Improved SLAM algorithm based on compressed-EKF

The compressed extended Kalman filter (CEKF)-based algorithm for simultaneous localization and mapping (SLAM) has low efficiency on state augment and map management. An improved algorithm (ICEKF) was proposed. The ICEKF algorithm achieves the state augment by only augmenting one auxiliary coefficient matrix, and the computational complexity is reduced from O(N²) to O(N_A), where N and N_A denote the number of the landmarks in the global and local maps respectively. A Euclidian distance-based method for map management was presented which selects the local map dynamically. In this way, the landmarks assignment and other problems issued from dividing the global map before the SLAM starts as in the CEKF algorithm were avoided. Simulations show that the ICEKF algorithm obtaines the same optimal results as the EKF algorithm, and its computational cost is greatly reduced in comparion with the CEKF algorithm.