A Monocular LiDAR Fusion SLAM for Indoor Environment

The research on SLAM (simultaneous localization and mapping) has greatly progressed in recent years. However, research gaps remain because of the nature of different sensors in SLAM applications. This paper proposes a fusion SLAM method for the robot platform in the indoor environment, equipped with a LiDAR (light detecting and ranging) and a monocular camera. The proposed method extracts point, line, and plane features from monocular images and LiDAR scans and uses points and line features for pose estimation and optimization. The experiment result on the recently released challenge dataset showed that the proposed method would recover the scale for monocular SLAM by 32-scan LiDAR with almost no accuracy loss and that it was more robust than state-of-the-art algorithms.