Chief Delphi

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)
-   Technical Discussion (http://www.chiefdelphi.com/forums/forumdisplay.php?f=22)
-   -   Swerve Drivetrain. (http://www.chiefdelphi.com/forums/showthread.php?t=93755)

Mk.32 19-03-2011 01:21

Swerve Drivetrain.
 
Hello all, Team 3647 here.
We are done with our 2011 season and moving to prepare for our next season.
One thing we saw that really neat was swerve drive systems.
So we want to build one on a test bot to experiment and test.

After looking around we found the andymark modules:
http://www.andymark.com/ProductDetai...ctCode=am-0760
We also have a basic idea of how rig everything up.

We are planning on having the left side wheels and right side wheels hooked to separate power trains. Then the top right and bottom left wheels linked together for turning and then the top left and bottom right linked together for turning. So we can have all the wheels at 45 degree angles so we can spin in place and then strafe with turning the wheels 90 degrees.

So my question what do you guys think? How do you run your swerve drives? Also how do you have your controllers setup? What gearing ratio do you use? Also any documentation would be greatly appreciated.

Thanks for all the help.

big1boom 19-03-2011 01:33

Re: Swerve Drivetrain.
 
We have done Right/Left Power split, Front/Back turning. It allows translation in any direction, good turning ability, on the spot rotation like a tank drive, and leaves the middle open for a manipulator.

We have also done all 4 powered together, Front/Back turning. Simple to control, difficult to chain up.

I would recommend Right/Left Power, Front/Back turning.

EricH 19-03-2011 01:37

Re: Swerve Drivetrain.
 
There are quite a few ways to do a swerve drive. There's fully independent steering (each module has its own turning motor). There's all turning at once. (This one really needs some other way to turn--or a turret, noting 118's 2007 robot.) There's front/back and left/right module linkages.

I don't think your turning solution is optimal--it'll work for turning in place, but it could be really nasty for anything else, programming-wise, especially if you get a wheel off from the rest by mistake. And the chain runs are going to be especially interesting--you're forming an X with the chains, and hopefully you find a good way to do that without risking fouling up. I've seen that done before, with 4 runs, and it worked OK, but I think it was the propulsion power runs from the center to the edge and the rotation runs on the outside.

What I might do, as an experiment (and I don't know if anybody's fully run this on coaxial), is to turn all the pods together, but have left/right or front/back motor units. Think of it like tank drive where the wheels can pivot for sliding sideways. It's somewhat easier to do with the Wild Swerve module (also available from AM/221 Robotics Systems), but could be really interesting to try a coaxial version.

Incidentally, I've never been on a team that used swerve drive. The closest was 2009, two years after I left, and if there hadn't been a trailer towing behind the robot it would have been a turreted robot instead of a pivoting drivetrain.

Mk.32 19-03-2011 01:45

Re: Swerve Drivetrain.
 
At the SD regional Team 1717 had a really nice swerve drive and a others i don't remember atm.
What is the difference between the Wild Swerve module and the ones i linked to in the OP?
And sliding sideways, we were thinking about that, but would it work?

EricH 19-03-2011 01:59

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by Mk.32 (Post 1041936)
At the SD regional Team 1717 had a really nice swerve drive and a others i don't remember atm.
What is the difference between the Wild Swerve module and the ones i linked to in the OP?
And sliding sideways, we were thinking about that, but would it work?

The difference is that the Revolution modules are coaxial, and the Wild Swerve have their motor in the module. (Oh, and they're based off two different teams' designs--Revolution is from 118 Robonauts, and Wild Swerve comes from 111 WildStang.)

Coaxial swerve means that the shaft located at the pivot axis (which the module tends to rotate around) also transmits power to the drive wheel. The motor is completely off the module, other than the chain run. Non-coaxial means that you have to be careful not to twist the motor power wires as the module turns (unless the GDC allows slip rings again), as the motor is in the module. Less chain, but more wire.

Sliding sideways=strafing. Sorry I didn't make that clear. And if you're not going to do that with a swerve drive, then you may as well not build a swerve drive...

Joe G. 19-03-2011 02:02

Re: Swerve Drivetrain.
 
The revolution module that you linked is what is known as a coaxial swerve module, meaning that the motors are mounted vertically, parallel to the module's rotation axis. The wild swerve module contains the CIM motor, and gear reduction, within the swerve module itself. The primary advantage of coaxial modules is infinite rotation, as there are no electrical components within the modules. Wild swerve style modules are limited by the wires running to the motors, and must use sensors to limit their rotation. This limited rotation can also cause changing orientation to take more time than it might otherwise.

