An SLAM Algorithm Based on Square-root Cubature Particle Filter

The lack of the latest measurement information and the Particle serious degradation cause low estimation precision in the tradition particle filter SLAM (simultaneous localization and mapping). For solve this problem, a SRCPF-SLAM (square cubature particle filter simultaneous localization and mapping) is proposed in this paper. The algorithm fuses the latest measurement information in the stage of the prior distribution updated of the particle filter SLAM. It designs importance density function by SRCKF (Squareroot Cubature kalman filter) that is more close to the posterior density, and it spreads the square root of state covariance. So, the algorithm ensures the symmetry and the positive semi-definiteness of the covariance matrix and improves numerical estimation precision and stability. The simulation results show that the proposed algorithm has higher accuracy of the state estimation when compared with the the PFSLAM (particle filter simultaneous localization and mapping) algorithm, EPF-SLAM (extend particle filter simultaneous localization and mapping) algorithm and the UPF-SLAM (unscented particle filter simultaneous localization and mapping) algorithm.