On the Use of a Maximum Correntropy Criterion in Kalman Filtering Based Strategies for Robot Localization and Mapping

One of the applications of the Kalman filter in the field of robotics is to solve the problem of Simultaneous Localization and Mapping (SLAM). The main drawback of the Kalman filter is that its performance can degrade in the presence of non-Gaussian measurement noise. In robotic systems using laser range finders such as the LiDAR, often optical properties of the beam-environment interaction introduce non-Gaussian noise into the system, which can significantly affect performance. In this paper, we investigate this problem and propose a SLAM algorithm similar to the Extended Kalman filter but based on the Maximum Correntropy Criteria (MCC), which aims to exhibit better performance than the classical Extended Kalman filter for some types of non-Gaussian noises. The performance of the proposed MCC-EKF SLAM and the classical EKF SLAM are compared by means of numerical simulations.