Non-coaxial modules, by nature, are independently powered, which can open up some flexibility in how the swerve is operated. For example, tank style turning will be much easier to accomplish in any orientation on a non-coaxial setup.

Finally, here are some teams that have produced quality crab and swerve drives over the years. Look into some of their past robots, its a great way to learn!

16 (most years)
71 (most years)
111 (most years)
118 (most years, but not the past couple)
148 (2008)
1114 (2004)
1640
1717

cire 20-03-2011 20:18

Re: Swerve Drivetrain.
 
I recently made a vex drive train that had 4 independently driven wheel. This allows it to have multiple drive modes because the wheels can be programmed to be at whatever angles you wish. I created 4 drive modes based off of a paper in 1640's wiki.

You can see a video of the robot working here: http://www.youtube.com/watch?v=s3gb5n_I7vc
I have it set up so I can switch between Twitch(tank), crab, automobile, and snake modes on the fly.

I would strongly recommend reading team 1640's Wiki and this paper before deciding on how you are going to power the wheels.

TEE 20-03-2011 20:28

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by Joe G. (Post 1041943)
Finally, here are some teams that have produced quality crab and swerve drives over the years. Look into some of their past robots, its a great way to learn!

16 (most years)
71 (most years)
111 (most years)
118 (most years, but not the past couple)
148 (2008)
1114 (2004)
1640
1717

Team 1625 had a 6-wheel swerve drive last year ;)

apalrd 20-03-2011 20:37

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by EricH (Post 1041941)
...(unless the GDC allows slip rings again)...

Not true.

Quote:

Originally Posted by FRC Game Manual, Section 4 (The Robot), Rev K
<R40> All active PD Board branch circuits shall be wired with appropriately sized wire:
Application
Minimum wire size...<snipped table>
The branch circuit may include intermediate elements such as COTS connectors, splices, COTS flexible/rolling/sliding contacts, and COTS slip rings, as long as the entire electrical pathway is via appropriately gauged conductors.

If they are COTS, and appropriately gauged, they're fine.

If you want to build a swerve drive, build one in the fall. You probably won't be able to do it in 6 weeks without prior experience in swerve-building.

Gdeaver 20-03-2011 20:56

Re: Swerve Drivetrain.
 
After a great debate, 1640 did pivot drive again this year. We cleaned sum stuff up and made easy to swap out modules. Still a few tweaks since our first comp. Programming also was cleaned up from last year. Drives great. Advice for any team thinking about going down this path. Start now and design - prototype. This is a very resource taxing endeavor. Evaluate your resources. If you don't have them do not do it. You need to go into a kick off knowing how to do this drive train. Deciding week two to do pivot drive I guarantee will lead to a heap of frustration. Link to this years effort.
http://wiki.team1640.com/index.php?t...II_Drive_Train
When everything is tweaked pivot drive is a joy.

Andrew Schreiber 20-03-2011 20:59

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by Gdeaver (Post 1042841)
After a great debate, 1640 did pivot drive again this year. We cleaned sum stuff up and made easy to swap out modules. Still a few tweaks since our first comp. Programming also was cleaned up from last year. Drives great. Advice for any team thinking about going down this path. Start now and design - prototype. This is a very resource taxing endeavor. Evaluate your resources. If you don't have them do not do it. You need to go into a kick off knowing how to do this drive train. Deciding week two to do pivot drive I guarantee will lead to a heap of frustration. Link to this years effort.
http://wiki.team1640.com/index.php?t...II_Drive_Train
When everything is tweaked pivot drive is a joy.


And read EVERYTHING on 1640's wiki. I was working on a pivot drive as a class project and I cannot tell you how useful their website is. Thanks guys, you saved me a ton of work.

EricH 20-03-2011 21:11

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by apalrd (Post 1042813)
Not true.
If they are COTS, and appropriately gauged, they're fine.

Read the entire post. Note that I said "GDC allows" and "again". And know your FRC rules history. Slip rings are an on-again, off-again legal part. (190 had custom ones in 2008 that were legal--there wasn't a prohibition on custom ones then. COTS ones were not legal at that time, IIRC.) COTS ones even more so. This is one of the few times that I remember COTS slip rings being legal for competition use.

And note that the rules change every year.

