CAD is not essential. As others have said, it is a tool that can provide a lot of benefit in the right hands. In the wrong hands, it can give false confidence in a design that is not practical.
Currently, the design process we try to follow boils down to:
- Defining requirements - These should be S.M.A.R.T.
- Basic 2-D mock-ups - This can be a sketch in CAD, a linkage made out of cardboard, whatever we can do quickly to get the basic geometry worked out.
- Prototyping - This can help narrow down designs if multiple 2-D mock-ups showed promise. It also gives us a chance to test things like ball compression or grip material.
- Basic 3-D CAD model - This is detailed enough to build a full working version of the mechanism, but usually missing things like lightening patterns. If we're unsure about mounting points, the model may contain an array of holes that we can choose from when assembling everything.
- Initial integration check - Before cutting metal, we like to make sure everything will fit together, at least on a high level. Sometimes this only happens as blocks in PowerPoint, other times it's a full 3-D model.
- Revision 1 mechanisms - These generally become our practice bot. There are always things to be learned by actually assembling and testing everything. We try to get these done with enough time before Bag Day that we can iterate on them at least once.
- Iteration cycle - The learning from revision 1 gets integrated into the CAD models, and when we feel that we have something we can put on the field at a competition we begin manufacturing revision 2. Inevitably, over the course of the competition season further improvements become apparent. We try to keep our CAD models as current as possible as the season progresses, in case additional parts need to be manufactured.
The farther along we get in this process, the more value we get from having an accurate 3-D model of the robot. This year, after our first event we decided our pickup wasn't robust enough, so we made a new one for our second competition. That iteration lasted through two events, but it's showing some wear. We're currently planning another iteration for Championship. Having a CAD model has allowed us to plan out these improvements without direct access to the robot.