[Chris is me]

I don't want to "slam" anyone who works this hard, so please don't take this the wrong way, but I think this system has a few problems.

First, your modules. Yes, cantilevered wheels offer easier access, but they are not the end-all be-all most important drivetrain feature in FRC. The reason my team cantilevers wheels is mainly for simplicity and weight. Cantilevering wheels in a swerve drive saves negligible weight (maybe a half pound?) and makes the whole wheel and module significantly worse supported.

In 2008, team 148 built modules much like yours, but with a fully supported wheel. Even then, the modules wore through bevel gears, I'm assuming due to the modules twisting causing excessive tip loading. You'll notice teams like 1625 and 1640, two popular coaxial swerve teams, go to great lengths to make their bevel gear setup rigid. 1625 used a "puzzle piece" method of connecting 1/4" plates while 1640 uses gussets to connect a piece of U-shaped extrusion together.

It also looks like your module support is a bit lacking. It looks like you have a single thrust bearing mounted to a piece of 1/8" metal plate supporting all of the forces on the module. That's a pretty extreme cantilever. A lot of teams use a lower support for at least extreme loading, or they support their modules in multiple places / over a wide area. Think carefully how you want to do this.

Swerve drives are tricky - that's why even well funded teams with many engineers don't get them right.