I mentioned that. I was just mentioning the bevels because the final ratio may not have enough space to be flipped. If so the desired change could be accomplished in two parts across the bevels and final ratio if that space constraint was an issue.
Here’s another quick sketch of a differential swerve, inspired by your gear layout. None of the structure or encoders are shown for simplicity and because I haven’t done it. The motors are mounted on 20:1 planetaries then further reduced by 18:84 to get a steering ratio of 93:1. After the differential, the ratio is sped back up to get a final drive ratio of 20:1 (somewhere around 15 ft/s on a 4" wheel).
It isn’t the lightest assembly, but everything is nice and COTS.
[edit] - You could also just bring the bevel gears right down to the wheel axle and avoid the belt stage entirely - closer to the original design, it would must make for really wide (maybe 6") forks.
You screwed up 
… OP’s version had the potential to build in a lightweight ballshifter on the shaft driving the driven bevel gear.
Oooh, missed that!
[edit] Well, maybe not so feasible. that shaft rotates with the module, so your shifting cylinder (and the air lines) would have to rotate with it.
No need, swash plate. (In the most basic sense)
(i.e. a cam follower on the top of the spring loaded ballshifter and a lowerable plate to engage it, so not really a swash plate (although you could use one if you really wanted…)… just bear with me)
2 inches??? you could go up to 16 inches this year!
(it would be hilarious to rotate all offset modules in sync and manage to get a hula-hoop to work on the lift assembly…)
OK, so I’m super confused. What edge does a caster drive have over a normal swerve?
P.S. the caster drive is hilarious, I lold
It’s a real drive configuration… not something to laugh at.
I refer you to the thonk at the beginning of this thread
really fantastic design! if I could ask, how much does this module weigh in cad?
I’m not actually sure, I’ll check the volume when I get home from school.