Overclocked's Rover Reveal

[Joe Johnson]

TLDR: Watch this video of the Overclocked FIRST Fall Practice Project.

Merry Christmas FIRSTers!

Not that this is a gift to anyone but either I post this today or it may not get posted at all as I am heading to my 90 year old mother's home and the internet is, well, how do you say Ye Ole' Dialup in Modern English?

BOOM DONE. and the Robot in Three Days experience last year really lit a fiter under my desire to get back on a real FIRST team for the 2015 Season. This and that, I ended up on Overclocked, team 246, as the lead technical mentor.

One of the things I brought to the team was my strong belief in doing Fall Practice Projects as a way of getting better at FIRST but also as a way of getting everyone on the team to understand the team's capabilities. I can't tell you how important it is for a team to have a realistic understanding of what it can and cannot do. Sizing your robot ambitions to match your teams abilities is really important.

For this year's project, we decided to try to advance the teams ability in the area of Chassis Design. Having been an early innovator in swerve drive with Chief Delphi, my tendency was to push the team to build a swerve drive. but, Overclocked had a much stronger CS team than mechanical or electrical and frankly, I felt it was just a bridge too far for this team.

But... ...I've known Anthony Lapp from back when we was a pup cutting this teeth on Team #1 when they were sponsored by 3D Systems*. When he's not doing cool stuff for NASA or designing clever mechanisms into Team 118's robot, he selling Swerve Drives at Team221.com.

I was bold enough to ask him and he was kind enough to offer us Wild Swerve modules. Of course, I would have rather and his sexy new Revolution Swerves, but beggars can't be choosers.

With the hollow shaft Wild Serves in our back pocket so to speak I felt we could take on the mechanism of a 4 wheel drive, 4 wheel independent swerve.

The software and electronics teams took on the rest.

Anyway, with out further delay. Here is Roever, Overclocks Fall Practice Project: Rover, a study in "unpinnable" robots. (13 minute vide of me yapping and the robot being great).

Features

• 4 wheel independent swerve. 6 inch wheels, 1 CIM per wheel, 10:1 ratio
• Steering with 775 Banebot motors plus VersaPlantary. 54:1 from motor to swerve.
• Nav6 "IMU" (really a high tech electronic "South Pointing Chariot") plus a ton of software provides driving in Real World Coordinates (as opposed to robot coords).

We will post the code and CAD online soon. Stay tuned.

Can't wait for the new season to start, glad to be back in the saddle.

Tick tock, Overclocled!

Dr. Joe

*Fun fact, that 3 in their name made them team #1 by the way -- in 1997 when FIRST assigned the permanent numbers for the first time, they sorted by Sponsor Name and in ASCII, 3 is before A. True Story...

[Jay O'Donnell]

Personally if I was building a swerve drive I would probably go with a bit faster speed through the gear reduction, 10:1 seems like too much to me. Besides that, the cleanness and execution of the design is pretty sweet. Glad to see that you and Paul both found a new team. Hopefully we won't have to deal with that swerve being against us this season!
See you at the Reading District!

[Joe Johnson]

The 10:1 ratio was not my first choice because I was hoping to have a higher top speed, but I wanted to be sure to have enough torque to break the tires free. I decided we had more margin on top speed than on torque. Gave up some of the former to get more of the latter.

That said, it is still tolerably fast. Not blinding but I think we could live with it if we had to.

Joe J.

[asid61]

I would go to 5:1. Your choice of course, but even something like 8:1 will still be traction limited.

This drive looks downright fearsome. I can't even see the wheels, so there's no way of telling which way it will go next. Could be a useful mind game tool.
Plus, extra chassis space.

[Joe Johnson]

I'm interested in feedback about round(ish) robots with swerve drive. At this point it is quite hard to plow any truly new ground in FIRST, so I'm sure others have done this. Can someone with experience comment on how effective the combination is at avoiding pins in the first place and getting out of them in the second.

My impression playing with Rover is that it seems like a very effective strategy, especially with a continuous bumper covered in sailcloth.

But, I've been wrong before. I'd like to hear the voice of experience.

Do tell...

Dr. Joe J.

[Pault]

Here is the code for Rover.

Source code can be downloaded here.

The robot was coded in Java using the CommandBasedRobot template. This is our team's first year using Java, and we have been extemely satisfied with it and the great libraries put out by Brad Miller and his crew at WPI. CommandBased robot gave us a strong basis to structure our code off of without limiting us at all. We absolutely plan on embrasing it during the season.

The code, although it may look like a lot, is actually fairly simple. The core of it can be found in the Drivetrain subsystem, which uses vector math to allow us to both drive straight in any direction without turning (crab drive) and simultaneously spin about an arbitrary point relative to the robot (snake drive).

Note that there are many features which we developed, but did not have time to test. These are not included in this release.

[Chris is me]

It's really cool to see people exploring fairly uncharted territory in FRC. There have been very few circular robots, and even fewer that incorporate a competitive swerve drive. Best of luck with further development on this endeavor and any resulting competition robots.

I certainly don't have the experience you're looking for here, but I do have some useless conjecture to contribute.

