@JacobD, I was giving this some more thought over the weekend. There are a handful of ‘design style’ guidelines that we follow which help us turn out designs more quickly than we used to. We were quite cognizant of these design process changes when we implemented them circa 2014, but they have become second-nature in recent years.
- 10-32 bolts and 3/16 rivets are our standard. Attachment point? 0.201in hole(s). No thinking needed.
- We do not make our own gearboxes basically ever. We just find a way to make OTS solutions work well enough.
- We use as many OTS parts as possible. We buy spacers and standoffs from McMaster, something we can easily make ourselves, but if we don’t have to we won’t.
- We have a sheet metal + brain/belly pan build style that we use from year to year. This dramatically reduces the number of design decisions to make when creating the chassis. We also try to use sheet metal everywhere else we can. Sheet metal parts are light, can be made quickly, and with good bending can dramatically reduce the total parts count.
- We try to find ways to reuse the same parts over and over within a given robot. Our chassis side plates are always the same, our 2018 elevator bearing blocks were left/right/top/bottom independent, our wheel modules are generally position-independent, things like that. It is super easy to make a chassis when you just build one side plate, pattern it to make a drive pod, pattern-array wheels modules into it, then pattern the whole drive pod to make most of the chassis.
For the sake of clarity, here are some screen shots:
Drive plate. There are a lot more holes than what will actually be used in any one position, but having all of the possible holes on every part lets them all be the same exact part. (That feels really weird to write out…)
Wheel module. Note that it’s bearings are captive to the axle/module and the whole thing can be installed, rather than being assembled on the chassis. The only ‘custom’ part here is the drive alxe, which is just the thunderhex model from Vex with an Extrude Cut to make it the right size.
Drive Pod Assembly. This is two chassis plates, three wheel modules, one transmission model downloaded from Vex, and some special sauce cylinders downloaded from McMaster.
Drive Base/Chassis Assembly. Comprised of two drive pods, one brain pan, and one ‘end cap’ part with 3x configurations to link the two drive pods together.
To get to this point we modeled 4x parts, one of which has 3x configurations, and downloaded less than 10x other models.
Does this style result in the optimal robot design every year? Absolutely not. We always compromise on one or more performance metric like CG, total mass, mechanism volume, etc. But it is FAST. Like, super stupid fast. We’re going to release models to our sheet metal sponsor tonight and would expect to have our chassis parts by Friday. We’ll also start machining some parts in house and maybe have elevator parts done by Saturday.
We also start cutting metal before everything is modeled in, we just try to give ourselves as much wiggle room as possible to fit things in later. We, as a team, accept this risk in favor of having our primary functions operating sooner to test and iterate. This mode of operation is a function of how small we are, we simply do not have the people-power to churn out a fully integrated model by the end of week 1 or 2. So, we work with what we’ve got!
I hope this was a little more informative/helpful than my last response.