Swerve drives

[bbradf44]

With the majority of our team graduating this year(about 80%) a couple of us have been thinking about doing one last project in the off season as a team(if we can get the mentors to bite) to both leave something behind that the younger guys can improve on in the coming years to maybe get on a competition bot by the time the current freshman are seniors and to teach them all that we've learned from our experiences. The idea of a swerve drive was brought up but immediately shot down, but then brought up again with some ideas. Sense then we have done some drawings and Inventor projects(nothing worthy of posting here) and are thinking it may be possible. Because we aren't a very high budget team our design will differ greatly from things you'd see on Einstein now but actually don't look half bad modeled out in inventor. My question here is what are some of the most interesting swerves you've seen? Working/not working, on/off a competition not. Post pictures or videos as well.
Also as a side question. I wasn't able to do much pit scouting the past 2 years at Peachtree as I've been stuck in our pit, but I've always been fascinated by swerves and told our pit scouts that if anyone says they have a swerve to come get me immediately, and I've never been told of any. Have there been any teams here in Georgia to pull one off yet?
Thanks!

[Foster]

There are a number of teams that have swerve drives (Not sure about Georgia). Most of the teams have posted decent pictures. 1640, Sab-bot-age has pictures, CAD, etc of the last 4 years worth of swerve drives. It would make a good place to start since they have done a good documentation trail of what worked, what didn't, where they made improvements, etc.

Swerve isn't for the faint of heart, but is amazingly cool when it works.

Sab-bot-age Team Site Swerve Central (http://wiki.team1640.com/index.php?title=Swerve_Central)

[Jay O'Donnell]

Here's team 2067's swerve drive info. Only swerve drive in New England that I know of and they really used it well this year! http://www.chiefdelphi.com/forums/showthread.php?t=123921&highlight=2067

[ekapalka]

The Bomb Squad's swerve is legendary, but I've never actually seen the wheels / modules themselves - just the completed robot with the bumpers on. Also, Team 2451 Pwnage (http://www.chiefdelphi.com/media/photos/40298) is doing their part in reinventing swerve drive with their highly compact swerve modules.

[Bryce2471]

Here's a link to a thread on team mean machine's swerve drive. The most recent BOM counted them at $221 each, so you may want to consider a similar design.

http://www.chiefdelphi.com/forums/showthread.php?t=121672&highlight=2471+swerve

There are a few other threads on them, if your interested. Hope that helps!:D

[Tyler2517]

It is possible to obtain outside funding for the production of a swerve drive. This year we got a grant just for the production of ours.

2471 has an amazing swerve drive that is both light and cheap to manufacture. Making the top out of one piece of extrusion is a great idea. The gear box they use for steering is much cheaper then any thing else i have seen and worked amazingly.

Team 1640 has a lot of info about the history of their drive train. Also a great team who does well with it.

Team 16 has a legendary swerve drive. They don't use a Co-axel design but acts as if they were by using slip rings.

We ran a swerve drive this year as well. We used a cool concept that a student from 2471 thought of(the long tube with the drive belt running around the steering motor). Along with a modified version of 1640's bottom pivot/yoke. We redesigned how to build it to make as many parts of it manual machined on a mill/layth.

https://fbcdn-sphotos-g-a.akamaihd.net/hphotos-ak-ash3/t1.0-9/559947_546786965435604_1829909614_n.jpg

[Gdeaver]

Ok, for the teams that did swerve for the first time this year, was it worth it. Do you feel that it gave you a competitive advantage?

[bbradf44]

Thanks for all the replies. Has anyone ever posted information on a failed attempt at a swerve? Also, this may be a ridiculous question but here it goes, has anyone made one from wood? Wood meaning the module without gears, belts, motors, axles, wheels, etc.. We have the ability to precisely cut wood(I don't remember what the machines called, we don't use it much) but not any kind of metal. Like I said, it is a very out there design

[Tyler2517]

Team 2102
https://www.youtube.com/watch?v=E3k72RVK8Hk
They use a lot of non traditional materials.
Wood is a great material in some applications. I am not sure if using wood for a swerve drive is the best idea at least not for some key parts that take a lot of load.

I know we built a failed co-axle crab drive in 2010. I can see if i can get some documentation up on that. Its a dark spot in my team.

