Adaptive Lattice Kalman Filter-SLAM for Robot Auto-navigation

This paper introduces the adaptive lattice Kalman filter (ALKF) into the SLAM problem to reduce the computational cost and improve the filtering stability. The lattice Kalman filter (LKF) algorithm based on the lattice rules has lower computational burden to maintain the state estimation accuracy by using fewer sampling points compared with other deterministic sampling method. The measurement noise is adjusted based on the measurement residual, and then the state variance is modified by using the fading factor under the modified measurement variance based on the lattice sampling points. Comparing with standard extended Kalman filter-SLAM, LKF-SLAM, the accuracy of the proposed ALKF-SLAM is valid and feasible in the simulations under the uncertain system noises.