artificial horizon

[dalbert02]

I accidently posted this in the UAV section and I really should have posted it here. Anyways, I think this device could be used to make an artificial horizon.
-dave
http://www.rcgroups.com/forums/showthread.php?t=658635

[MatC]

I'm not sure that the $299 price tag includes the display readout, and even if it is, it won't be genlocked to your camera.
When I build mine, it will be like this:
1. Horizontal detection. Really hard. Easiest is to build your own gimballed gyroscope. Better might be a combo of solid state rate gyros with some sort of horizon/light/IR detector.
2. Display. Relatively easy to overlay onto a composite signal with a microcontroller or two, phase locked loop and perhaps some SRAM. (I did this back in the old days with a 16F84 [overclocked a bit], PLL, Sync Detector and not much else).

[tommyt]

dave check PM's
Tom

[deh6]

Is an artificial horizon really needed? Thinking back to my full-scale flying days and IFR training, the artificial horizon makes life easier, but one can flying a plane with just a rate gyro (yaw), "ball" (slip indicator), airspeed, and compass. If one wants to fly straight, keep the needle on the rate gyro centered (not turning), check the ball (no slip/slide (which is not a problem with models that are deisgned to be stable), check the airspeed (increasing means add some up elevator, decreasing add some down), and check the compass to keep on heading.

MEMS rate gyros drift, so they have to be constantly corrected; for yaw the compass can be used to do this. If needed, the angle of bank can be calculated from airspeed and rate of turn. That gives part of the artificial horizon.

The other part of a true artificial horizon is pitch. I suppose one could combine airspeed and engine/motor rpms with some sort of calibration for the plane to give a reasonable estimate of pitch.

A lot depends on the purpose of the artificial horizon. If it is for UAV and an autopilot it really isn't needed. If it is for RPV where someone wants to fly it in a virtual reality, then it is worth some effoirt, though not really necessary.

[dalbert02]

I was thinking RPV at night where the video image sent does not clearly show the horizon. Such as: http://www.bijouxdesigns.com/video/night%20flight_final.wmv
-dave

[deh6]

Cool video! It demonstrates the need for a reference, especially during turns over the water (JFK Jr lost his orientation). With the situation in the video the durations when there is no horizon reference are relatively short. (Now if you are going to fly to the Bahamas or Marthas Vineyard it would be a different story.)

The simplest solution that comes to mind is a rate gyro (yaw) that moves a servo with a thin wire in the field of view of the camera. The deflection of the wire would be proportional to the rate of turning. That way one would not inadvertantly tighten up the turn into the classic death-spiral. More complicated would be to send the rate gyro to the ground station and insert something into the video.

The zero point of the gyro will drift a little during the flight. If that drift is too large it could be re-zeroed, by flying the plane level when the horizon is in view and resetting the zero point of the gyro by using a flap switch, or spare channel.

[MatC]

If you just want to indicate the output of the rate gyro, you don't need to zero it.
If you plan on integrating the rate gyro output, drift becomes a major issue, I don't think you can get away without at least an intermittent absolute orientation reference.
So that leaves... IR sensors to look at the sky (won't work at night, amongst the mountains, or above water), gimballed gyroscope, accelerometers, other.

Accelerometers: I'll think more about this one.

Other: anyone tried measuring the strength/direction of incoming GPS satellite signals? That might work, esp. as GPS units output the positions of the satellites. Just a random thought.

[deh6]

If you just want to indicate the output of the rate gyro, you don't need to zero it. True for the mechanical gyros. For the MEMS type there is a voltage for zero rate (about 2.5v) which is initialized, and that has some drift, though it may not be enough to be a problem.

IR sensors to look at the sky (won't work at night, amongst the mountains, or above water) I did some searching on these and someplace (it may have been patents) they showed the temp difference between the sky and horizon and claimed the IR sensors will work at night, in clouds, over water, and over snow, though the temp difference with snow was getting low. Valleys/mountains would be a problem. I don't know how well the claims hold up in practice, but if they are right, they work a lot better than I expected. They are extremely accurate for a satellite detecting the horizon.