A Fusion Localization Algorithm Combining MCL with EKF

The Monte Carlo algorithm uses a random and weighted sampling set to represent and estimate the possible and position distribution of the mobile robot. To improve the accuracy of localization, a new localization algorithm combining the original MCL (Monte Carlo Localization) with EKF (Extended Kalman Filter) is proposed in this paper. First, according to the initial set, the needed particles are collected in the space and the mean value of particles are calculated. Second, the best global features LG are extracted from the sensors' measurements. Finally, EKF is used to update the current state and covariance of the robot and exclude the useless particles. Simulations and experiments proved that the proposed algorithm is superior, for the localization particles distribute tightly around the moving robot with lower location error.