pic: camera locked on light 143ft away!!!



we saw a post that said that camera could only find the light to about 20ft, well needless to say we disproved that…we found it 143ft away. After that point the camera would only lock on for a second or 2, but before that it would find and lock on. We are very excited about this recent decelopment

thats pretty sweet

what were your config parameters (AGC gain control, blue gain control, red gain control, saturation control, brightness control, auto-exposure control)?

just wondering, because so far i’ve got the cam to track out to about 15ft

thanks!

We use a modified version of Kevin Watson’s default camera code. The modification only improves the scanning algorithm. Actual camera configuration parameters are left unchanged.

Please add your input to the Camera Tracking Maximum Distance poll and thread.

excellent!

now to get a ball fired at 12M/S to travel that far is would have to be spinning at…[reaches for calculator]

:^)

this really is awesome - you could use the camera to triangulate your position from anywhere on the field, by looking at both beacons!

I’m not an engineer, but don’t you need three known points to triangulate?

Well, you know the heights of the goals, and you can get the angles to both goals from the camera. However, short of using two cameras, (which is possible), you’d need a really complex/thought out light/switch algorithm.

What would happen once the other target gets shut off? You’d probably be out of luck then.

Yeah, but you really only need to know your absolute position during autonomous.

Actually, you really only need to know your relative position during autonomous. Simply how far away and at what angle you are to the goal is enough information to score.

Yeah, really. For this game, there is no “real” reason to know your absolute position. If you wanted to, it could be done though. cough waste of timecough:slight_smile:

the two lights are a fixed distance apart. Thats one side of the triangle. If you can measure the distance to both of them you have defined the lengths of all three sides, so you know where you are on the field

or knowing the angle to both lights will determine your position on one side of the field.

If your robot wanted to navigate itself to any of the corner goals for either an offensive or defensive play during autonomous mode, it may be beneficial to know your absolute position on the field.

I’m no expert, but it seems to me that with only two points, there would be two possible positions you could be in on the field, and it would be impossible for you to determine which one you were in without another reference.

I have attached an image attempting to demonstrate this, someone tell me where I am wrong.

robot position.JPG


robot position.JPG

The lines going to the robot position should be rotated 180 instead of mirrored.

That’s where a gyro comes in handy!!! (as so graciously provided in your kit!!! )

[/IFI_rant]

wow thats really awsome…we’re having a hard time trying to get the camera to lock within 30 feet or something wit out any interference from sudden movements of the robot. every time the “test” bot moves or jerks around the camera would freeze up…which really suck. i wonder if we got it working now? well 143 ft is impressive…very impressive. good luck to u guys :yikes:

Ah, I see what you mean. But wouldn’t you still have the uncertainty?

Absolutly, a gyro would be needed. Or some other crude form of figuring out which way you are facing. I suppose if you could get a consistent lock with the cameras, the cameras would know when they’ve exchanged vision sensors when you lose your lock for a few seconds be cause of turning your robot.

Oh and to whomever posted this picture, please bring down the size of your canvas in paint for future pics.

Remember that you would still be abel to be in 2 possible positions assuming that your not collinear to the 2 lights