I’ve posted about something similar to this before, but now I want to ask this question specifically. Looking through some online resources I’ve seen a lot of high-level teams that clearly CAD their entire robot before they build it. However, on our team, our coaches are really encouraging us to push individual subsystems and begin fabrication (in-house) before the rest of the robot is cadded and then go back and fix our mistakes in later iterations.
It depends how intricate your robot is tbh. For us, we can start manufacturing and assembling swerve modules and other generic things that the current iteration of the CAD will require. Typically, we don’t start assembling or attaching anything to the unmade robot frame until we have the dimensions finalized.
Last year We starting cading the base on the first Sunday and started manufacturing on the first Friday. We had literally no clue what the rest of our bot would be like at that point. However, we did use versa frame everywhere and we had an overall design that made it easy to integrate stuff
Historically our team has done cad after we design and make things which is very counterintuitive and unproductive but for this upcoming season, we’ve made changes to our process to ensure we cad as much as we can before we start making parts to go on the robot.
We CAD our drivetrain within the first few days of build season, and aim to have it built in around a week and a half. Once the drivetrain is CADed, the mechanism subteams do their CAD on top of the chassis and then we manufacture the final iteration of those when they are refined.
I’ll offer what 100 often does: We traditionally fabricate after CAD, but not on a full-robot basis; rather, individual assemblies (such as drivetrain or manipulator) “ship” to manufacturing. Typically this means that in the 3-7 days of downtime while our fab kids handle making and assembling, the designer(s) move on to a new subsystem. Once an assembly is put together, it’s tested for a few minutes and then modifications (usually requiring single-digit numbers of new parts to be fabricated) are made in CAD, and then sent off in another wave to fabrication.
This technique works because it allows fabrication to get a head start on making whatever it is that’s already working in CAD - for instance, if you’re clever about including mating points on an assembly, you can pretty easily design to mate with it even though you’re locked into specific features. If those integration points change, typically we might fabricate 1 or 2 integration parts and do some match drilling. This is particularly well suited to our manufacturing capabilities, as we rely very heavily on CNC machined parts. This way, we can keep the mill running for as much time as possible during the season.
A primary downside of this is that it’s difficult to reshape the overall robot paradigm later on, and unforseen issues may require serious rework of an assembly. It’s simply an unfortunate consequence of the time constraints of the build season.
Nothing on the robot should ever be considered final. Just because something is built doesn’t mean it can’t be thrown out and replaced. In some, maybe most, industry environments there is a clear delineation between the design, analysis, and fabrication phases of a project. In order for that to work you need time. In FRC, time is the resource which is in shortest supply. Prototyping, design, fabrication, and testing tend to happen nearly simultaneously. A team culture of iteration and continuous improvement will serve you well. Transitioning to this mindset is sometimes a shock to our new team members, who think that when something is assembled for the first time, it’s “done”. No, you’ve just finished the initial assembly of something that will be tested, broken, modified, redesigned, and sometimes discarded for something better.
All that said, you can be smart in prioritizing what gets built first, and what can wait. As others on the thread have pointed out, many teams start by building a drivetrain. Veteran teams tend to have a proven drivetrain design they return to year after year, with small modifications to fit the current game. Scoring mechanisms can be prototyped and developed on previous years drivetrains before being “designed” to integrate with the current year’s robot. Often, tradeoffs must be made to make everything integrate, and that will require some amount of rework. Adopt the mindset that iteration and rework are a desirable method for making your robot better, not a waste of time. If this program should teach you anything, it’s that all the planning in the world doesn’t compare to iterative testing as a means of rapidly optimizing a competitive robot.
The real world rarely precisely matches the assumptions you need to make to design a robot. You need to make a lot of guesses which will turn out to be wrong. Embrace prototyping and iteration as a means to evolve your robot to better operate in the actual world, with the game as it changes over time.
If only we could get all the stuff done that we need to, in a nice progression from strategy decision, robot concept, subsystem prototyping, subsystem design, overall integration, detail design, manufacturing, assembly, testing, and be programming on “something” all the while…
instead, we end up taking two weeks to develop one or two working subsystem prototypes, and by that time we need to have built stuff, so we have something to program and test before build season ends. It’s a symphony of chaos, usually.
Typically for us… (may vary a bit from year to year)
After our brainstorming sessions on game play and robot design, and after some prototyping…
Once we nail down the basic frame size and drive train type, we start cadding it up to be built upon.
Then as systems get figured out and refined we add to the model and fabricate parts.
So the cadding is slightly ahead of the building, but if we need to nail down major design pieces to be fabricated, we get to about 90% before fabricating.
In a good year our Cad is 75-85% finished at the end of the season. Some things are just worth the trouble with the resources we have available. Parts that have to be manufactured to mate with other parts get more attention. Obvously anything that gets CNC has to be complete. If you keep track of the weight and CG of the sub parts, you can get a good estimate of the robot weight and CG.
So you have to do what is best for your team and resources. If you have no experience in CAD, Build season is not the place to learn. What we did was work into it over a number of years. Start with a part & build from there.
So, I think a lot of people have offered some good tidbits, so I’ll offer some slightly different perspectives.
Keep in mind that you’re looking at high-end teams - high-end teams have an amazing amount of skills and resources which allow them to get further along in the CAD process before beginning “manufacturing.”
Separate perception and reality - often times the documentation may have a clear delineation between CAD and manufacturing, but you don’t have visibility into the actual wrinkles of the season. So, you don’t see all the iterations they’ve done in prototyping, and you don’t necessarily see all the iterations they manufacturing after the CAD is “done.”
Design for change - as ToddF has mentioned, you should be open to things changing. After all this is the foundation of iterating a design. However, there are certain design decisions you can make to ensure that you don’t end up needing to re-manufacturing everything. For example, if you have an arm and a claw, ensure that they attach easily so that you can just upgrade the claw instead of having to swap out the whole upper structure.
Ultimately, your coaches are on the right track.
It’s better to make a decision early, learn from it, and make improvements rather than anguish over finalizing the perfect design.