[alopex_rex]

This is interesting, I look forward to reading the next article.

Last year we started out designing the controls from the "programmer's perspective." After a scrimmage, where we discovered that our robot was very difficult to drive, we designed a new semi-automated scheme. I spent several straight hours on my computer at home implementing it on top of our existing code, which was not structured well to accept the automation. It... worked, but it wasn't pretty.

This year we began with semi-automation in mind, and started brainstorming the control scheme as soon as we knew all the actuators to control. We settled on a very simple-to-control system; the copilot basically just has to press a button now and then, whenever we decide to pick up a tote or a recycling container. We have another "test" teleop mode for manually controlling each mechanism, which we use heavily for, well, testing. Just today we finally got the "official" teleop mode to be self-sufficient, i.e. we can use it continuously without having to switch to manual controls to fix issues. Today we were preparing for driver tryouts, and it was gratifying to see how easily everyone picked up the copilot controls. We haven't gone over to one-driver control, for various reasons, although I've certainly considered the notion from time to time.