|
Posted by Adam Chapman on March 20, 2008, 6:21 am
> You are diving into the deep end. I imagine that controling an
> aircraft with this sort of precision is an extremely big challenge.
> But, regarding your question;
> First, I am not certain you understand how the Sharp distance sensors
> work. They emit a beam of IR light that is seen as a dot on the
> object being measured (well, you would see it if you could see IR).
> Then, a lens focuses the image of that dot onto a linear sensor (sort
> of like a camera, but with only one dimension, not two). How far left
> or right that focused dot falls on the sensor is what determines the
> distance.
> Other, cheaper, IR sensors work by simply emitting an IR light (fairly
> broadly) and then a sensor simply detects any reflected IR light. The
> brighter the reflected IR light, the closer it is presumed to be. So,
> it doesn't really detect distance. Rather, it is detecting if
> something is close.
> Neither of the above are good methods for distance measurements if you
> are tracking a small point.
> Ultrasonic range finding is another option, by perhaps having the
> drogue emit a pulse of light and an ultrasonic "ping" at the same
> time. Your airplane would then look for the light, and listen for the
> ping and measure the time difference between the two to determine the
> distance. But, I suspect that there will be too much noise in the
> environment for this to work.
> This leaves using two cameras to see your infrared beacon on the
> drogue. The difference in the postion on the image on the two cameras
> is the method of determining distance. It would also tell you the
> direction of the drogue in relation to the airplane. The CMUcam has
> the ability to automatically track colored objects, and communitate
> the coordinates to your CPU over a serial connection. Perhaps simply
> provding a red surface on the drogue is enough to target it.
> Joe Dunfee
Thanks, I was of similar mind about sonar. Last night I was thinking
about using maybe some webcam type sensors dotted around the aircraft
needing fuel and finding the 3D location of a bright beaon like the
drogue and you confirmed this morning that it would probably work.
Thanks again
Adam
| 
Yahoo! 
Google 
Windows Live 
del.icio.us 
digg 
Netscape 
XML site map
> aircraft with this sort of precision is an extremely big challenge.
> But, regarding your question;
> First, I am not certain you understand how the Sharp distance sensors
> work. They emit a beam of IR light that is seen as a dot on the
> object being measured (well, you would see it if you could see IR).
> Then, a lens focuses the image of that dot onto a linear sensor (sort
> of like a camera, but with only one dimension, not two). How far left
> or right that focused dot falls on the sensor is what determines the
> distance.
> Other, cheaper, IR sensors work by simply emitting an IR light (fairly
> broadly) and then a sensor simply detects any reflected IR light. The
> brighter the reflected IR light, the closer it is presumed to be. So,
> it doesn't really detect distance. Rather, it is detecting if
> something is close.
> Neither of the above are good methods for distance measurements if you
> are tracking a small point.
> Ultrasonic range finding is another option, by perhaps having the
> drogue emit a pulse of light and an ultrasonic "ping" at the same
> time. Your airplane would then look for the light, and listen for the
> ping and measure the time difference between the two to determine the
> distance. But, I suspect that there will be too much noise in the
> environment for this to work.
> This leaves using two cameras to see your infrared beacon on the
> drogue. The difference in the postion on the image on the two cameras
> is the method of determining distance. It would also tell you the
> direction of the drogue in relation to the airplane. The CMUcam has
> the ability to automatically track colored objects, and communitate
> the coordinates to your CPU over a serial connection. Perhaps simply
> provding a red surface on the drogue is enough to target it.
> Joe Dunfee