Competition's today. We're really really happy with our robot. It tracks balls nearly perfectly, and after some more tweaking the roller and lift work great. Unfortunately, we never had time to implement goal finding and dumping, so hopefully we can score a lot from possession alone. To this end, we've added a second tier to the ball hopper so that its capacity is increased. Once in a blue moon the code can crash, but we seriously doubt that will happen.
We got stuck for more than 12 hours on a bug today. We know it had to do with threading, but really don't know why. For some reason, in the thread that tracks balls, calls to Motor.set() would lag for 1 or 2 seconds, causing the robot to drive out of control. Since the ball thread stops all other threads and then restarts them when it's finished, we can't imagine how it could conflict. After trying everything, we just gave up and put all of the code from the image thread into the wandering thread. Now it works, and that's a good thing since we have to impound.
Elvio woke up this morning and realized that the image is being re-Rasterized and getData()'d every pixel. Plus, since we do averaging in a really stupid way, it's more like 4 times per pixel per image. Once we fixed this (stupid) error, the image processing went down to about 100ms(!). That makes it useful enough to actually implement some ball-tracking in addition to wandering.
AUGH AUGH although I am not the primary coder, I am responsible for implementing threading (john) and elvio's image code just isn't cutting it. For some stupid reason the image processing is taking 8 SECS!!! its getting late, most mechanical stuff is done, why! any ideas guys... we know dheera is better than 100ms now....
The ramp we built for the lift is pretty cool. Its a v shaped wedge with aluminum brackets at such an angle that the ball stays in contact with our roller for about twice as long now and kicks right into our lift. Yesterday we actually wired up the lift and bought a threaded rod to replace one of the pins that holds it together (to lift the ball). We turned everything on and it works really nicely now. The image code is coming on but that is Elvio's department and he will update about it when its good and done.
The wood robot has worked out so much better then the lexan, supporting the weight better. We attached the roller yesterday, but we need to mod the lift ramp so it works better tomorrow. Elvio has worked on the code, specifically he has optimized the color code so now it see's colors reallly efficently. He can see cyan vs blue!
Today we rebuilt the entire robot out of wood. The lexan was too wobbly since we didn't have the thicker stock. We mounted the now-working lift, and constructed a ball roller to feed balls into the bot. At the end of the day, we tested it all. Miraculously, it worked quite well. Now we need to do lots of coding. Note: Team 6 STILL officially thinks that the new size regulation is lame. Only now we think it's lamer.
Today we tried the lift. It didn't work. We made a lot of changes, and new things didn't work. (GOTO: HOME)
Eventually, we began work on a new chassis out of polycarbonate and aluminium (for reinforcement). It's still a 3 tier design. At the very end of the day, we made new wheelmounts, and began attaching the ball hopper. All seems well. Note: Team 6 officially thinks that the new size regulation is lame.
Today Johnnie unscrewed some things, added a chain thingie, and worked his tush off in general. Elvio is on vacation =/
Today Johnnie used the metal hole puncher thingie on one of his robot parts. He did other stuff too.
We spent a bunch of time at home getting our color reading perfect :) yay, and we also came up with our basic concept we wish to achieve. Now we have to continue working on the code so we can zero in on a red ball and start thinking about a prototype of our design.
We did a bunch of stuff today. We got OrcSpy working with the help of the team next to us, doh we weren't successful at getting it to run on the mac so we ran it on athena. We also wrote the code and were stuck on that forever, thinking we were inadequate until we realized it wasn't our mistake and the TA's told us the solution (happened to all the groups around us too). Other then those issues however it was a breeze getting the basic 'robut' up and running (and stopping) like we were supposed to. Should be fun tomorrow.
Team Hagopian