Machine learning and IMU sensors?

Question

Machine learning

Is it possible to use the IMU readings of gyro sensors and machine learning to replace the PID and stabilize the aircraft model in space, because the readings are quite noisy even after filtering with the Majevic filter (+/- 5 deg) and also train the network to perform certain actions, for example, turn on the course in the shortest possible time and how best to do this?

Answer 1

I think you don't need a network, but you need an adequate mathematical flight model.
Based on it, you can generate a bunch of examples of entry-exit, etc.
train the network to behave.
Well, or use millions of copies of an aircraft model to break 99.9999999999999% of them.