Drive Train Choices

[efoote868]

Originally Posted by DampRobot

Butterfly, swerve, and to a large degree mecanum are comparatively hard to learn to drive in a short amount of time.

I strongly disagree. The easiest drive to learn in a short amount of time is some form of field-centric omnidirectional drive. Just point your joystick in the direction you want your robot to go, and it goes.

[T^2]

Originally Posted by efoote868

I strongly disagree. The easiest drive to learn in a short amount of time is some form of field-centric omnidirectional drive. Just point your joystick in the direction you want your robot to go, and it goes.

I think Damp was assuming that the team was somewhat new, and therefore not quite at the point where they could program good field-centric drive code.

[KrazyCarl92]

Originally Posted by efoote868

I strongly disagree. The easiest drive to learn in a short amount of time is some form of field-centric omnidirectional drive. Just point your joystick in the direction you want your robot to go, and it goes.

Even with a field-centric drive, I have witnessed many first-time drivers struggle to grasp omnidirectional driving in the time they can easily grasp simple arcade drive. It's the mind set of properly using the drive train by translating in multiple axes rather than always rotating to change direction that most seem to struggle with. Re-orienting the robot to change direction is intuitive since that's how most things we are used to work (humans, cars, etc.). Keeping the robot facing a constant heading while changing direction is counter-intuitive for most, while it may seem easy for others.

[DampRobot]

Originally Posted by KrazyCarl92

Even with a field-centric drive, I have witnessed many first-time drivers struggle to grasp omnidirectional driving in the time they can easily grasp simple arcade drive. It's the mind set of properly using the drive train by translating in multiple axes rather than always rotating to change direction that most seem to struggle with. Re-orienting the robot to change direction is intuitive since that's how most things we are used to work (humans, cars, etc.). Keeping the robot facing a constant heading while changing direction is counter-intuitive for most, while it may seem easy for others.

Exactly.

I've never driven an omidirectional robot, but have talked to and observed many of their drivers. The best ones (1640, 1717, etc.) all highly prioritize driver practice, and through this extra work can bring out the advantages associated with the extra maneuverability. However, the worst ones just drive their mecanum or kiwi drive as if it was a glorified tank drive with huge wheels and rollers.

I have driven both Cheesy Drive (arcade) and regular tank, and can say that even Cheesy was more difficult to learn than tank. With tank, pretty much any old kid can walk up to the driver station and start driving, while with Cheesy drive, you have to take a few moments to figure it all out. There just appears to be something extremely intuitive about one joystick controlling one side of the drivetrain. However, I will admit that Cheesy Drive and other non-tank driver setups have higher ceilings than tank. Drivers (with practice) can do things with those systems that are amazing.

But just like in drive trains, to take advantage of fancy driver setups, you need experienced and practiced drivers.

[magnets]

Originally Posted by DampRobot

Exactly.

I've never driven an omidirectional robot, but have talked to and observed many of their drivers. The best ones (1640, 1717, etc.) all highly prioritize driver practice, and through this extra work can bring out the advantages associated with the extra maneuverability. However, the worst ones just drive their mecanum or kiwi drive as if it was a glorified tank drive with huge wheels and rollers.

Exactly. I know for a fact that 1640 has their drivers practice on a robot from the first day of build season, to the last day of their competition season.

[Ether]

Even with a skid-steer vehicle (6WD for example) it is straightforward to program a driver interface wherein the vehicle will turn in an arc so as to go in the direction of the joystick angle with a speed proportional to the joystick radius. For example: http://www.chiefdelphi.com/media/papers/2438

With an omnidirectional drive (like mecanum or omni or swerve), it can be programmed to immediately go in the direction of the joystick angle while simultaneously rotating to align itself with the direction of travel. Or, with the push of a button, instead go in the direction of the joystick angle without aligning.

[efoote868]

Originally Posted by DampRobot

The best ones (1640, 1717, etc.) all highly prioritize driver practice, and through this extra work can bring out the advantages associated with the extra maneuverability.

I'm not aware of any powerhouse team that doesn't prioritize driver practice, which is to say I doubt the "extra" work the best omni-directional teams do is really "extra" at all.

What I witnessed in 2010 was that once field-centric drive was demonstrated (this robot doesn't know it has a "front" side), people with no driving experience had the robot moving around the room much faster than I witnessed others with driving experience during 2008, all without accidentally banging into desks, chairs, or freshmen.


From what I've seen and what I've read on CD, there are many teams that have poor experiences with holonomic drive trains because they don't use field-centric drives; field-centric makes a world of difference.

