Abstract: The traditional unscented Kalman filer is lack of on-line adaptive adjustment ability，and probably decreases the filtering accuracy under the influence of erroneous noisy model. An improved simultaneous localization and mapping (SLAM) algorithm based on robust unscented Kalman filer is proposed. An multidimensional measurement noise scale factor is introduced into the proposed algorithm to adaptively adjust each sensor’s noisy model according to the real changing condition of noisy statistic characteristics, and then the filter gain is rectified to an appropriate value, thus improving the estimation accuracy of the filter. Simulations are performed by using different SLAM algorithms with the time-varying noisy statistics, results show that the proposed algorithm is of better adaptability and estimation accuracy compared with other SLAM algorithms, and its robustness is also improved.

Key words: Simultaneous localization and mapping (SLAM), Unscented Kalman filter, Scale factor, Filter Gain, Innovation