So, IF the GDC chooses to allow slip rings again, then you don't have to worry about motor wires twisting. Otherwise, you do.

Mk.32 20-03-2011 21:33

Re: Swerve Drivetrain.
 
That team 1640 doc was exactly what i was looking for, information wise.
Thank you very much.
If we do go this route we are definitely building a test robot off-season first.
And what rule exactly is it that dis-allows slip rings?

Clem1640 20-03-2011 21:42

Re: Swerve Drivetrain.
 
2011 is our 2nd year of doing Pivot Drive. 1640 is very satisfied with the performance of this drive-train. Our drivers love it! We've designed and built our own. Info is available at:
http://wiki.team1640.com/index.php?t...II_Drive_Train

1640 has taken the approach of independently driving and steering each wheel. This puts a lot of investment into the drive-train, but also provides a fly-by-wire flexibility. Drive modes are programmable. The robot has mecanum agility (or better) with tank traction. A hell of a combination.

This is a great drive-train, but not easy to build or program. Many teams which start a Pivot program fail and abandon it. The AndyMark modules are probably a good starting point.

GGCO 20-03-2011 21:50

Re: Swerve Drivetrain.
 
We're using the AM swerve modules, and they're fantastic! The hardest part though is the code base. Make sure you have a TON of time and good programmers to do this.

EricH 20-03-2011 22:26

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by Mk.32 (Post 1042870)
And what rule exactly is it that dis-allows slip rings?

This year, there is not one. In fact, they are explicitly allowed.

In past years, rules varied on legality. In future years, the rules may or may not allow them. Hopefully, they will be allowed. But, we can't be certain.

Mk.32 20-03-2011 22:34

Re: Swerve Drivetrain.
 
k that makes sense.
Thanks for the clear up.

Mk.32 20-03-2011 22:39

Re: Swerve Drivetrain.
 
Also another question,
with the wild swerve drive modules
http://www.andymark.com/ProductDetai...ctCode=am-0496
do we need to blot both the top and bottom plates down?
or would just blotting the top plate be enough?

cire 20-03-2011 23:18

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by Mk.32 (Post 1042922)
Also another question,
with the wild swerve drive modules
http://www.andymark.com/ProductDetai...ctCode=am-0496
do we need to blot both the top and bottom plates down?
or would just blotting the top plate be enough?

You need to bolt both the top and bottom plate, otherwise it will just bend the top bracket.

Ether 20-03-2011 23:19

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by GGCO (Post 1042893)
The hardest part though is the code base. Make sure you have a TON of time and good programmers to do this.

There are a couple of short papers here which explain the kinematics of, and give example code for, 4-wheel-independent steering and speed control.



PAR_WIG1350 21-03-2011 00:49

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by Joe G. (Post 1041943)
The revolution module that you linked is what is known as a coaxial swerve module, meaning that the motors are mounted vertically, parallel to the module's rotation axis. The wild swerve module contains the CIM motor, and gear reduction, within the swerve module itself. The primary advantage of coaxial modules is infinite rotation, as there are no electrical components within the modules. Wild swerve style modules are limited by the wires running to the motors, and must use sensors to limit their rotation. This limited rotation can also cause changing orientation to take more time than it might otherwise.

Non-coaxial modules, by nature, are independently powered, which can open up some flexibility in how the swerve is operated. For example, tank style turning will be much easier to accomplish in any orientation on a non-coaxial setup.

Finally, here are some teams that have produced quality crab and swerve drives over the years. Look into some of their past robots, its a great way to learn!

16 (most years)
71 (most years)
111 (most years)
118 (most years, but not the past couple)
148 (2008)
1114 (2004)
1640
1717

:confused: I was under the impression that coaxial module swerve was (usually*) the more flexible of the two due to the deletion of electrical connections to the module thus allowing the module to rotate directly to the desired position (for example, a 10 degree rotation in a coaxial setup could translate into a 350 degree rotation in a MOM (motor on module) setup). I am currently working independently on coaxial swerve and how to unlock it's full potential. Technically, due to the allowance of slip rings this year, MOM swerve is superior due to the greater efficiency that can be gained by eliminating the 90 degree transmission necessary for coaxial modules and the removal of MOM's limitations by the slip rings allowing unlimited rotation of the modules. One key design feature to fully utilize the abilities that swerve can give you is the independent steering and drive of all modules. After venturing into the 'branch' of swerve that I have, It is difficult to look at front/back // left/right linked swerve and say that it is even approaching all of the benefits that a fully functional swerve drive train has to offer. (NOTE: I plan on releasing that white paper I promised eventually).

buchanan 21-03-2011 15:22

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by PAR_WIG1350 (Post 1042973)
:After venturing into the 'branch' of swerve that I have, It is difficult to look at front/back // left/right linked swerve and say that it is even approaching all of the benefits that a fully functional swerve drive train has to offer.