Originally Posted by Ether

Even with a skid-steer vehicle (6WD for example) it is straightforward to program a driver interface wherein the vehicle will turn in an arc so as to go in the direction of the joystick angle with a speed proportional to the joystick radius. For example: http://www.chiefdelphi.com/media/papers/2438

I actually had that in the back of my mind when I was posting about field-centric, but I was trying to be concise

[Andrew Schreiber]

Originally Posted by efoote868

From what I've seen and what I've read on CD, there are many teams that have poor experiences with holonomic drive trains because they don't use field-centric drives; field-centric makes a world of difference.

I've built and driven a few omni-directional drives in my day and have never put field centric code on them. Rarely had major problems driving them. However, many students have had problems. A lot of it will come down to the driver.

I guess I need to sit down and make a vex kiwi drive and put field centric drive on it.

[apalrd]

I dunno about you guys, but several of our Vex robots ran non-field-centric slide drives and at least one of the drivers was able to master it fairly quickly and drive it well without too much practice.

The hardest part for me about omnidirectional test chassies is usually that I can't figure out which side is the front, because it's just a chassis

[nicholsjj]

Originally Posted by apalrd

The hardest part for me about omnidirectional test chassies is usually that I can't figure out which side is the front, because it's just a chassis

That's when you mount a VEX flag on the chassis and call it the front

[Clem1640]

Originally Posted by MICHAELABICK

I highly reccomend perfecting your 6wd before you look any further into special drivetrains. A good 6wd will carry your team for many years. Once you have a solid drive, you can expand from there, based on what your team wants.

While mecanum, swerve, octocanum, and nona(and many others) are known to provide extra "manuverability", I personally find that this manuverability is often not used correctly, and better performance will usually come from more driver practice instead. I think it would be more worthwhile your time to look into a drivetrain that uses less resources or improve your current drivetrain(reduce resources too).

One promising drivetrain is butterfly. Butterfly can be built using very little resources. In addition, it offers performance increases over typical 6wds like getting out of t bones, straighter tracking for auton, and being unturnable(also butterfly can be built to have two different speeds).

There are many additions you can do to a 6wd too. We personally have added motors, used wider wheels, reduced resources to build our 6wd(primarily using new COTs parts), and reduced weight. Other notable additions that we are testing during the offseason are belts and drop down omnis/ball casters to get out of tbones.

As a team that likes swerve and has done swerve for the past four years, I would strongly recommend building your base competencies with 6wd first. 6wd is pretty darned good. Master this first. After you've got 6wd nailed, expand to other options.

We use off-season (summer-fall) projects to experiment with new concepts, both drive-train and in other areas. This allows us to build institutional knowledge outside the frantic pace of build season.

[Oblarg]

Originally Posted by Clem1640

We use off-season (summer-fall) projects to experiment with new concepts, both drive-train and in other areas. This allows us to build institutional knowledge outside the frantic pace of build season.

This, here, is the crux of good FRC drive development. The time for crazy ambition and reach-goals is the offseason, when you have neither strict time constraints nor immediate demand for success.

As a rule for drive, I promote the following: During build season, you should never consider a drive system which you have any doubt about your ability to execute. Any sufficiently complicated drive system should not even be on the table for build-season unless you have successfully executed it within the working memory of the team. The best way to build up your catalog of usable drive designs is thus to devote time in the off-season to experimentation with new designs.

Keep in mind that it is human nature to be overly-optimistic (often to silly extents), and that you must make a conscious effort to place your judgment of what you are able to do significantly below your initial feeling. This holds true for all design goals, but is especially pertinent for drive. For a drive, reliability trumps every single other concern, no exception. If your robot is unable to move, you are not able to play the game.

[KrazyCarl92]

Originally Posted by Oblarg

For a drive, reliability trumps every single other concern, no exception. If your robot is unable to move, you are not able to play the game.

Rules compliance. Sadly this is overlooked nearly as often as reliability. I would say those are the two criteria that every single drive train in FRC should meet. The kit bot meets both if done according to directions, so there is truly very little reason for a drive train that is not reliable and rules compliant.

[magnets]

What oblarg said about reliability is very true. Unless you are one of those teams who has really mastered swerve (111, 1640, 1717) you won't see any benefits. Swerve is just so hard to do right. You don't get drive practice time until the season is almost over, and unless your swerve modules are 100% perfect in every way possible, you'll have one fail at competition, and you'll drive in circles for a match! For 95% of teams, the kitbot drive setup (or at least a simple 4/6wd) is the way to go.