Feature Extraction and Matching Algorithms to Improve Localization Accuracy for Mobile Robots

Localization of indoor mobile robots is an important field in Simultaneous Localization and Mapping (SLAM). SLAM is a technology that generates map and estimates the current locations as robot explore random space. So, that is commonly used in indoor environments where GPS is not working. We propose the method of feature extraction and feature matching for localization. Features are represented wall and corner in line and point. We extract lines and corner points with observed data by 2D lidar sensor and match extracted features with a stored feature in the map. Finally, we show to increase the accuracy of localization by calculating differences between coordinates of matched features.