A good point. While it's unlikely anyone would build a tank/skid steer drivetrain without a software/control system capable of exploiting all its physical capabilities, the same cannot be said of fancier systems like swerve and omni/meccanum. It's entirely possible to build one of these that exploits some, but not all, of what's physically possible with its hardware.

You can look at a 6wd, tank, etc machine and know pretty well what kind of maneuvers it's going to be able to make (speed aside). You can't make the same assessment of a swerve or meccanum platform without knowing what's in the software.

apalrd 21-03-2011 15:33

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by buchanan (Post 1043235)
...While it's unlikely anyone would build a tank/skid steer drivetrain without a software/control system capable of exploiting all its physical capabilities...

I disagree.

Including the ride height automation, I have almost as much code in the drivetrain of our robot as I do in the elevator.

Ignoring the ride height automation, I still have a lot of code. The driver inputs come in, are run through filters if certain conditions are met (to provide better response and less oversteer), can be flipped if the driver requests invert, and are fed through a special algorithm that handles turning and arcing better (to fix some issues we found where the driver requests a (1,0) but really means (1,coast). We then run the commands to the closed-loop speed control (which in itself has automation of coasting to stops and holding position), and then run the shift scheduler.

The simple way of programming a skid steer would be to set the motors to the joysticks.

Joe G. 21-03-2011 15:55

Re: Swerve Drivetrain.
 
Quote:

Originally Posted by PAR_WIG1350 (Post 1042973)
:confused: I was under the impression that coaxial module swerve was (usually*) the more flexible of the two due to the deletion of electrical connections to the module thus allowing the module to rotate directly to the desired position (for example, a 10 degree rotation in a coaxial setup could translate into a 350 degree rotation in a MOM (motor on module) setup). I am currently working independently on coaxial swerve and how to unlock it's full potential. Technically, due to the allowance of slip rings this year, MOM swerve is superior due to the greater efficiency that can be gained by eliminating the 90 degree transmission necessary for coaxial modules and the removal of MOM's limitations by the slip rings allowing unlimited rotation of the modules. One key design feature to fully utilize the abilities that swerve can give you is the independent steering and drive of all modules. After venturing into the 'branch' of swerve that I have, It is difficult to look at front/back // left/right linked swerve and say that it is even approaching all of the benefits that a fully functional swerve drive train has to offer. (NOTE: I plan on releasing that white paper I promised eventually).

My statement is based around the fact that non-coaxial modules are, by nature, independently powered, while many coaxial swerves power the modules in pairs, or even from a single "power plant" swerve, as 118 often does. Putting multiple modules on a single power source limits some of the motions you can make.

For example, say you have your modules steered in front/back pairs, and powered in left/right pairs, as many (most?) coaxial swerves do. This setup excels at akerman-style steering. But tank style steering is impossible in any module orientation other than "forward," and in long orientation, with four traction wheels, may even be difficult there. Motor-in-module setups, however, would allow tank-steering in any orientation; forward, sideways, and probably diagonally to some degree.

This downside of coaxial is considerably more apparent in a 118 style setup. With all wheels tied to the same motor, there can be no variance in their speed. This makes any kind of tank steering just about impossible, and for this reason, all robots that I know of built this way have been either turreted, or had no real "front" (think 148 in overdrive). Which is something that a team may or may not be able to fit into their design.

Admittedly, it's not a huge downside. And you're correct, ignoring the possibility of slip rings, having infinite rotation of the modules is a big plus for coaxial. If you can do a independently powered, independently steered "pivot drive," more power to you, but that's not an easy undertaking. Which system is right? It depends on your robot, what you have experience making, and which set of pros and cons you think is best.

The Analog 21-03-2011 17:34

Re: Swerve Drivetrain.
 
this year we (Team 79 KRUNCH) went with swerve drive again after a few years away from it. We are using a coaxial system with all 4 wheels driven independently and the front and rear wheels are steered separately (fronts steer together and rear steer together). So far this year, this has proved to be a very versatile drivetrain.


All times are GMT -5. The time now is 16:02.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi