LIDAR-based SLAM implementation using Kalman filter

Abstract

The capability to navigate accurately is one of the features, that a mobile robot should have to be able to perform tasks autonomously. In a GPS/GNSS-denied environment, for example inside buildings, localization of a mobile platform is an especially challenging problem. In such cases, to provide a robot with the ability to determine its position and to analyze its surroundings, Simultaneous Localization and Mapping (SLAM) algorithms could be implemented. In the article, we present a SLAM system that uses a Kalman filter together with data gathered by a 2D LiDAR. Our approach applies the ICP algorithm to calculate the localization and employs clustering and shape recognition technics to build the map of the environment. The article contains a detailed description of the individual elements of the proposed SLAM solution. Furthermore, it presents the results of experiments during which the system was validated.