抄録
In this paper, we propose a sampling technique with Unscented Transformation for FastSLAM algorithm. Introducing the Unscented Transformation into the particle sampling, we make that the sampling accuracy becomes more accurate and stable than using the formula of Extended Kalman filter like the FastSLAM2.0. Besides, unlike the Extended Kalman filter, the Unscented Transformation eliminates the cumbersome derivation and evaluation of Jacobian matrices. Experimental results of SLAM in the dynamic outdoor environment are presented, we show that the accuracy of the SLAM with the Unscented Transformation is more accurate than the FastSLAM2.0.
