UAV newbie

[nevada]

Hey,

Sorry if this has been asked earlier, please direct me if it has, I did go through the forums though..

I'm starting off with a UAV project based on a airplane which I intend to use for the prophanging maneuver apart from regular flight...

Since I really don't have much time to prototype and experiment, I'm looking at sparkfun for the components like the IMU (http://www.sparkfun.com/commerce/product_info.php?products_id=9184), Servo Controller (http://www.sparkfun.com/commerce/product_info.php?products_id=8897),

The idea is to integrate the board with my plane, which is going to be controlled using a futaba 6 channel and basing a wireless camera on it, which'll be controlled directly by the futaba. The driving signals from the futaba shall be fed to a custom made atmega 128 board running at 16 MHz, which will accept the futaba values and the error values from the various sensors on the IMU and give an appropriate output to the servo motor controller.

I intend to use the XBee to calibrate my sensors on the PC itself and possibly log flight data on the go, although the 1 mW Xbee gives a maximum range of only a 100 metres outdoors, its a good enough platform to experiment with methinks.

However this sounds great in theory, I'm very certain there are some flaws in practice, any ideas/inputs on the products I'm buying etc? And further, on how to implement this in the best way?

1. Signal Mixing - The idea so far is to figure this part as we go along. We've worked with gyros alone in making a human interface device, but not with accelerometers. Obviously with a low speed processing as on the mega 168, we wont be doing any kalman filtering (or for that matter even if we did have an ARM, it'd take us a very long time to do that :) ) , but the idea is to implement something basic, any ideas here are welcome.

2. We've made RC controlled gliders etc before, but obviously a UAV is a completely different ball game. Inputs on construction are appreciated. Currently we've built our own airplane because the commercially available ones did not meet our design expectation. The plane is made partly out of balsa and partly of carbon fibre rods.

3. I'm not putting in a GPS as of now, I'm concentrating on a controlled well adjusted flight, and then I'll move on..The current objective is level maintenance in flight and prop hanging.


Thanks a lot

[Gary Mortimer]

Well if its prop hanging your after you will need a pretty complex autopilot!

Deans (attopilot) playing with a 3D type airframe, he would be the chap to talk to.

Not sure if hes got it prop hanging though.

[wonginator1221]

3. I'm not putting in a GPS as of now, I'm concentrating on a controlled well adjusted flight, and then I'll move on..The current objective is level maintenance in flight and prop hanging.


The new FMA Copilot (http://www.fmadirect.com/new_applications/copilot2.html) is allegedly suppose to fill those requirements.

[Gary Mortimer]

Not sure how thermopiles would be accurate enough.

Someone here will know.

It must come with the Z sensor as standard this time as well.

[Nimski]

Not sure how thermopiles would be accurate enough.

Someone here will know.

It must come with the Z sensor as standard this time as well.

I agree. I doubt thermopiles would be quick enough in their response. Because of the delicate nature of a prop-hang, if your IMU is too late in telling your control system that the system has drifted, you'll end up oscillating or not getting a clean hover. But I've never tried this, so purely speculation. The IMU from sparkfun, coupled with the correct filtering should provide the necessary response time and accuracy.

Also, are you intending to the autopilot to initiate the prop-hang as well? How do you plan to acheive that.

[airmcn_3]

The MP2128 heli version would do the trick but you will need to use the GPS....

http://www.micropilot.com/products-mp2128heli.htm

We were looking at them for a VTOL project.