. (This is what led us to change the turning controls)

Not sure how you are monitoring the wheels turning, but we have a limit switch set up to make sure the wheels don't go past 90 and it allows us to straighten our wheels if communication is lost midmatch.

Biggest improvement by far this year is we went with a gear driven wheel direct from the CIM Motors. Last two years we went with chain and it was awful. Still have chain doing the turning.

I'm not the driver or the programmer so I am just spouting off what I know from being around the team so if you want more specifics/code ideas, whatever let me know and I can hook you up with the right people on the Sprockets

[Ankit S.]

My team is using mechanum wheels, and i am the driver so i might have some information. I dont know how similiar swerve and mechanum are, but when i saw your robot, it looked like it is possible for them to be configured the same way.

We have 2 joysticks, one which controlls forwards backwards and strafing, the other which controlls rotation. The only speed changing we have is when the trigger on the joystick was pressed, the robot would travel at half speed.

That config works for me, and it might work for you too.

Hope this helps, and sorry for the not-so-great grammar.