altitude sensors anyone?

[reversed_rocker]

for the collegiate first pilot program we would really like a way to be able to tell the altitude of our quadcopter. since i dont to trust GPS signals inside a building and i don't know of any altimeters that work at this low of an altitude, does anyone know of a good distance measuring sensor?

We're looking at measuring distances of up to 10m and to be accurate within about 6 inches or so. all of the sonic range finders i'm familiar with do not have a large enough range. the rules state that we're not allowed to have lasers of any kind, so that throws out laser range finders.

[kamocat]

You're correct that altimeters are nowhere near accurate enough.
I would recommend a MaxBotix SONAR sensor.
I'm guessing you want the narrow beam width, but it may take some testing to find out.
maxbotix.com

[DonRotolo]

The Maxbotix I have is limited to about 255 inches.
But surely other types have the range; perhaps laser-based.

[kamocat]

How high in the air are you? 252 inches is 21 feet.

With a laser-based sensor, you'll have to do some processing to deal with the noise in your reading (because you're measuring distance in a straight line, and there's ALWAYS clutter on the ground)

EDIT:
I suppose the other option is to use the ground-based vehicle as a distance reference. You can triangulate.
(To be honest, time-based methods with light are a pain to measure because it's such a minuscule length of time)

[theprgramerdude]

Whats the size of this aircraft, and what are it's processing capabilities? I've been told that with a little filtering and math on the inputs, it's possible to properly extend the Maxbotix sensors out to well over 100 feet if using more than 1.

[davidthefat]

Tried a camera? One of my mentors has this quadcopter that uses a camera point directly down to stabilize itself or something. So I guess if it uses edge detection to find the edges. It can find out if they moved or something.

[Bongle]

Kinect? (http://www.youtube.com/watch?v=4EGQZsu-kkk&feature=related)

Failing that, maybe a poor man's kinect? Maybe with a strong IR emitter on the bottom and an IR sensor, you could determine from the brightness of the received IR how far away from the ground you are. This would depend on how much the ground below absorbs IR so it might not work if you don't know what you'll be flying above.

[DavidGitz]

Is it possible to change your competition strategy? i.e., since you are indoors, fly closer to the ceiling and monitor your clearance to the ceiling itself.

What exactly is your motivation behind wanting to get your range at that far away? Their may be ways around this limitation, depending on your reasons. And since it is an indoors environment, are you able to place static objects ahead of time? Probably not, but it's a thought.

[Bryan Herbst]

Is it possible to change your competition strategy? i.e., since you are indoors, fly closer to the ceiling and monitor your clearance to the ceiling itself.

What exactly is your motivation behind wanting to get your range at that far away? Their may be ways around this limitation, depending on your reasons. And since it is an indoors environment, are you able to place static objects ahead of time? Probably not, but it's a thought.

The collegiate FIRST pilot is happening at championships. The field will be very similar to the FRC field, and will have a net over it (if I remember correctly). Though definitely a fantastic new way of looking at the problem, that won't work.

As for other suggestions- Kinect would be so much fun, but these are light quadrocopters (powered by arduinos), so I have a feeling that would be too heavy.

The other problem here is that there are platforms that robots must land on, so I am not sure how much help sonar will be. Not exactly sure what the altitude will be used for, but this may cause an issue.

[Chris Hibner]

For a sonar or light-based ranging system, I would recommend a Kalman filter strategy for error/noise rejection.

If you want to get fancier, add 6-axis inertial measurement unit along with the ranging system, and tie it all together with a Kalman algorithm strategy. This can work quite well.

[reversed_rocker]

thanks for all the help

so first of all, i think it's especially important to mention that we're using 2 controllers. one processor for controls and one for fancy sensor inputs (image analysis and whatever else we throw on there). the second processor may be an ARM 3530, so i think kalman filters on the matbotix sonic rangers will do the trick. After reviewing the rules, i found out that the ceiling is 16 feet high, so that should be within the 255 in range of the range finder.

i was going to play around with using the camera and an optical flow algorithm to do position hold, but unless you're looking at a fixed object on the ground with a known size and shape, you cannot use the camera to determine altitude.

[Gdeaver]

There are some high resolution pressure sensors available. With the proper calibration you might get 1 foot accuracy. Turbulence with a copter could be the killer. A tube with properly designed holes may help. They use SPI or I2C so the audrino low analog is not a n issue.

[Tom Line]

The maxbotix sensors are accurate over long distances (VERY accurate) but only if they are held still. They also have a noticeable lag time along with an overshoot-then-leveling-off issue when used with the analog (I'm not sure if the digital connection has the same problem).

I would consider combining two sensors. The ultrasonic for longer distances and the sharp for closer in where speed is important.

Sharp does have some longer range rangefinders (1.5m) that might work:

http://www.robotshop.com/sharp-gp2y0a02yk-ir-range-sensor.html?utm_source=google&utm_medium=base&utm_campaign=jos

[kamocat]

I have found that using the measuring the digital pulse on the Maxbotix SONAR is much less noisy than the analog signal. (It also doesn't have the trouble with maximum slew rate).
Granted, this is with FRC hardware. The characteristics of your control system will be different. (You may have little enough noise on your 5v power supply, and a pulse-length-measurement that is not as accurate as an FPGA, and so analog may be a better choice.)

EDIT:
about those rangefinders: They have a MINIMUM distance around 5-10 inches. (it will say on the datasheet). If you get closer than that, you will get false readings.