So it’s all done and dusted, we’ve made it home on a high, and yet we’re already thinking about what we would do differently. Well for one, Glitterator II really didn’t glitter enough. We had a Blinkt for status indication, but that’s only 8 APA102s. Nothing compared to the dazzle that Glitterator managed. So definitely more lights. LOTS more lights. And that despite Dr Lucy Rogers saying today:
You don’t get any points for blinding the technical merit judge. #justsaying
— Dr Lucy Rogers (@DrLucyRogers) April 23, 2018
The controller issue we had was a shock. Last year we had serious problems with our bluetooth PS3 controller dropping out, and so this year we opted for a Rock Candy (wifi dongle) controller. I’m not sure what the solution here is, but at least with the genuine controller you get a unique pairing. It might be worth trying to properly isolate just what was causing our PS3 dropout issues and reverting to that. Or finding something else that will do a unique pairing. Much thought needed there.
The vision challenge was obviously something we struggled with, but mainly because we ran out of time. Actually, given how little time I spent on it, I was quite happy with how it went.
In general, I’d like to do a better job of error handling. The serial communication between our two pis occasionally got out of sync and died fatally. I knew that this was an issue but didn’t have time to fix it. Thankfully it didn’t happen on the day, but that was just luck.
Having said that, I’m thinking again about the two-pi solution. I still like the idea of having the remote-controlled robot getting “taken over by a parasitic brain,” as Angus described it on the day, and being able to connect/disconnect it as we wanted did make life easier. So on the whole, I think I still like it.
Power is another issue that we will revisit. We had three separate things to power this year: the motors, the pi zero, and the pi 3B+. Each one was powered by LiPos – the pis via LiPo shims, and the motors by a hacked Lego battery pack (with a voltage monitor hacked into it). (Sheesh, we forgot to show off that feature in the technical merit judging!) Biggest problem was that the pi zero was running off a 400mAh battery – it went flat way too quickly. It would be easy to get a bigger battery, but then that wouldn’t fit in the hacked battery box. So another solution might be called for. I am eyeing up the PiJuice hats…
And in general, reliability is something we want to do better at. And that means being ready earlier, then testing, testing and more testing. So there’s the key part of the plan for next year.