Integration of accelerometer data: Filter analysis and design using Riccati solutions

The adjustment and integration of noisy accelerometer data is formulated as a discrete-time optimal control problem which requires the minimization of a quadratic performance measure subject to linear dynamics. The algebraic Riccati equation associated with this problem is solved by expressing the entries of the Riccati matrix explicitly in terms of the weights in the quadratic performance criterion. Steady-state feedback gains are thus easily computed as functions of the weights. The frequency domain analysis of the adjustment procedure provides the basis for selecting appropriate weights.