[Electronica1]

This (http://www.chiefdelphi.com/media/papers/2578) is a pretty good paper on swerve that could be useful.

[bbradf44]

This (http://www.chiefdelphi.com/media/papers/2578) is a pretty good paper on swerve that could be useful.

Their early versions is actually where I'm getting some of my ideas, mainly the sprocket made as part of the module for the turning and partly the design overall. Currently our design is a cylinder with a cim close to the top with gears going through the inside and a chain going from the last gear axle to the wheel. It will sit on a lazy Susan bearing(if I can snag a couple before first choice closes) and a large sprocket an inch or so above it built into the cylinder. I'll post the inventor file once I get it cleaned up enough to be judged by the far more knowledgeable eyes such as those found on CD

[kylelanman]

Ok, for the teams that did swerve for the first time this year, was it worth it. Do you feel that it gave you a competitive advantage?

YES!!!

We decided day 2 of the build season we were going to go with a swerve drive. The general decision making paradigm was that due to the lack of a crazy pyramid to climb or any end game for that matter the rest of the robot was not going to be that complex. Therefore we thought we would have enough time to figure out a swerve drive. Having never done one before we reached out to a new company in the area for funding so we could build 2 drive trains. One was built the first week and given to the programmers and the second became the actual robot. We bought 118 style swerve modules from AndyMark and wrote the code from scratch. We did reference 16's code base to ensure we were on the right track from time to time.

We firmly believe and demonstrated in Wisconsin that our swerve drive gave us the additional competitive edge needed to win. The extra agility that the swerve drive provided allowed us to play very effective defense and quickly out maneuver our opponents to score. We took a gamble and it paid off with our first blue banner ever.

[sdcantrell56]

With the majority of our team graduating this year(about 80%) a couple of us have been thinking about doing one last project in the off season as a team(if we can get the mentors to bite) to both leave something behind that the younger guys can improve on in the coming years to maybe get on a competition bot by the time the current freshman are seniors and to teach them all that we've learned from our experiences. The idea of a swerve drive was brought up but immediately shot down, but then brought up again with some ideas. Sense then we have done some drawings and Inventor projects(nothing worthy of posting here) and are thinking it may be possible. Because we aren't a very high budget team our design will differ greatly from things you'd see on Einstein now but actually don't look half bad modeled out in inventor. My question here is what are some of the most interesting swerves you've seen? Working/not working, on/off a competition not. Post pictures or videos as well.
Also as a side question. I wasn't able to do much pit scouting the past 2 years at Peachtree as I've been stuck in our pit, but I've always been fascinated by swerves and told our pit scouts that if anyone says they have a swerve to come get me immediately, and I've never been told of any. Have there been any teams here in Georgia to pull one off yet?
Thanks!

We will most likely be building one this summer and fall and we are in atlanta. Hopefully we will have it together for grits.

[Bryce2471]

Ok, for the teams that did swerve for the first time this year, was it worth it. Do you feel that it gave you a competitive advantage?

Yes,

We did some off-season development of swerve drives, and we were very glad that we did. We already had the code nailed down, and enough swerves for the practice bot, so we just adjusted the design slightly, and pumped out six more for the comp bot. They have had great performance overall, and have had no mechanical failures!(knocks on wood):D
One thing we have been displeased with is the optical encoders. The ones we use have an index pulse, but can still get off by a significant amount for a noticeable portion of the match. Unless we stop to run a homing routine.
We are testing magnipot absolute encoders from Digi-Key on the practice bot and we will probably be changing out the comp bot's encoders for them at champs.

[Gdeaver]

I like the 6127's. We did have problems losing calibration. We used plastic beam couplers from Automation Direct. They would some times slip requiring calibration. This year we took advantage of the good AL beam couplers that were on First Choice and they do not slip.

[bbradf44]

http://i.imgur.com/QqofDII.jpghttp://i.imgur.com/hWQGu2s.jpg
Heres a couple images of our inventor creation. Its still missing the gears, chain, axles, and belts but this is the basic design. This is only a rough and dirty concept. Below the piece of wood is the lazy suzan bearing. about 1/2 inch above the wood is a sprocket for the turning

[Bryce2471]

I like the 6127's. We did have problems losing calibration. We used plastic beam couplers from Automation Direct. They would some times slip requiring calibration. This year we took advantage of the good AL beam couplers that were on First Choice and they do not slip.

