Reliable 2006 Camera Settings?

[Travis Hoffman]

I'd like this thread to be a one stop shop for useful camera settings used by those teams who've used the camera most successfully. For you camera geniuses out there, please consider posting your "reliable" settings here.

I personally use Kevin Watson's bells and whistles code. Everything works pretty well; I'm just looking for tweaks to the settings that can make the performance more reliable and repeatable from location to location.

I'm looking for answers to the following questions, along with any other suggestions you can provide:

1. Which settings work well for you without requiring modification from event to event?

2. Which settings do you have to tweak at each event? Are there particular ranges of values you use, or could they be all over the map?

3. Which settings are unimportant to green target tracking and what are they set to on your machine?

4. Has each event provided a list of camera calibration values at pit admin? If so, are those values useful?

5. Have other similar-colored and sized arena lights caused any vision light tracking problems at your events?

6. How do you think the brighter lighting conditions in the Georgia Dome will affect camera performance?


Thanks in advance for your input.

[Tom Bottiglieri]

Originally Posted by T. Hoffman

6. How do you think the brighter lighting conditions in the Georgia Dome will affect camera performance?

We're not using a camera this year, but I know 237 was having problems tracking the vision target at Chesapeake, which was lit by daylight.

[Bharat Nain]

Originally Posted by Tom Bottiglieri

We're not using a camera this year, but I know 237 was having problems tracking the vision target at Chesapeake, which was lit by daylight.

Default values worked wonders for us. We wrote our custom tracking code which did no use servos but directly controlled the motors.

[Rickertsen2]

I have used the default values without ever thinking twice about them. I have rewritten just about every other aspect of the default camera code, but never had a signifigant problem with the calibration. There have been cases where our camera has tried to lock onto overhead lights, but this was resolved by limiting its range of motion and never actually presented a problem on the feild.

At the peachtree regional, it was obvious that precautions had been taken to prevent team's cameras from locking onto field lighting. I am assuming measures will be taken at the nats as well.

[EricS-Team180]

1. Which settings work well for you without requiring modification from event to event?
2. Which settings do you have to tweak at each event? Are there particular ranges of values you use, or could they be all over the map?
3. Which settings are unimportant to green target tracking and what are they set to on your machine?

We just haven't diddled the settings enough to provide data on 1-3. We've used both the default #'s and some re-calibrations we made in our shop. In our gym, the defaults worked best. At UCF the hardware was working, but we edited Kevin's code a "bit much" and spent the regional shooting manually. At Palmetto we used the default #'s and found out the camera had come way out of focus. In competition, the camera worked well from about 15ft., which is about the distance from which we were shooting, anyway.

4. Has each event provided a list of camera calibration values at pit admin? If so, are those values useful?
I didn't check at UCF, but did ask at Palmetto. The answer we got was "use the defaults"

5. Have other similar-colored and sized arena lights caused any vision light tracking problems at your events?
At Palmetto, the large flourescent arena lights were off on the field side, but on for the pits. The practice field was illuminated with flourescent lights, too. The camera quickly and consistently locked on these lights - with the default #s - any time they were in the camera's field of view. The lighting on the field was OK for competing. Once, while aiming the camera across the pit area (hooked up to a PC for focusing), it recognized an open doorway through which sunlight was entering, and on the pit video screen.

6. How do you think the brighter lighting conditions in the Georgia Dome will affect camera performance?
There'll be an effect, just how much I don't know. I've read suggestions about cutting way down on the saturation, so we'll be experimenting. I got a suggestion from 108, to use 2005's java based program for calibrating ... anybody got a link?

For ATL, we may rig up a portable camera/PC arrangement with our spare camera from last year, and send a crew around, "surveying" the fields from however close they can get. If we do, we'll gladly share anything we find out.

see ya there!
Eric

[Donut]

We only attended one event (AZ Regional), so we've only used the camera tracking there, and in a classroom and a gym at our school.

For questions 1-3: we used the default light and color settings, but changed alot of the tracking numbers. We adjusted the max/min tilt and pan values, as well as the step sizes between them, but how much you want to vary that depends on the camera placement on your robot and the range of your shooter/rotator, so our numbers won't help you. We did effectively cut our camera tracking from 3 tilt and 4 pan positions to 2 tilt and 3 pan positions with the effect of actually locking onto the target faster and more accurately, so I would say that you can easily cut your camera's search time by half without affecting it's reliability with a little testing. You can also cut down on the time the camera spends at each position by a little bit.

4. AZ provided camera numbers, but our camera was already working fine so we ignored them.

5. We have had no problems at competition with outside/arena lighting; at our school we had problems tracking our flourescent lights at first, but this was easily remedied by changing the condition for having found a target to also include a check that the camera confidence was above a certain number (in our case, 20). Since then we have had no problems, even with flourescent lights only 1 foot above our green light.

6. If you use the solutions teams have been using to distinguish the light more effectively, and prevent your camera from looking straight up at the lights, I think you'll have no problems.