Localization System Through 2D LiDAR based Semantic Feature For Indoor Robot

In this paper, we propose a semantic feature extraction based on the light detection and ranging (LiDAR) sensor of an indoor driving robot and a location recognition method using the extracted features. After extracting semantic features based on the corner position and direction and shape of the corner for a wall or door in an indoor driving environment, and matching it with the corner information of the map, position recognition is performed using the collinearity method. It shows excellent performance with low computational complexity in embedded computers. We tested the proposed method in a real indoor environment using real robots and sensors. The performance of the location recognition system was verified by comparison with the widely used AMCL (Adaptive Monte Carlo Localization) algorithm.