The Larry Project

Here it is: I won’t integrate the accelerometers at all. If I even use them, they will just be used to measure instantaneous acceleration for direct use, not as part of an INS. The camera platform will be leveled with the FMA sensor. If I can get away with it, I will even use the FMA controller to level the platform. Then I can measure the platform angles relative to the airplane to get the pitch and bank angles. The rate gyros will then be used as inputs to the flight controller. I think I can get away with using my 1 Hz GPS, but I think I will also need either a pitot-static system or an angle of attack sensor. Taking GPS track angles and speeds for two different steady headings, I can calculate the wind speed and direction, assuming the airspeed is known. If I do this for three different headings, I can find wind speed and direction, and airspeed. This assumes the airspeed is a constant. This comes from the following equation:
a^2=g*w-2*g*w*cos(wind_direction - GPS_direction)
where
a = airspeed
g = GPS speed
w = wind speed
and 2 = 2

Today I talked to my advisor. He says some kind of feedback integration of the inertial readings should work. The GPS will provide feedback to the accelerometer-derived position and velocity. The horizon sensors will give feedback for the attitude determination integrators. Right now, my main concern is the 1 second lag in the GPS. I am looking at ways to compensate for this. Right now I’m working on a simulink model for the GPS/accelerometer system. The errors are bounded (using a constant-biased accelerometer siginal of various forms) by the GPS feedback to the velocity and position integrators, but the lag in the resultant position and velocity estimates is still about 1 second. I did try using a lagged version of the integrator to use with the lagged GPS values, but that system was unstable for useful gain settings.

I’m done working on this until finals are over.