Hi all,We are using the thermopile sensors to estimate attitude of our airplane. The problem is that the paparazzi wiki mentions a nonlinear relationship between the angle and the voltage. Therefore, we can't use it unless we calibrate it experimentally and find the values for different angles and store them in a table. However, that would be cumbersome. It would help greatly if there was a mathematical relation or a lookup table between the voltage and angle, but so far we haven't been able to find it. Any ideas on this?- aravind, yogesh

You need to be a member of diydrones to add comments!

The people from Monash University seem to have used the IR sensors for quite a while. One of the ways, as described in that paper is that you assume a sinusoidal profile and use the peak to peak values voltage you get on the field before flying to calibrate the sensors. Then use the vertical sensors to constantly monitor the temperature range and update the peak to peak values and re-calibrate them again. I will have a look at the paparazzi equations.

There is a series of technical reports by Egan/Taylor from Monash University describing the characterization of infrared sensors (they use the FMA IR head modules) for UAV autopilots. They are all available via Professor Egan's website, but here's one (attached) with some info:

"Use of Infrared Sensors for Absolute Attitude Determination for UAVs"

From my reading, it seems the basic relationship (assuming a wide FOV for the thermopiles) is a "chopped" sinusoid. I haven't done any testing yet, but I plan to use a sinusoidal relation which would underestimate the inclination at most angles as described in one of the papers.

My approximation is:

Vir = (Voffset-Vmax)*sin(x) + Voffset

The curve is parametrized by Voffset (a function of the hardware, I think 1/2 of the source voltage on FMA heads) and Vmax which should be calibrated. I also attached a figure from my Master's report.

This is all untested, mind you. I would be interested to see how this shapes up against the Paparazzi model. Anybody know that?

I would suggest that you look at the paparazzi source code, I was able to tease out the equations that they use to estimate attitude from the IR sensor information. I don't remember which file you need to look at off the top of my head (estimator.h maybe?). I have the file with the equations on my other machine, but if you pm me with your email address, I can send you that information (it may take a few days since I will be traveling all week).

BTW, this seems more a forum topic than blog topic.

## Replies

The people from Monash University seem to have used the IR sensors for quite a while. One of the ways, as described in that paper is that you assume a sinusoidal profile and use the peak to peak values voltage you get on the field before flying to calibrate the sensors. Then use the vertical sensors to constantly monitor the temperature range and update the peak to peak values and re-calibrate them again. I will have a look at the paparazzi equations.

"Use of Infrared Sensors for Absolute Attitude Determination for UAVs"

From my reading, it seems the basic relationship (assuming a wide FOV for the thermopiles) is a "chopped" sinusoid. I haven't done any testing yet, but I plan to use a sinusoidal relation which would underestimate the inclination at most angles as described in one of the papers.

My approximation is:

Vir = (Voffset-Vmax)*sin(x) + Voffset

The curve is parametrized by Voffset (a function of the hardware, I think 1/2 of the source voltage on FMA heads) and Vmax which should be calibrated. I also attached a figure from my Master's report.

This is all untested, mind you. I would be interested to see how this shapes up against the Paparazzi model. Anybody know that?

UseOfIRsensorsForAbsAttDet.pdf

VImodelPPT.png

I would suggest that you look at the paparazzi source code, I was able to tease out the equations that they use to estimate attitude from the IR sensor information. I don't remember which file you need to look at off the top of my head (estimator.h maybe?). I have the file with the equations on my other machine, but if you pm me with your email address, I can send you that information (it may take a few days since I will be traveling all week).

BTW, this seems more a forum topic than blog topic.

Cory