Real-Time Optimal State Estimation Scheme With Delayed and Periodic Measurements

This paper proposes an efficient real-time optimal estimation scheme that uses accurate but delayed measurements obtained periodically from high-performance sensing devices. For real-time optimal estimation, we employ two Kalman filters: one to conventionally estimate the current state and the other to precompute for the future state estimation to be carried out, when a new, accurate but delayed measurement arrives. The precomputing Kalman filter does the necessary computation in advance, for the future state estimation, from the available measurements for distributing the computational burden over time, thereby obtaining an optimal estimate in real time. By optimally incorporating accurate but delayed measurements, the optimality is preserved at all times, without imposing a heavy computational burden in a short sampling time interval. It is demonstrated through experiments that the proposed scheme can significantly improve the estimation performance with the least detriment to the real-time computation and memory size, when delayed and periodic measurements are available.