Ours were really easy to set up compared to the USD encoders we were using. We thankfully decided to get the fattened shaft version. This means that the setscrew we use to pin them down will not slip, and we don't have to find the junction between 0v and 5v on each one individually. We know that when the flat of the shaft is facing towards the BI symbol of the encoder, it is at "zero". We then mount them to the drive so that when the encoder is at zero the set screw is accessible. We then turn the module to a 45 degree angle,(witch is where we used to put the index) and slide on the plastic encoder gear.

We haven't done extensive testing, but the process seams promising.

PS. How do you interface with the output pins on the 6127's? We just solder PWM cables to them.

[ekapalka]

I'm not sure if this is true or not, but I've been told that a good thing to use for swerve drives are ten-turn potentiometers. The benefit being that you can easily plug it into the following angle=(potentiometer_voltage-inherent_zero)*voltage_to_degrees to get to the angle, whereas with other potentiometers you need to account for some sort of wraparound

[Orion.DeYoe]

Swerve is something I've wanted to build in the offseason for a while. I understand the mechanical and control concept of it and I know how it's driven and mostly how the modules need to be constructed.
The point that I have a really hard time understanding is how the pivot is set up. Does it just ride metal on metal? Or is there some sort of bearings, rollers, or slip pads/blocks? How do you contain the rotating module (the yoke) up and down and side-to-side while still being able to disassemble it? No one seems to have any good cross section views of the modules in an assembled state.

[Tyler2517]

Swerve is something I've wanted to build in the offseason for a while. I understand the mechanical and control concept of it and I know how it's driven and mostly how the modules need to be constructed.
The point that I have a really hard time understanding is how the pivot is set up. Does it just ride metal on metal? Or is there some sort of bearings, rollers, or slip pads/blocks? How do you contain the rotating module (the yoke) up and down and side-to-side while still being able to disassemble it? No one seems to have any good cross section views of the modules in an assembled state.

We use thrust bearings in the top of the yoke so that it can carry weight but still rotate freely.
What do you mean by contain the rotating module? We have a retaining ring at the top of the pivot shaft that keeps it from falling out the bottom. The sides of our yoke have 3 bolts that have threads on the top of the yoke to connect. On ours at least it takes about 2 hours to get every thing to go together, to take it apart it can take up to 6 after a little use(with the help of a hammer). Our pivot shaft is thermal press fit in to the top of the yoke. We can not get it out even if we wanted to.

[The Doctor]

I'm not sure if this is true or not, but I've been told that a good thing to use for swerve drives are ten-turn potentiometers. The benefit being that you can easily plug it into the following angle=(potentiometer_voltage-inherent_zero)*voltage_to_degrees to get to the angle, whereas with other potentiometers you need to account for some sort of wraparound

Team PWNage was using a $60 Absolute value encoder (http://www.usdigital.com/products/encoders/absolute/rotary/shaft/ma3) to figure out the angle.

[Jefferson]

I'm not sure if this is true or not, but I've been told that a good thing to use for swerve drives are ten-turn potentiometers. The benefit being that you can easily plug it into the following angle=(potentiometer_voltage-inherent_zero)*voltage_to_degrees to get to the angle, whereas with other potentiometers you need to account for some sort of wraparound

If you are using the WPILib PIDController, it's as simple as setting the PIDController to continuous when using a continuous rotation encoder/potentiometer.

You have to keep up with turns and unwind if you get close to the limit on a 10 turn pot.

[ekapalka]

Team PWNage was using a $60 Absolute value encoder (http://www.usdigital.com/products/encoders/absolute/rotary/shaft/ma3) to figure out the angle.Believe it or not, our team (your team) is quite familiar with those. We have them on more than one of our older robots.

[mman1506]

Believe it or not, our team (your team) is quite familiar with those. We have them on more than one of our older robots.

Any tips for preventing the tiny wire on the connectors breaking?

[ekapalka]

Any tips for preventing the tiny wire on the connectors breaking?We connected ours by simply removing one wire (and 1/4 of the connector itself) from one of the 4 wire connectors that we had acquired for our OEM encoders (because we bought the 3 pin analogue versions). We've never had any problems with this setup, but we probably would have been better off with one designed for this task. We also cut the top off of a Mountain Dew bottle and used it to shield the sensor because it was located a bit precariously