I recently saw a matlab sensor fusion example which combined IMU and GPS data for drone localization.
What was surprising about it was that the IMU was used to predict future behavior of the system as if it was a control signal. Basically the IMU reading was read and then integrated over a time step to create a prediction for the next location of the drone which was then fused with the GPS reading.
This is confusing to me as in principle both the IMU and the GPS are sensors and not control inputs.
Could someone explain in more detail when you are supposed to consider something a control input to a system vs a sensor input? I always assumed control signals were things like motor velocity control which evolved the system state using linearized dynamic equations but this matlab example has kind of blurred the line for me.
In principle could any sensor reading be viewed as a control input if you can evolve the system state from it?
