Towards Image-Grade LiDAR Odometry and Mapping

The recent developments in LiDAR technologies has promoted several super high intensity image-grade LiDAR sensors, receiving great interest in the intelligent transportation applications. Although objects detection and high-resolution map construction can benefit from this characteristic, there are several underlying challenges arising from such LiDARs. In this contribution, we present a robust and accurate odometry and mapping scheme for image-grade LiDARs, which addresses many key issues including outliers rejection, moving vehicles filtering, feature extraction and selection. Results from extensive experiments performed in various environments, including feature-poor tunnels and busy urban roads, demonstrate the high accuracy and robustness of the proposed method for state estimation in the real applications.