Artificial horizon using MEMS gyroscope?

Question

I have an application akin to an aircraft "Artificial horizon", and am wondering whether I can use a MEMS gyroscope. By integrating the output I should have a measure of angle to the vertical, but I understand that the integral would drift off over a relatively short period. Measuring the gravitational direction is obviously not possible, as a correctly banked aircraft always sees the combined gravitation and centrifugal force acting downwards with respect to the aircraft.

Has anyone any ideas how to correct or at least measure the drift?

DAB · Answer

Hi Peter,

To maintain an artificial horizon, you have to keep track of the position and atitude changes of the vehicle.

The gyroscope will help give you changes in yaw, pitch, and roll, but you then need to project those changes onto your current position and attitude to put them into proper perspective.

You might need to implement a basic inertial measurement unit to establish your baseline coordinates. Then you need to update that virtual position as the device moves.

I know that some people use a Kalman Filter approach to those inputs to reduce the amount of noise, but I am not sure that it would help your current problem.

DAB

vsluiter · Answer

Hi Peter, As mentioned by DAB, the most used solution for this is Kalman Filtering, using both an accelerometer and a gyroscope (IMU), or even gyr/acc/magnetometer (often called IMU, but in fact an AltitudeHeadingReferenceSystem). These filters use data when there reliability is best. You'll be able to find a lot of info on Kalman filters on internet. The amount of drift you can expect is specified in most datasheets, they'll give a 'zero angular speed offset'. Since the angular speed is integrated to get the angular position, you can integrate this value over time to get a worst case estimation. Also be aware that if you get clipping (hard impact with environment), your integration value will get way off....