Why is swerve so slow?

[Tyler2517]

Something else to note swerve drives are typically heaver then a skid/west coast ect decreasing acceleration.

How does the mechanical efficiency of a swerve drive compare to that of other drives?
How much does Mechanical efficiency matter any way? Does removing a gear set or a chain really do that much?

[Meredith Novak]

Originally Posted by Andrew Schreiber

To be fair, while he's MUCH taller now he still does that...

And he doesn't actually sit on the inspectors' table now...the under 12 rule went into effect the year he turned 12. Could have been ugly.

[Kevin Leonard]

Originally Posted by Tyler2517

Something else to note swerve drives are typically heaver then a skid/west coast ect decreasing acceleration.

So you're telling me the 120 lb robot with swerve will accelerate slower than the 120 lb robot with 6WD because it weighs more.



Swerve drives are not inherently slower than any other drive system.
/thread

[Tyler2517]

Its a know fact that swerve is heavy.
Comparing a robot that is identical to another with a swerve drive. The one with the swerve drive will almost always be the heavy one.
This year was a big point in that when robots would typically only weigh 90ish pounds.
I'm not saying that swerves will always be slower I am just pointing out they are heavy in comparison.

[Greg Woelki]

Originally Posted by Kevin Leonard

So you're telling me the 120 lb robot with swerve will accelerate slower than the 120 lb robot with 6WD because it weighs more.



Swerve drives are not inherently slower than any other drive system.
/thread

Correct but note that in high gear drive trains are power limited so 6 cims in a tank drive will lead to faster acceleration.

[page2067]

What may give the impression of slow for swerves (4 wheel independent) is what happens when rotation and translation (spin and strafe) actions are combined. Here each motor is going at a different direction and speed - the fastest motor is at the limit - the other are less - net result is a slower translation motion. But, If you drive pure translation (all 4 motors at their limit speed) and no rotation you go faster.
(You can see this effect if you use Cory's 2067 swerve simulation shown on a different thread)

A second consideration is most swerves are not shifters - so you will typically see in the 10 - 14 fps range. We were around 12 fps which was actually faster than we have been before (with tank, and were at ~10 fps in our development swerve off-season robot). Tank shifters will often have high speeds at 14 - 18 fps (true speeds) (there are shifting swerves out there though).

Thirdly most are 4 CIMs - compared with the ease (now) of implementing 6 CIM shifting tank drives - will typically show less acceleration for a given gear ratio.

On the other hand a swerves ability to directly go in the direction desired and to spin around a defender can speed up getting from A to B.

I believe the open field this year favored faster. NE, I believe, has traditionally run slower with more emphasis on defense/pushing. I think this year CA taught us that fast and maneuverable was the way to go - though with 6 CIM shifters to maintain pushing capability.

As some have posted, in 2013 on a "shorter" field, 610 showed a conservative 10-11 fps cycler that was "Faster" than most. Going with single speed 6 CIM drive has advantages for simplicity; mechanical and driving, and low weight.

In the end you may be perceiving a swerve to go slower but if applied (correct gear ratio(s) used, and reliable) and driven well can be more efficient and "faster" in executing its goals.

Anyone seeing 1717 in 2012 would say they were clearly the fastest (IMO) in collecting balls and shooting reliably, though you watch there slalom video and would say they looked slow.

[Tom Bottiglieri]

It seems the inability to carry momentum through a change in translation direction causes a swerve to be a bit slow.

For example, on a skid steer setup, you can be moving very fast and basically pin one side to the ground. This allows the robot to use its own momentum to swing around the pinned wheel side. You exit the turn with almost as much energy as you had before, just moving a different direction.

I think most swerve drive software implementations are pretty naive and will just turn the wheels 90 degrees when the driver wants to start moving to the side. Field centric control only makes this worse as there is another controller between your driver and the wheels.

If we ever built a field centric swerve, I would almost certainly allow a drive mode that allows quick momentum saving direction changes. (Maybe a single button click? Maybe a press and hold button that locks 'forward' to the direction of your last translate command then allows throttle/turn steering?)

[AdamHeard]

Originally Posted by Tom Bottiglieri

It seems the inability to carry momentum through a change in translation direction causes a swerve to be a bit slow.

For example, on a skid steer setup, you can be moving very fast and basically pin one side to the ground. This allows the robot to use its own momentum to swing around the pinned wheel side. You exit the turn with almost as much energy as you had before, just moving a different direction.

I think most swerve drive software implementations are pretty naive and will just turn the wheels 90 degrees when the driver wants to start moving to the side. Field centric control only makes this worse as there is another controller between your driver and the wheels.

If we ever built a field centric swerve, I would almost certainly allow a drive mode that allows quick momentum saving direction changes. (Maybe a single button click? Maybe a press and hold button that locks 'forward' to the direction of your last translate command then allows throttle/turn steering?)

16 does this and it's why everyone thinks they are so fast.

We had variations of it in 2012 but never quite practiced them enough to utilize it.

[Michael Hill]

Just throwing this out there: I've got a drivetrain calculator spreadsheet on here.

http://www.chiefdelphi.com/media/papers/3038

It used Ether's algorithms, but it's in Excel format. There are some pretty graphs as well.

[asid61]

Originally Posted by Tom Bottiglieri

It seems the inability to carry momentum through a change in translation direction causes a swerve to be a bit slow.

For example, on a skid steer setup, you can be moving very fast and basically pin one side to the ground. This allows the robot to use its own momentum to swing around the pinned wheel side. You exit the turn with almost as much energy as you had before, just moving a different direction.

I think most swerve drive software implementations are pretty naive and will just turn the wheels 90 degrees when the driver wants to start moving to the side. Field centric control only makes this worse as there is another controller between your driver and the wheels.

If we ever built a field centric swerve, I would almost certainly allow a drive mode that allows quick momentum saving direction changes. (Maybe a single button click? Maybe a press and hold button that locks 'forward' to the direction of your last translate command then allows throttle/turn steering?)

I was thinking about this too. It is very possible to do this using a swerve drive. However, not only can swerve turn like that, it can also do a "basketball roll" around a defensive robot by spinning while going around the opposing robot to keep its momentum.

I recall 254 made some swerve modules in 2008. Where did that go?

[Electronica1]

148 did swerve in 2008, and I doubt anyone can call that robot slow.

[AdamHeard]

Originally Posted by asid61

I recall 254 made some swerve modules in 2008. Where did that go?

Those modules were for the 2008 world champions

[DampRobot]

Originally Posted by AdamHeard

Those modules were for the 2008 world champions

I always assumed that they were a prototype, or for the EWCP swerve project!

*Mind Blown...*

[AdamHeard]

Originally Posted by DampRobot

I always assumed that they were a prototype, or for the EWCP swerve project!

*Mind Blown...*

They were definitely 148's in 08. 254/968 just posted the picture, they didn't make them.

254 isn't even involved in EWCP either. Did they ever do a swerve, I don't recall that.

[Tyler2517]

There are maneuvers that allow the momentum to transfer they just are not instantaneous.
The faster you want to translate the less momentum you can save. Mind you this is for pure translation only you can save a marginal amount of momentum by translating and rotating at the same time. http://youtu.be/1oVNrp2L1EQ?t=2m27s For instance this is the best example of us translating with the rotation.
Using banked turns on a swerve drive is best as it allows for the most momentum saved over instantaneous turns.