Accelerometer for stabilizer

[vpatron]

Hi folks,

I wanted a simple system to keep a plane flying level. Can a 2-axis accelerometer be used?

I know it can be fooled by g-forces on the plane but basically the plane would be flown and stabilized before turning on the simple autopilot which just keeps the plane flying straight.

Has anyone tried this?

-Vince

[Unterhausen]

I think you can integrate and average the accelerometer outputs to do this. The Analog Devices accelerometer outputs that I have seen look pretty good. I wonder if a gyro would work better? Aren't they about the same price? Sparkfun has some interesting looking products along these lines.

I studied aircraft control in the mid-80's and never used it, so I'm very rusty. I've resolved to go back and look it over.

[vpatron]

The Analog Device 2-axis accelerometer is less than $5 (in 1,000 pc quantities). I figure if 1) the plane is inherently stable (polyhedral wing electric glider), 2) I have GPS data, 3) no rapid movements are needed, I can use this simple accelerometer to keep the plane steady, and use GPS to adjust the course.

I figure the GPS data can also correct the accelerometer data. For example, if the plane is banking right, the accelerometer will indicate banking but the controller can ignore the roll sensor and just monitor the pitch axis during turns.

I looked at the absolute pressure altimeter and it needs more post-processing. For gyros, I don't know the price, but seemed like the 2-axis accelerometer might be simpler.

THanks,

-Vince

[LukeZ]

Vince, although I too am partial to the accelerometer/gyro method of state estimation, it sounds like for your needs horizon detection might work just as well and be a lot easier to set up. Have you looked into this? The FMA Copilot is used by a lot of people. Also RadioHound on these boards is selling just the IR detector array for I think $10 (http://radiohound.com/ecommerce/os/catalog/), that you could interface to your PIC and so have a bit more control of what you do with the output.


Luke

[vpatron]

Luke, Thanks! That's an excellent site and I now understand how the FMA Copilot works.

Yes, the copilot might work just fine for what I'm doing in the "Phase 1" plans. In fact, I just ordered it yesterday. I'll get some airtime with that and see how well it works.

FMA Copilot coupled with GPS might even work fine for what I'm trying to do. THanks,

-Vince

[LukeZ]

Vince, by the way, what are you trying to do? ;)

My personal goals involve long distance/long duration flights - for that reason I don't think horizon detection will be the best solution, given that I will certainly be flying at night and quite possibly also flying in inclement weather. I understand that the CoPilot is purported to work at night, but it wouldn't in fog or clouds, and even the claims about night I'm not sure I completely trust. Also it seems to me it would need to be re-calibrated when transitioning from day to night, but that wouldn't be very practical in flight.

Therefore I'm going to have to figure out the accelerometer/gyro/Kalman approach... which I'm dreading.


Luke

[clolson]

Something to consider (or maybe not consider) is that if you pick a stable plane as your platform (such as a typical trainer) it's possible to depend on it's inherent stability for wing leveling and just steer gently with the rudder. They don't necessarily have spectacular tracking, but people have demonstrated that this very simple approach can work.

As I understand it, the co-pilot simply stops doing anything when it can't sense much differential between sky/ground temps.

So in the back of my head I'm thinking that my Kadet Sr. (which I hope to eventually migrate into a UAV) will track quite well with the co-pilot working, and only have degraded performance when the copilot can't tell the difference between sky/ground. That could be wishful thinking, but manually I've been able to steer the Kadet Sr. around using rudder only (no copilot at all) and it stayed relatively stable and if I stuck to gentle rudder inputs it happily stayed sunny side up and went where I steered it.

I don't intend to ever let it get out of my sight, but gps only navigation with no autopilot is something I hope to explore when I get that far ... if the theory works it would allow you to possibly fly into clouds or weather and continue your mission.

And also bear in mind that just because something is possible or can be done, doesn't necessarily mean it's a good idea ...

Curt.

[sesat]

Vince,

No.

Ram.

