Abstract: Backward-smoothing square-root cubature Kalman filter(BS-SRCKF) is used to caculate observation data of GPS dynamic single point positioning,and the gross errors' bad influence on backward-smoothing filter is analyzed in this article. In order to reduce gross errors' impacts, robust SRCKF is put forward referring to robust Kalman filter. And when coarse errors don't exist,backward-smoothing filter can be used to improve the precision of the results and avoid iterative computation if using robust filter each epoch.Measured data is adopted to prove that the algorithm is effective.

Key words: nonlinear filtering, SRCKF, robust filter, backward smoothing