Why is swerve so slow?

[asid61]

Just wondering why swerve drives all seem to be geared for very low top speeds on the order of 10-15fps. Even 1717 with their shifting swerve has a top speed of ~16fps. Why is this? Is it due to the 4 cim drives?

[z_beeblebrox]

16fps is not a low speed. The (few) fastest FRC robots are geared for a little more than 20fps and the kitbot is geared for about 10fps. Normal low gears for FRC robots are about 4-7fps.

Most drivetrains, including swerve drives, have a < ~20fps top speed because a drivetrain geared too high will accelerate slowly, turn badly and possibly trip breakers.

[asid61]

Not sure about turning, but yeah, acceleration would be a problem. Thank you!

[Tyler2517]

Speed is a interesting concept and a top speed is a really weird way to state it.
We ran swerve this year using something almost identical to what 1640 uses. We ran something at like 12.5-13.5 fps single speed. Acceleration matters much more when you are supper maneuverable in a tight space then when you just want to out run your competition. We wanted something just above the median speed range but not so fast that we would never reach the speed. So a acceleration that we can reach top speed in about 3ish robot lengths. The key is not to be to fast to out run your opponent nor to slow only be where you need to be in such a way that it would appear effortless.
https://www.youtube.com/watch?v=Lpb2...DIBcQ&index=37
Us team 2517 were always where we needed to be to play defense in just the right time in such a way that it would be hard to see where we were going.

Now there is some cool things you can do with swerve that make it appear that you are travailing much slower then you really are. By making small circles where a normal tank drive would drive back then ram you can be essential ramming at full speed even though it takes up the same amount of space as a tank drive would take to break and start moving again.

[asid61]

Originally Posted by Tyler2517

Speed is a interesting concept and a top speed is a really weird way to state it.
We ran swerve this year using something almost identical to what 1640 uses. We ran something at like 12.5-13.5 fps single speed. Acceleration matters much more when you are supper maneuverable in a tight space then when you just want to out run your competition. We wanted something just above the median speed range but not so fast that we would never reach the speed. So a acceleration that we can reach top speed in about 3ish robot lengths. The key is not to be to fast to out run your opponent nor to slow only be where you need to be in such a way that it would appear effortless.
https://www.youtube.com/watch?v=Lpb2...DIBcQ&index=37
Us team 2517 were always where we needed to be to play defense in just the right time in such a way that it would be hard to see where we were going.

Now there is some cool things you can do with swerve that make it appear that you are travailing much slower then you really are. By making small circles where a normal tank drive would drive back then ram you can be essential ramming at full speed even though it takes up the same amount of space as a tank drive would take to break and start moving again.

Exellent driving in that match! You definitely look faster than you are.

Thinking about acceleration, if you could take the battery load then you could "warm up" the motors by spinning in place until it's time to move.

One thing to note is that acceleration apparently barely changes from 10fps to 20fps. One parent on our team made me a spreadsheet (with graphs) that detailed acceleration given motor specs and robot weight, although it did not include friction in the calculations. It showed that for speeds up to ~30fps the distance vs. time was almost the same. By going at lower speeds you would get an advantage on the order of a few inches.

Increasing the number of cims did help acceleration a lot, but nothing else except robot weight helped that much (according to the spreadsheet).

[brennonbrimhall]

Originally Posted by asid61

Exellent driving in that match! You definitely look faster than you are.

Thinking about acceleration, if you could take the battery load then you could "warm up" the motors by spinning in place until it's time to move.

Most teams don't "spin in place" because that is also known as "stalling your motors."

One thing to note is that acceleration apparently barely changes from 10fps to 20fps. One parent on our team made me a spreadsheet (with graphs) that detailed acceleration given motor specs and robot weight, although it did not include friction in the calculations. It showed that for speeds up to ~30fps the distance vs. time was almost the same. By going at lower speeds you would get an advantage on the order of a few inches.

Increasing the number of cims did help acceleration a lot, but nothing else except robot weight helped that much (according to the spreadsheet).

A look at Ether's Drivetrain Acceleration Model might be worthwhile.