[vpatron]

Luke,

I think you're right about FMS CoPilot. I also heard it can get fooled when flying with the sun very low on the horizon. Just the same, I just bought it to test it out and get the plane in the air ASAP. I'll sell it later on if I come up with a better system.

I'm also starting to lean on a gyro for bearing hold, and a pressure altimeter for altitude hold.

You guys are way ahead of me. I'm in "Phase 1a". I'm doing this in phases so I can get at least some intermediate satisfaction and to test the market for aerial photography. But I'm doing the system planning now so that I just keep adding, not have to change planes, change controllers, change sensors in mid-development.

Here's the plan:

Phase 1a: Aerial photo plane, video downlink, pan/tilt camera servos
Phase 1b: Add 2nd control for camera operator
Phase 1c: Add GPS downlink + live mapping

*Begin some commercial work to test out the market and determine needs.

Phase 1d: Add telemetry (battery level, radio link level, RPM, airspeed, etc)

Phase 2: Add altitude hold and bearing hold

Phase 3a: Add uplink data path and GPS autopilot. Do simple nav: fly in a level circle around target, fly a 2-D route at fixed altitude. Add in-air route upload capability.
Phase 3b: Full autopilot, fly several mile 3-D route and take photos autonomously for large area photography.

-Vince


Vince, by the way, what are you trying to do? ;)

My personal goals involve long distance/long duration flights - for that reason I don't think horizon detection will be the best solution, given that I will certainly be flying at night and quite possibly also flying in inclement weather. I understand that the CoPilot is purported to work at night, but it wouldn't in fog or clouds, and even the claims about night I'm not sure I completely trust. Also it seems to me it would need to be re-calibrated when transitioning from day to night, but that wouldn't be very practical in flight.

Therefore I'm going to have to figure out the accelerometer/gyro/Kalman approach... which I'm dreading.


Luke

[vpatron]

Hi Curt, yes, I had the same thoughts. The platform I'm using is the Push-E Cat electric EPP foam glider, inherently stable like your Kadet Sr. but will take lots of crashes with no damage to the airframe... :) I've got the inventor's prototype twin motor "cargo" edition for even more room.

For the basic aerial photography, an altitude hold that I can kick in and then just steer around with rudder will be perfect. The Slow Stick I'm using now tracks horribly and a lot of photos are crooked. Given the altitude hold, add the GPS and I will be able to do this at very high altitudes.

I just ordered the FMS CoPilot. I think it will be so-so, but I think true altitude and bearing hold (not pitch/roll stabilization) will be better.

I'd still like GPS autopilot at some point to handle large areas or for repeat customers. For example, a construction project may need photos monthly or weekly to show progress. The plane can fly the same route and shoot the same exact angles over and over.

Regards,

-Vince


Something to consider (or maybe not consider) is that if you pick a stable plane as your platform (such as a typical trainer) it's possible to depend on it's inherent stability for wing leveling and just steer gently with the rudder. They don't necessarily have spectacular tracking, but people have demonstrated that this very simple approach can work.

As I understand it, the co-pilot simply stops doing anything when it can't sense much differential between sky/ground temps.

So in the back of my head I'm thinking that my Kadet Sr. (which I hope to eventually migrate into a UAV) will track quite well with the co-pilot working, and only have degraded performance when the copilot can't tell the difference between sky/ground. That could be wishful thinking, but manually I've been able to steer the Kadet Sr. around using rudder only (no copilot at all) and it stayed relatively stable and if I stuck to gentle rudder inputs it happily stayed sunny side up and went where I steered it.

I don't intend to ever let it get out of my sight, but gps only navigation with no autopilot is something I hope to explore when I get that far ... if the theory works it would allow you to possibly fly into clouds or weather and continue your mission.

And also bear in mind that just because something is possible or can be done, doesn't necessarily mean it's a good idea ...

Curt.

[helibel]

Maybe an interesting URL for realizing your dream might be :
www.u-nav.com/

The pico pilot looks nice and adequate for your project.

Helibel

[vpatron]

Thanks Helibel, but I saw their system already and it seems interesting, but I want to make my own... :)

-Vince