I am using a extended Kalman filter for the state estimation of a nonlinear system. The real system has accelerometers, so I need to include the acceleration of the system as part of my measurements with noise added (position and angular rates are other measurements). All of the Kalman filter design I've done in the past only had measurements based on state variables. Obviously the acceleration is not a state variable, so I am unsure about how to get my simulated acceleration measurement. I can get acceleration since I know what my inputs are, but all of my EKF references don't have inputs as part of the measurement function: That's the only way I can think of getting my acceleration measurements...what am I missing?