[Orion.DeYoe]

Originally Posted by asid61

One thing to note is that acceleration apparently barely changes from 10fps to 20fps. One parent on our team made me a spreadsheet (with graphs) that detailed acceleration given motor specs and robot weight, although it did not include friction in the calculations. It showed that for speeds up to ~30fps the distance vs. time was almost the same. By going at lower speeds you would get an advantage on the order of a few inches.

Increasing the number of cims did help acceleration a lot, but nothing else except robot weight helped that much (according to the spreadsheet).

There's something wrong with that spreadsheet...

[Chris is me]

Originally Posted by asid61

One thing to note is that acceleration apparently barely changes from 10fps to 20fps. One parent on our team made me a spreadsheet (with graphs) that detailed acceleration given motor specs and robot weight, although it did not include friction in the calculations. It showed that for speeds up to ~30fps the distance vs. time was almost the same. By going at lower speeds you would get an advantage on the order of a few inches.

What model for acceleration are you using? What factors is it taking into account? Are these results backed up by empirical data? There are models available for acceleration in FRC, but they are just that - models. There isn't a difference "on the order of a few inches" in acceleration between 10 and 20 FPS, unless you perhaps mean the time it takes to make a 40ft sprint (and if it's a difference of a few inches, the 20 FPS drive is extremely inefficient and performs worse at close range).

I've made the mistake before of using spreadsheets to design drivetrains without really understanding what the values on the screen meant. I'm not saying that's what you're doing here, but that model seems to conflict with my empirical data so I'm curious how it works.

[asid61]

Here, I'll put the spreadsheet into google drive. Like I said, it doesn't take into account friction so the results might be a little off. Please wait a moment.
See here: https://drive.google.com/file/d/0BzU...it?usp=sharing
You can download and open it in excel.

Note that everything is in meters/second, not feet/second. 10m = 32ft.

[Samwaldo]

The only team to do swerve in NE was 2067. They were one of the fastest on the field (didnt see too much pushing robots out of the way from them due to speed. Instead they would go around.)

[Ether]

Originally Posted by asid61

By going at lower speeds you would get an advantage on the order of a few inches

Are you reading the graphs correctly?

At 2 seconds, the 10fps bot is more than 9 feet ahead of the 20fps bot.

[asid61]

Originally Posted by Ether

Are you reading the graphs correctly?

At 2 seconds, the 10fps bot is more than 9 feet ahead of the 20fps bot.

Ether, it's in meters/second. 3.04m and 6.08m is what should be compared for 10fps vs. 20fps.

[Ether]

Originally Posted by asid61

3.04m and 6.08m is what should be compared for 10fps vs. 20fps.

I think what you meant was 3.04m/s and 6.08m/s.

If so, the difference in distance traveled at 2 seconds is almost 13 feet, according to the spreadsheet.

The maximum distance traveled at 2 seconds occurs with 30fps gearing, according to the spreadsheet. The lack of friction and electrical resistance in the model probably is responsible for this unrealistically high number.

[Richard Wallace]

If this discussion is about calculations, then please carry on -- theory is always amusing, and frequently provides insight.

However, if anyone seriously believes ALL swerve drives are slow, then I think there are several highly successful robot drivers who can offer very convincing, practical refutation. <insert well known team numbers here>

Just as one example, Team 16 has not built a robot that anyone could reasonable consider "slow" for a quite a long time. Circa 2004, I recall a very young John Taylor Novak (all 42 inches of him) seated on the St. Louis Regional inspection station table, explaining to a group of grey haired engineers why swerve drive is the best thing since sliced bread.

[Andrew Schreiber]

Originally Posted by Richard Wallace

If this discussion is about calculations, then please carry on -- theory is always amusing, and frequently provides insight.

However, if anyone seriously believes ALL swerve drives are slow, then I think there are several highly successful robot drivers who can offer very convincing, practical refutation. <insert well known team numbers here>

Just as one example, Team 16 has not built a robot that anyone could reasonable consider "slow" for a quite a long time. Circa 2004, I recall a very young John Taylor Novak (all 42 inches of him) seated on the St. Louis Regional inspection station table, explaining to a group of grey haired engineers why swerve drive is the best thing since sliced bread.

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