I’ve just released the second part of my over-ambitious project to make an auto correcting mini golf club. This time I focused on the software required to enable the club to handle bounces, breaks, etc. You can check it out here!
This was a really fun project, but it’s also a relief to have it finally working. At some point it turned into a software engineering slog with the big range of (brittle) programs I had to build and maintain. The software ranged from low level embedded stuff, to computationally intense algorithms, all the way up to networking and system stuff.
I’ve been considering why this took so much software compared to other similar projects and I think it has to do with the device being handheld and requiring higher precision than other builds. Ultimately it required much better data collection and analysis tools (to see what is going on), more complex control and prediction, and more involved planning algorithms that needed offloading to other computers, etc.
I’m not sure what the takeaway is (perhaps don't fly so close to the sun :), but it’s interesting to consider the seemingly nonlinear relationship between precision / handheld and overall complexity. I’ll certainly be keeping it in mind when thinking about other projects!
Hopefully you all enjoy seeing the project taken to completion. Keep your eyes peeled for an upcoming Patron only commentary where I'll be discussing more of the details and behind the scenes things that didn’t make it into the final video. As always thank you so much for your support!
⁃ Shane