I think there is a decent argument for going faster, "beyond" a traction-limited gear ratio, for this base. The lack of turning scrub in almost all situations really reduces the amount of load you'll put on a battery and the need to slip wheels at low currents. (I think this is why a lot of 8-motor coaxial swerve designs have been popping up?) The circular design of the frame and the smooth sailcoth is going to make pushing a robot without rolling off very difficult, so initiating pushing matches is going to be difficult. As for pushing through defense, is that something this robot will ever have to do? It seems like the robot is purpose built to avoid pushing at all costs, finding a way to slip out of any pin. I'm not sure if the situations you need to brute force your way out of will happen enough to justify compromising the agility you use to avoid those defenders. Perhaps I'm not thinking enough about the various defensive scenarios this robot could encounter, such as being pinned in a corner or by multiple robots.

All that said, from what I have heard (again, little experience in swerve here) it's easiest to start swerve development at a lower top speed and inch things upward. And swerves can be agile and competitive at surprisingly low top speeds, relative to a standard drive needing to start and stop more often. So it's certainly an open question for debate. The Wild Swerve system has a variety of gear ratios for you to choose from, and if you incorporate Vex gears into the mix you can get even more flexibility, so I'm sure this is the kind of thing you'll play with and tune if you build this in-season.

I'm excited to see how this plays out in real matches, if you guys end up going with this design. Any idea how an arm or other manipulator would fit on top? Perhaps even a turreted one, if you're willing to go that complex?

[asid61]

Originally Posted by Chris is me

I think there is a decent argument for going faster, "beyond" a traction-limited gear ratio, for this base. The lack of turning scrub in almost all situations really reduces the amount of load you'll put on a battery and the need to slip wheels at low currents. (I think this is why a lot of 8-motor coaxial swerve designs have been popping up?) The circular design of the frame and the smooth sailcoth is going to make pushing a robot without rolling off very difficult, so initiating pushing matches is going to be difficult. As for pushing through defense, is that something this robot will ever have to do? It seems like the robot is purpose built to avoid pushing at all costs, finding a way to slip out of any pin. I'm not sure if the situations you need to brute force your way out of will happen enough to justify compromising the agility you use to avoid those defenders. Perhaps I'm not thinking enough about the various defensive scenarios this robot could encounter, such as being pinned in a corner or by multiple robots.

+1 to all of this, and something else to consider. If you just want to play heavy defense, insttead of pushing against the opponent's robot, you can just lock your wheels into an X formation and sit there. Nobody can push past that without lifting your entire robot. /then when they move, you move too. You can gear it as high as you want too.

[Deke]

We had some experience playing against a round swerve last year that 2767 stryke force had, and there were pros and cons to it IMO.

It was very hard to t bone and push the drive around, the circular base did a good job of rolling off the t bones. However, it was fairly easy to stay in front of them to block their path since the robot was so big. The diameter was optimized to meet the frame perimeter, so it had a fairly large footprint on the field.

A final observation is it was hard for them to line up balls to push in the low goal. The large diameter didn't fit well in the corner and sometimes it took awhile to maneuver where they needed to be in the corner of the field.

[Gregor]

Another thing to keep in mind with circular robots is they won't fit through a standard doorway with wheels on the floor (with 112" frame perimeter). If you were to build a tall circular robot you might have a problem.

[Abhishek R]

Originally Posted by Infinity2718

We had some experience playing against a round swerve last year that 2767 stryke force had, and there were pros and cons to it IMO.

It was very hard to t bone and push the drive around, the circular base did a good job of rolling off the t bones. However, it was fairly easy to stay in front of them to block their path since the robot was so big. The diameter was optimized to meet the frame perimeter, so it had a fairly large footprint on the field.

A final observation is it was hard for them to line up balls to push in the low goal. The large diameter didn't fit well in the corner and sometimes it took awhile to maneuver where they needed to be in the corner of the field.

I would imagine the converse is true as well, a circular robot would not be able to friction pin a rectangular one.

[Joe Johnson]

Originally Posted by Abhishek R

I would imagine the converse is true as well, a circular robot would not be able to friction pin a rectangular one.

Yes, I think you're right but swerve drives do still have an advantage in a pushing match: the ability to have all the pushing vectors pointing exactly in the direction that the driver wants them.

I think with enough driver practice, this can be a net positive when playing defense, more than making up for the slippery bumper materials and the roundish shape. Time will tell.

One thing that people have already pointed out is that a roundish robot with a 112" frame perimeter has a diameter that is bigger than typical FRC robots. This means we'll take up more space on the field (Probably not a good thing). More bothersome is that the robot can't fit through standard doors even with the bumper removed. For Overclocked, that is a major pain as we have single door access to our workshop. Could be a deal breaker for our team...

Dr. Joe J.

[MrBasse]

I agree that Strykeforce's round swerve was impressive to play against. After some time behind the controls they seemed much more able to do what they wanted to do with their robot. Pinning them was near impossible, and thy were able to keep us stuck in low gear a fair amount of the time we were against them to try to push through. they were able to simply roll along with us and keep up. The only issues I saw them face was lining up to things and attempting to push the ball.

[sduffley]

Hey all. For those curious, the CAD can be found here as a STEP: https://dl.dropboxusercontent.com/u/..._frame_asm.stp

[TCMJ1816]

Looks great, We have prototyped with wildswerve too. We also had trouble with the wheel mounting system on the bottom plate, so we laser cut Lexan rings that span between all of the mounting points. images here (sorry about the potato quality). They worked like a dream, so much better than the included black blocks. They did a great job dealing with the scrub that comes from the wheels being even just a few degrees off.