Square Root Unscented Particle Filter with Application to Angle-Only Tracking

This paper develops a novel filter for nonlinear angle-only tracking issues. This new filter is implemented according to the procedure adopted in generic particle filter (PF). But unlike the generic PF that simply uses the prior as the importance density function, this new filter uses a recently developed square root unscented Kalman filter (SRUKF) to generate the importance density. That is why the new filter is called the square root unscented particle filter (SRUPF). In SRUPF, the importance density features in embodying the latest observation information, so it can well approximate the distribution of the state variable. Moreover, using SRUKF to calculate the importance density avoids factorizing the state covariance at each time step. This makes the new filter have a good numerical stability. A simulation-based angle-only tracking example is used to demonstrate the effectiveness of the proposed algorithm.