Wednesday, February 11, 2009

wednesday [FIVE DAYS LEFT.]

i fear these posts will become progressively more incoherent as the number of days grows less.

-we tried that piecewise thing, first using a whole mess of case structures and such. it semi-sort-of-maybe worked, but in order to add PID loops, we had to add 4 or 5 of them. and something went badly wrong there - the robot just would not stop on time. it seemed like we'd added a lead time rather than a lag time (or maybe i've got it backwards). either way it DEFINITELY didn't work - after you pulled the joystick back to center (0,0) it would take many seconds for the motors to actually stop.
-so we attempted to use a formula node rather than uglygrossmess of case statements, as karlin suggested. the formula node essentially allows you to type code in c (!) into labview. as most of you can guess, this feature makes me really really happy. except then, the robot COMPLETELY did not work. at all. somehow, the motors received their values at different times. so if you moved the joystick really really quickly, then one wheel would start spinning, and almost as soon as it stopped, the other motor would start. they just weren't aligned, time-wise. although the control on the front panel thought they were getting the motor speeds at the same time. we tried to add a bias, which kind-of worked but not in a good enough way. according to karlin, there are Issues with the formula node.
-so tomorrow we attempt to write subvis which we can just place into robotmain. essentially, the subvi will take joysticky and joysticks and the output will get fed into the motor speed. that subvi will likely be the mess of case statements. although karlin says she thinks that there are smart ways to do it.

jimmy should be done welding tonight.

5 days. keep your chin up, we're almost there.

No comments: