So I have finished (for the most part) the design for this thing.
Welcome to the FTC Imbedded Wheel Swerve Drive (IWS Drive) Google+ Album here:
https://plus.google.com/u/0/photos/101718392754230534136/albums/5722384081366804769/5722384077534644098
As some of you might have seen
http://www.chiefdelphi.com/media/photos/37673
This module, it inspired me to try to do the same, but at a different scale.
It uses alot of resources, so I dont think a competition worthy version will actually be built. However If it does get built I want to see it and make sure you email me pictures and video! 
What sets this apart from most FTC systems is that it uses absolutely no Tetrix structural metal. Why you might ask? Simply because I think it would be awesome to see a team attempt to build an entire robot using as little Tetrix material as possible. It would teach alot to students and has potential for some really awesome mechanisms 
As for the construction of this thing.
The Frame.
The frame is 1" x 1/2" x 17.8" C Channel Riveted to a 17.8" x 17.8" x 1/16" water-jetted piece of AL. (doesn’t have to be, I just have those resources available)
Bearings, Turntables, and Custom Gears.
They are made from laser-jet sheets of 1/8" and 1/16" thick Delrin, reason for this is because of the offset of the 1/16 inch sheet metal belly pan. All gears are compatible with the Tetrix gears as well this allows one to use them elsewhere if need be.
Motor mounts.
They are modified as well. Since I find the Motor mounts in FTC kind of large and bulky in this application I have made them thinner. However one chooses to do that is ok However I would use a mill personally. The motor mount also has gearbox mounting holes in it which is made from laser cut Lexan or Delrin (It doesn’t really matter)
Motor Sleeve.
The motor sleeve is a piece of schedule 80 PVC It is machined out in the center to a couple thousandths smaller than the OD of the motor. It is then pressed onto the motor ensuring a very tight connection. On the outside before it is pressed on two grooves are machined into the sleeve at predetermined distances that act as “tracks” for the bearings. Which brings me to the next topic.
Imbedded Motor Inside Wheel.
How this was accomplished was by drilling out the center of the wheel with a considerable bore which is larger than the motor. From there there are through bolt holes drilled. Which attaches the bearings. (Yes Bearings If you remember the Pivot bearings) This drive uses 16 per wheel However can be reduced to 8 per wheel if need be. The Drive gear also needs to be heavily modified and drilled with the same bolt pattern at the wheel and threaded to a 4-40 threaded bolt that is 1.5 inches long (I think)
Now that I think I’m done with the mechanical portion, its onto the sensor/electrical/programming portion.
Drive Wheels.
Each drive wheel has enough space allotted so that you can attach the FTC encoder to the drive motor as usual to give you a feedback on how far you have gone. The wires would be suspended above the center of the wheel to prevent extreme tangling and give you 360 degree rotation (with limits of course)
Pod Rotation.
This is done through the continuous rotation servo, since lets be honest. None of us want to dedicate 8 motors to the drive train. Even I say 6 is max, even this year. But nonetheless, If the specs I found on the Cont. servo are accurate the 5.5:1 would have rotated it something like 1 per minute or something ridiculously slow. So what I did (not shown in renders) is made a small custom gearbox for the Servo. This increases the speed from the ridiculously long time it would have taken to about ~1 second rotation times. This would allow for incredibly diverse programming options and quick maneuvering of the field.
Now How do you predict where its at when its rotating you ask? Well this is where I’m stumped, but I have a few solutions. My first idea was to take a sheet of Paper or different colors lexans and crate a ring with ticks on it these ticks would be large enough for a Lego light sensor to sense, but small enough to make the system accurate. That would be done through a sensor Multiplexor to the NXT.
My second idea was using the Protoboard. Now I’m not sure how expansive the Protoboard can be feel free to enlighten me if this is possible. How it would work is you would use the encoders provided in the FTC kit through a custom gear system to basically sense how far the pod has rotated over time, this would be applied one per pod thus giving you 4 encoders for the pods total. Encoder count now currently at 8 for the drive train.
Anyway I think thats everything I can think of for now.
Comments, Questions, Concerns?
Enjoy
