Camera Losing Sight Close Up

[ebowla]

Well, so far we have been able to get the camera to work properly to a point. It tracks the light perfectly, and when it loses sight of it it does its search pattern to find the light again. Heres where the problem comes in. When we move it close enough to the light (which is at the height it will be at the competition) it starts acting odd. It will find the light for a moment, then instantly lose it and refind it, without doing its normal search pattern. It repeats this constantly until we move it further back. I made sure that it could see the light at all times, nothing was blocking it. I also checked the max tilt value and it is far enough back that it would allow for it to go directly under the light if need be. Any ideas of why this pattern of locking on and losing sight over and over when close to the light is happening, but not when the camera is further away?

[marccenter]

I can't answer your question, but just wanted you to know that Team 1286 has similiar camera problems. We can get the camera to work and then suddenly it stops working. In order to get it to work today, we had to remove the green light from the box and expose it so (turn the green lights around) the camera could see it directly about two feet from the camera. This whole camera/light business does not appear to be a very robust system at this point.

I will reading more threads to help figure out our issue. One thread I just read suggested making sure the 12V battery powering the light is fully charge. Another, that the 7.2 V backup battery (which powers the camera) needs to be fully charged as well.

[gnirts]

Originally Posted by ebowla

Any ideas of why this pattern of locking on and losing sight over and over when close to the light is happening, but not when the camera is further away?

Is what you are talking about almost like the camera is rapidly twitching back and forth between the light and another point? Our camera does this sometimes, yet fortunately never at the height of the lights in the competition.

Can anyone comment?

[T3_1565]

it may be from the angle at which it is seeing the light, when we started testing it on our laptop and you do that test in which it shows you a blue box of what it is tracking as green light. depending on the angle that box gets bigger and smaller try hooking your camera up for bench testing and postion it at the point your having problems and use the program to see what the camera is seeing ( if thats confusing in any way I'm sorry, I'm not the programmer so not sure what to call the programs)

[petek]

Have you taken a look at what the camera is seeing when it's close to the rack (with LabView)? You may find that the light appears as a very thin line of pixels, with a low tracking confidence value. If you have filtering set above 0 (and therefore subtracts pixels), when the camera is very close to the rack the light may "disappear". If that is the case, you may get better results with filtering set to 0, or by placing the camera farther back and/or higher on the robot to give it a better viewing angle.

[Tom Line]

Hey - check what your maximum tilt angle is defined as. It's in your tracking.h file and orginally it isn't set to allow your camera to look nearly straight up - assuming you are using Kevin's code.

[ebowla]

OK heres the deal, Yes I'm sure the max tilt is fine. It is set more than needed. And yes it does twitch, in other words it moves a lot, but doesn't do it's normal search pattern. Remember this is only when it is really close to the light. As for a backup battery, we are using a (not sure what it actually is but) power source thing that lets you choose the Voltage, so I'm sure it has charge. And same for the main battery.

Petek thanks, your post probably is the problem or at least it makes the most sense, I'll check it the next chance I get.

[pheadxdll]

We've had that problem to. The CMU cam looks up to find the light and while it moves to a level it sees it but it still moving to the level. Once it gets on the level it realizes it has lost it and returns to the regular position. Thus in a loop. Right now the camera is mounted on the RC on the base of the robot. Hopefully when we move it a couple feet higher to the arm, this problem will solve itself. I don't want to go in and start editing Kevin's searching code. :X

[lkdjm]

Very similar problems here at 564. We are considering only using the camera to get close to the rack, and then use other methods to reach the rack. Check out our thread about alternate ways of reaching the column without the light. http://www.chiefdelphi.com/forums/sh...ad.php?t=54024
Please add your ideas!

[ptan]

Also, don't forget to check the electrical as well. When the camera is up close, it tilts a lot more, and that can put a strain on the power cable to the CMU camera.