Swerve designs

[Dominick Ferone]

One of our off season projects was the possibility of building a swerve drive and add manipulator to test new things we have never done before. We have experience in meccanum but noticed issues strafing at times and obviously pushing fights can be a pain. We'd like to test swerve and were wondering what are some tips needed and other easy ways to get a swerve drive going that's light weight. Preferably the weight of or not lighter then a meccanum chassis.(we may also experiment with a custom chassis or modify a versachassis as we have only done meccanum and c channel the last two years.)

While we enjoy it and our drivers got good at meccanum we don't want to stick ourselves in a rut and feel like we have to use it.
I know some downsides with swerve can be weight, lots of motors/moving parts and battery draw.(largely in part of most use high traction wheels meaning more friction.)
Any advice would be greatly appreciated and anyone could correct me where wrong or educate us where we are lacking.

[asid61]

It's hard to get a swerve as light as a mecanum. Getting one to that points could take iteration and a CNC to make optimal lightening patterns in aluminum.
For ideas, look at 2451's cim-in-wheel swerve drive, or their two speed swerve drive design posted last year. Aren Hill also developed a tiny swerve module and a single-speed design similar to 2451's, as well as countless other designs.
If you don't have a CNC, I would recommend buying the Revolution swerve modules from team221.com and fitting them to your application. Those can weigh very little if you avoid a lot of external gearboxes and such, and only require a little work to integrate into a chassis.
Alternatively, for a fast and easy solution, the AndyMark Swerve and Steer is a fantastic choice as it offers all the motors, gearboxes, encoders, and the module itself for less than $350. And it comes already assembled. For a first time swerve, those are the easiest, however, it is not the lightest solution out there.
My team might pursue a custom swerve design in the offseason, but it remains to be seen. We tried it last year but bit off more than we could chew in the machining sector and could not complete it by build season.

[nathannfm]

1640 has a great website dedicated to their work with swerve over the years

http://wiki.team1640.com/index.php?title=Swerve_Central

When we designed our swerve last summer their team was an invaluable resource. Chief Delphi is also a great resource, there are dozens if not hundreds of threads about swerve design, pictures, and models of what others have done or thought of. Just search for "swerve" or "crab"

How much did your mecanum drive weigh? 1640 has modules that weigh around 8lb, MOE was able to trim it down to 7.2lb (lightest in MAR that I am aware of), AndyMark sells an assembled module that comes in at 8.2lb so keep these numbers in mind and decide if a 30-40lb drive is worth what swerve can do for you.

Also keep in mind I would say at least 70% of the work is programming after you have a working set of modules, be prepared to leave time for your programmers to play with it after it's up and running. We are still making programing improvements and will probably spend this summer working on even more.

[Gdeaver]

1640 will be at champs. We would love for teams attending to stop by and talk swerve.

[Dominick Ferone]

Thanks for all the resources, unfortunately we won't be attending champs this year so we can't go see 1640's swerve modules. I have seen them in past years in either 2013 or 2014.

Also our drive train this year weighted around 35 pounds this year if I recall. We had c-channel frames with the corner connects,tough box nanos,mecannum wheels, our custom bearing blocks and extra supports this year to run the belts.

We could make it light by adding holes in the c-channel, make custom lightweight corner connects, and other odds and ends.

[MStump]

Another thing to keep in mind is with swerve comes a lot of extra parts: eight motors, eight motor controllers, four sets of gears, and another stage for rotating the gearboxes.

You will also almost certainly need at least four encoders (maybe even eight depending on how you do it!). Encoders can often be difficult to work with so if you decide to use a swerve drive during the season, it would probably be a good idea to be confident with the use of these sensors.

[nathannfm]

Originally Posted by MStump

Another thing to keep in mind is with swerve comes a lot of extra parts: eight motors, eight motor controllers, four sets of gears, and another stage for rotating the gearboxes.

You will also almost certainly need at least four encoders (maybe even eight depending on how you do it!). Encoders can often be difficult to work with so if you decide to use a swerve drive during the season, it would probably be a good idea to be confident with the use of these sensors.

The only thing not included in the weights I was talking about are the extra 4 motor controllers. Everything else is on the modules. I also did not consider any of the framing because with a swerve it is hard to define where the module support rails end and the structure of the rest of the robot begins.

Encoders were our biggest issue this year, make sure you can mount them in a way that allows for "wobble" so you don't damage them.

[asid61]

Originally Posted by MStump

Another thing to keep in mind is with swerve comes a lot of extra parts: eight motors, eight motor controllers, four sets of gears, and another stage for rotating the gearboxes.

You will also almost certainly need at least four encoders (maybe even eight depending on how you do it!). Encoders can often be difficult to work with so if you decide to use a swerve drive during the season, it would probably be a good idea to be confident with the use of these sensors.

How would you use only 4 encoders? You could probably live with 5 on a 4-wheel crab, or 4 encoders on a 3-wheel crab, but I thought you needed at least 6 for a 3-wheel swerve?

[Ether]

Originally Posted by asid61

How would you use only 4 encoders?

4 encoders for steering 4 wheels.

Voltage control of drive motors.

[Gdeaver]

For a 4 wheel drive independent swerve we use 4 0 - 360 continuous rotation absolute encoders with 10 bit analog output for stearing. We do not control individual drive wheel speed with encoders. We do have a tach on each drive wheel. We use a cheap analog IR analog reflectance sensor plugged in to a digital input. The roborio does the digital chopping. Pololu QTR-1a 's for $4.25 a pair. We use the tach's for distance measurements. This year we went with a navx-mxp board for autonomous navigation. Swerves do not like to drive straight.