Re: CMUCam Next Year?
 

My limited experience as the team's lead teacher/sponsor was that we could get the camera to recognize color blobs and track but we lacked sufficient time with a working robot to program navigation. IMHO it was the dual challenge of autonomy coupled with looking for a variety of colors on different planes as closed loop system for navigation that was the challenge and not the camera.

Six weeks is all that Botball teams are going to have with their vision camera set up once again this season. Those robots are completely autonomous so once again I think the specific application of the cmu system was the challenge.

I echo what others have suggested, include the camera with applications that would give closed-loop feedback to robot systems less challenging than autonomous navigation. Get the robot or manipulator close via a driver and then engage the software lock. Those types of grabbing, dropping, color sorting functions would increase robot performance reliability dramatically.

APS

Re: CMUCam Next Year?
 

FIRST needs to put more emphasis on autonomus, and less emphasis on arbitrary sensors. Teams don't need "cool" additions to the kit to motivate them, they need a game which makes autonomus a worthwhile pursuit. If you don't do autonomus well, you should be in a whole lower class of competition. The games should be designed such that the top ranking teams are teams with effective autonomous operation.

Oh, and I was very dissapointed last year to find my CMU cam was D.O.A., as were seemingly a large percentage of them. Though looking back, I'm glad I didn't order a replacement because it seems as though the varying lighting conditions made the whole thing fairly unworkable...

Re: CMUCam Next Year?
 

I would be very, very surprised if the CMUCam was not in the kit. It was the most advanced sensory piece of equipment and yet it wasn't used. Why? Becuase it wasn't worth enough strategically. Why spend all that effort for a bonus that will barely last 15 seconds past autonomous in the finals? If FIRST had said "Oh and by the way, if you cap the center goal in autonomous, you unconditionally own that goal for the rest of the match, we would have seen people cap it autonomously. Also it doesn't make sense to say, "very few people could accomplish a difficult task, so let's give them an easier one."

You don't have to do "the task" for autonomous so why make "the task" easier? For exampe in 2004, our robot went forward, grabbed a goal, and brought it back for the ball dump. Does it accomplish "the task"? No. Is it awe-inspiring? No. Did the programmers-in-training learn something while accomplishing it? Yes.

Re: CMUCam Next Year?
 

EXACTLY!!!

:D

This is, unequivocaly, the BEST POST I've EVER SEEN on ChiefDelphi. EVER!

Re: CMUCam Next Year?
 

Well, I have heard that the autonomous mode will play a much more vital roll in this year's game which my team coach never fails to remind me. So, consequently, it would be advantageous to show the prestige of any robot's autonomous mode by using the CMUCam, and possibly have an upper hand in the score early on in the game to a larger extent than the previous years.

If it isn't going to be used, then I'll just be flabbergasted. Simple as that.

Re: CMUCam Next Year?
 

Once again, very well said!

My only concern with the CMUCam is that it is color-based. Last year, even the practice field at the St. Louis regional had different lighting, prohibiting any testing / refinements of the autonomous mode. Also, FIRST only distributed the exposure calibration values to teams. That is NOT all the information needed to get that system working reliably! The teams (mine included) that used RGB windowing (much more stable) were out of luck.

All that being said, I definitely want to see it come back this year. :D

Re: CMUCam Next Year?
 

I might be wrong, but i believe that at Sillicon Valley they gave us the RGB values for each side. Of course, we were one of the last regionals, so they may have changed it by then.

Re: CMUCam Next Year?
 

We were at St. Louis. I helped to calibrate the camera at the beginning of the day, and in the process got the RGB numbers I needed. :D

Re: CMUCam Next Year?
 

i haven't used it but iv'e been reading the help and reading the code and it all looks so confusing to try and program a robot to follow something. yes its probably easy for u vet teams but for a team only in it for a couple years and the fact that i'm programming and i just barely understand how our past codes work i don't want to use it. our team (1075) used a pre programmed code in '03 and '04. both years we had now problems. in '04 we were one of the few teams that could manage to knock the ball at the side off during autonomous. with pre programmed theres only the battery power that plays effect whereas with the cams and infra red sensors and gyros and compasses you have lighting conditions, battery power, parts to fail, other sources of interference. we experimented with a compass but simple metal objects would completely throw it off. the only way they could MAKE us use the cam would be to actually check our code everytime we go to the field. otherwise anyone could program the cam to do stuff and make the bot look like its following the cams instructions. so theres something so u don't have to worry so much about it. all there trying to do is give us a different way of doin autonomous. honestly i think its probably alot more reliable than the infrared sensors and line tracking and maybe the gyros(i was watching the '05 video and the gyro bot ended up about 45degrees turned from where it was supposed to be.