Driving: Tank, Arcade or FPS?

[Ether]

Question for those teams using XY on one Joystick and rotation on the other: do you use the left or the right joystick for XY?

[RunawayEngineer]

Originally Posted by JamesCH95

Sure, lots of practice always helps. Perhaps that is not the best anti-two-stick-tank argument.

Still, how can a two-stick system drive straight at part-throttle? This is VERY difficult to do.

Consider this thread from last year: http://www.chiefdelphi.com/forums/sh...d.php?t=125759

Team 33 found that all drivers wound up preferring FPS-style control to two-stick tank and found it "way easier" to control their robot(s).

Well, I was a base driver in 2008 - so if you are familiar with the game, then you know that driving at less than full throttle was not a scenario that I encountered
I have no question that 2-stick tank requires finesse; but from running driver training with RC controls, I know that that is not a unique situation.
If people find something that they like, more power to them. I haven't discounted any method of control.

[JamesCH95]

Originally Posted by RunawayEngineer

Well, I was a base driver in 2008 - so if you are familiar with the game, then you know that driving at less than full throttle was not a scenario that I encountered
I have no question that 2-stick tank requires finesse; but from running driver training with RC controls, I know that that is not a unique situation.
If people find something that they like, more power to them. I haven't discounted any method of control.

Have you found a story of any team that switched to two-stick tank from any other method of control? I can only find threads like this, where teams ditched two-stick tank for another method they found to be superior.

[cadandcookies]

Here was 2220's progression (with skid steer robots):

2012: Joystick tank drive
2013: Controller-based tank drive
2014: Controller - based FPS drive

Whatever paradigm of control you use (tank, arcade, FPS), for a lot of kids (at least in my experience), a controller is most intuitive. And FPS is almost always most intuitive from a controller. At least in my experience/opinion. Personally I'd shy away from telling the driver how they'll be controlling the robot

[zinthorne]

Originally Posted by Ether

Question for those teams using XY on one Joystick and rotation on the other: do you use the left or the right joystick for XY?

In most video games with an xbox controller the left stick controls turning and direction and the right controls your view. (Like which way you are looking)

I am a big forza fan and other fps games. Forza drives with the triggers controlling forward and brake and reverse. Left is reverse/break, and right is forward. The left joystick controls turning and the right stick is generally not used much, because if you are doing manual transmission you must use x and b for shifting. In Halo it is pretty much the same with the right stick being used to control where u look.

I personally am going to drive with an xbox controller if the proggramers will let me. (We have no previous experience with one) I would like to merge both forza and halo and us the two triggers to drive and control speed. (Forward and reverse are only on the triggers) and then use the left stick for turning. Then the right joystick would be used to control all driving. (Forward, backwards, turn left, turn right) When used though the robot would only run at 25-50% speed, while all other drive controls are then not used during that time. The low speed would allow for fine movements without worry of over correcting or accelerating too fast when you are in a crammed place.

[RunawayEngineer]

Originally Posted by JamesCH95

Have you found a story of any team that switched to two-stick tank from any other method of control? I can only find threads like this, where teams ditched two-stick tank for another method they found to be superior.

Honestly, no.

[IronicDeadBird]

On a semi related note does anyone have experience with creating there own controllers or joysticks if you could sling me a PM that'd be swell

[JamesCH95]

Originally Posted by RunawayEngineer

Honestly, no.

I think that 33's experience (documented in a thread I linked earlier) is basically a microcosm of all of FRC's experience with two-stick drive. A lot of drivers think that it's intuitive at first, so they ask for it without knowing any better. However, once they get used to 'split arcade' or 'fps' or some other more sophisticated control scheme (where the computer is doing the drive control math instead of the driver, where drive transforms can be used) they wind up preferring it to two-stick. I believe this is why teams rarely, if ever, change back.

[Gregor]

Originally Posted by Ether

Question for those teams using XY on one Joystick and rotation on the other: do you use the left or the right joystick for XY?

XY is on the left joystick, turning on the right.

[JesseK]

Originally Posted by JamesCH95

As a (relatively) old-school driver I liked single-stick arcade control for FRC robots. ...

I absolutely cannot stand 'tank drive' where two joysticks are used. It forces the driver to do the drive 'math' in their head, something a computer is MUCH better suited to do. I never felt that it was intuitive, fluid, or an efficient use of controllers. I know some people/teams have been successful with it, but I can only imagine how much more successful they could have been with any of the controls described above.

I too am somewhat old-school, preferring single-joystick arcade for things like the Descent series, several racing games and a whole slew of flight simulators from Falcon 3.0 to Lockheed's F-35 simulator. Yet those are all first-person simulators with relatively accurate perspective-based flying modes. Except for Descent*, they are also simulators where going in reverse and/or 0-radius turns are not plausible.

