CMUcam initialization time
 

Hey all,

I was just wondering what the average initialization time with, lets say, Kevin's frc_camera_2 code was.

Thanks much,
/dan

Re: CMUcam initialization time
 

The camera initializes very quickly, usually in a matter of seconds.

If what you are worried about is the camera "booting up" in time at the start of autonomous mode, there is a very clever reason this won't be an issue. The robot is ON before the start of a match, just the master processor is preventing it from sending any signals to the motors. However, the CMUCam is linked to the RC by serial, and is talking to it this whole time.

Essentially, the camera will initialize as soon as you flip your robot on, well before the start of the match.

Re: CMUcam initialization time
 

Thanks for the answer Greg, but it's not the one i was looking for.

My question is, if I switch color parameters and register values, how long would I need to wait until i could poll the data (i.e. Until the values registered in the camera, and it would be returning data based on those values).

thanks again,
/dan

Re: CMUcam initialization time
 

Old thread, but...

I haven't tried it personally, but other people on these forums are reporting that switching colors rapidly is inefficient. I know everyone has heard the rumors about multiple colors last year. I have something else in mind...my code will deal with multiple colors. :D

I can't wait to see what the GDC is going to dish up!

JBot

Re: CMUcam initialization time
 

Based on testing (in easyC) the reset time is about 2 seconds. Which would be fine as long as the colors don't change rapidly. Like if you had to pickup an object that was blue and drop it off in the green square. That would be fine as long as you reset the camera right away. Hopefully they don't use red or blue as the robots might go after the LEDs on the side of the robots. But, that may be why they went to flags last year.

Re: CMUcam initialization time
 

They'd also have to ditch the drop lights in the alliance station. Given that those aren't just decorative (they're status lights for the robots--when the light is flashing, something is wrong), I don't think they would do it.

Re: CMUcam initialization time
 

I would personally like to see an updated firmware for the CMUcam, like people say it takes a few second to change color parameters on the fly.. i have a hard time believing that this is a hardware problem causing this huge delay

Re: CMUcam initialization time
 

i toaly agree with u on that 1 .. im pulling my hair out half the time when my camera doesnt boot up properly.. which is really strange cuz its stops the master thing to stop working as well:S:S werid! .. anyways .. i've been able to switch colors just not in 2 secs:P

Re: CMUcam initialization time
 

Not if those drop lights were the vision targets themselves. If the goal was positioned right in front of the driver's station, then you would be able to use them just as regular lights. They might have to add diffusers, and maybe tweak the field controller code a bit, but they could become active elements themselves.

Re: CMUcam initialization time
 

Does anyone know if the cmucam's firmware is opensource?

Re: CMUcam initialization time
 

No, it is not, but if you Email the CMUCam developers they might share it with you.

Re: CMUcam initialization time
 

My guess would be probably not. Based on their site:
http://www.cs.cmu.edu/~cmucam2/downloads.html
the firmware looks closed-source. They offer .hex files for those who have access to an SX-Key programmer to upgrade the firmware. I would guess it would take a lot of reverse-engineering to write custom firmware for the CMUcam. You might consider firing off an e-mail to IFI just to see what they have to say about it.

JBot