Extraneous Lighting

[R.C.]

Cd,

When we were at the Long Beach Regional, a fellow judge, also the owner of Mechanical Concepts stopped by our pits. We had a lengthy conversation in the pits about our bot and the camera. He told us that the camera should be mounted at the peak of the bot (highest spot possible) and put a hood over it. That way the lightening does not affect the camera TOO MUCH. We never got around to do this but were wondering how affective it would be?

-RC

[Francis-134]

Placing the camera at the top of the robot will ensure that it is not blocked by anything (robots etc.). I would think that placing a hood over the camera may help a small amount, but I would think it will still not help in long-distance lighting issues such as washout/inconsistant lighting.

[nathanww]

Yeah, the problem with this is it's fairly hard to position the camera so that you can get a full view of the vision targets at fairly close range without looking directly into the lights--so having a hood doesn't help you so much.

[Cow Bell Solo]

I would think the hood is more for protection for the camera so you don't have a ball come flying or for some reason a robot hits the camera and busts it off the robot and breaks it.

[Greg McKaskle]

The higher the camera is mounted the better -- I agree. Fewer objstacles, the cross section of the cylinder is closer to a rectangle, and distance calculations may also be simplified with a smaller angle. The biggest reason, though is to avoid having the camera stare into the lights. Substitute your eyes for the camera, and give it a try. Height helps.

As for a hood. Again do the eye substitution and put on a ball cap. It will help with light from a certain direction, but you still may not be able to avoid looking into the lights in some situations. Continuing the analogy, you may think putting black smudge beneath the camera will help, and it will make the robot to look meaner, but a better implementation would be to avoid shiny materials like lexan in front of and beneath the camera that will reflect overhead lights back into the camera.

Continuing down this line of thought, you may be tempted to put sunglasses on your camera. Good thought, but testing it out, you will find that if set to auto-expose, the camera will just adjust the exposure to rebrighten the entire scene, pretty much the way your eyes do. What about polarizing lenses? Unfortunately, the light from the field isn't polarized, and the target material doesn't polarize the light either, so in that case, polarized lenses work the same as colored ones, and they can't selectively filter out glare or bright spots.

If you find something that works miracles, please share.
Greg McKaskle

[Daniel_LaFleur]

Originally Posted by Greg McKaskle

The higher the camera is mounted the better -- I agree. Fewer objstacles, the cross section of the cylinder is closer to a rectangle, and distance calculations may also be simplified with a smaller angle. The biggest reason, though is to avoid having the camera stare into the lights. Substitute your eyes for the camera, and give it a try. Height helps.

As for a hood. Again do the eye substitution and put on a ball cap. It will help with light from a certain direction, but you still may not be able to avoid looking into the lights in some situations. Continuing the analogy, you may think putting black smudge beneath the camera will help, and it will make the robot to look meaner, but a better implementation would be to avoid shiny materials like lexan in front of and beneath the camera that will reflect overhead lights back into the camera.

Continuing down this line of thought, you may be tempted to put sunglasses on your camera. Good thought, but testing it out, you will find that if set to auto-expose, the camera will just adjust the exposure to rebrighten the entire scene, pretty much the way your eyes do. What about polarizing lenses? Unfortunately, the light from the field isn't polarized, and the target material doesn't polarize the light either, so in that case, polarized lenses work the same as colored ones, and they can't selectively filter out glare or bright spots.

If you find something that works miracles, please share.
Greg McKaskle

Greg,

Do the spotlights that FIRST uses put out full spectrum white light?

If not, then can a team (Knowing the height of it's camera, and knowing the size of the target) filter out all but 1 of the 2 colors and track on that?

The reason I'm asking is because (if you know your camera height) you can tell whether your target is on top or bottom by angle within the frame, and (know the size of your target) you can tell how far away it is by the area it fills within the frame.

By only tracking 1 color you can filter out all other light (you cannot easily filter out optically for 2 colors not next to each other within the spectrum) which should let you filter out even the brightest spotlights, unless they put out full spectrum white light.

[Greg McKaskle]

The field typically has around 30 floodlights on each side of the field, which output warm white light. If you were to plot their power vs spectrum from 400nm to 700 nm, they would be pretty flat, but higher in the red end of the range. So any analog optical filter will see them as a really bright spot of your filter color plus a haze of the filter color as the light from the flood hits the entire sensor

I think it is pretty safe to think of the color threshold as a color filter operation, but it can differentiate between white and other color combinations, not just a single peak like optical filters. So the purpose of doing this would be to replace the digital filter with an analog to get rid of the processing, but as noted, you can't substitute them one-for-one.

Greg McKaskle