Personally, when driving robots in 3rd-person, (SeaPerch, FTC, FTC, QuadCopters, and even a R/C fork lift), I find it far more intuitive to do the 0-radius turns and the forward/reverse via 'tank drive'. I also know that not every turn is 0-radius, particularly in warehouse operations where 0-radius means a large movement of the load at the extremes of the robot. In these situations, I can most definitely maintain finer direct tank control over the object than any software (that isn't fully autonomous with 3D sensing capability) under two conditions: (1)the inputs are scaled to 'S' curves rather than linear (2) I can maintain visual contact with the vehicle and the objects around it. QuadCopters, of course, are a bit trickier since there are a variety of types of flight control available and usually full autonomous is almost always preferred.

Yet this isn't an argument for/against tank drive - but rather an anecdote that each driver will have preferences after trying the different styles. Hopefully the software team isn't force-feeding something to the driver, but rather letting multiple potential drivers try out the different modes.

*Come to think of it, I could probably write a joystick emulator that converted 2-joystick 'Tank' to Descent's 1-joystick 'Arcade' to see what how the experience changes.

[MrForbes]

Originally Posted by JesseK

Yet this isn't an argument for/against tank drive - but rather an anecdote that each driver will have preferences after trying the different styles. Hopefully the software team isn't force-feeding something to the driver, but rather letting multiple potential drivers try out the different modes.

That's the whole point of this thread.... if you don't do some research, or ask the right questions, you're mostly "force fed" the options of tank drive or arcade with the canned drive software we get.

[JamesCH95]

Originally Posted by JesseK

Yet this isn't an argument for/against tank drive - but rather an anecdote that each driver will have preferences after trying the different styles. Hopefully the software team isn't force-feeding something to the driver, but rather letting multiple potential drivers try out the different modes.

I think it has to go beyond "trying it [drive mode] out" to where the driver(s) develop some level of proficiency. Some, like you, find it intuitive, which is quite appealing initially. But if a little bit of time is spent get build some proficiency I think that other options are likely to be preferred by a driver.

[DarkRune]

Originally Posted by Ether

Question for those teams using XY on one Joystick and rotation on the other: do you use the left or the right joystick for XY?

XY on the left joystick, turning Z on the right joystick.

Originally Posted by JamesCH95

I think that 33's experience (documented in a thread I linked earlier) is basically a microcosm of all of FRC's experience with two-stick drive. A lot of drivers think that it's intuitive at first, so they ask for it without knowing any better. However, once they get used to 'split arcade' or 'fps' or some other more sophisticated control scheme (where the computer is doing the drive control math instead of the driver, where drive transforms can be used) they wind up preferring it to two-stick. I believe this is why teams rarely, if ever, change back.

This is my personal experience, but after switching to arcade and split arcade, I was really unsatisfied since it seemed at least to me to be more work and math (though it probably isn't), whereas tank (for me) seemed to be completely intuitive and required no mental math. The problem I had with split arcade was how turning was completely independent of of the other drive joystick. So I tried out arcade and I liked it a little, but there was more of a thought process for what angle to press the joystick at, how much to turn, etc. With tank this doesn't require any thinking.

[Ken Streeter]

Originally Posted by JamesCH95

Have you found a story of any team that switched to two-stick tank from any other method of control? I can only find threads like this, where teams ditched two-stick tank for another method they found to be superior.

Over the past few years, we've used both two-stick tank drive and the FPS-style drive (although we call our version "Speed Racer Drive" even though it isn't actually the drive we used for Speed Racer in 2008 - for Speed Racer we actually used a Steering wheel and foot pedal).

However, there is one maneuver that the two-stick tank drive makes easy that I haven't found a good way to replicate on the FPS drive -- the "pivot turn."

For two-stick "tank drive" on a skid steer, the "pivot turn" is easy to execute -- to pivot about the left side of the robot, move just the right stick; to pivot about the right side of the robot, move just the left stick. We haven't found a good way to get this motion using an FPS, as such a motion requires a coordinated synchronization of the forward/back stick and the rotation stick.

Any teams have some good suggestions on how to do "pivot turns" with an FPS control approach?

Essentially, this is the complementary problem to that of a two-stick tank drive having trouble making arcing turns (or, to a lesser extent, driving straight at low power settings).

Back when we used two-stick tank controls for a skid-steer robot, we solved the "drive straight" problem by having a "trigger" on one of the sticks be a "drive straight override" -- in that case, only the stick with the trigger held is used to determine robot speed, and both sides of the robot were given the same motor commands to "drive straight."

In any case, why do I ask about doing "pivot turns" with an FPS control approach? Well, we used "Speed Racer Drive" (FPS) on our robot last year, but have found that the "pivot turn" maneuver is often helpful for lining up on totes. Getting the robot to gently do a "pivot turn" is very hard for FPS, but would be pretty easy for a two-stick tank. (So, yes, we're considering switching back for this year, due to the usefulness of the "pivot turn" when aligning against a tote.)

[MrForbes]

can you just add a couple buttons for that? The gamepad has lots of buttons. you don't need speed control, just make one pair of buttons turn the right motors at a slow speed forward or back, and another pair turn the left motors at a slow speed forward or back.

But I'm not a programmer, so I don't know how difficult that is to actually implement.