MapCleaner: Efficiently Removing Moving Objects from Point Cloud Maps in Autonomous Driving Scenarios

• DOI: 10.3390/rs14184496
• Publication Date: 2022-09-09T08:54:41Z
• Authors: Hao Fu; Hanzhang Xue; Guanglei Xie
• Keywords: map cleaning; 03 medical and health sciences; 0302 clinical medicine; autonomous driving; Science; Q; LiDAR point cloud; dynamic object

Three-dimensional (3D) point cloud maps are widely used in autonomous driving scenarios. These usually generated by accumulating sequential LiDAR scans. When generating a map, moving objects (such as vehicles or pedestrians) will leave long trails on the assembled map. This is undesirable and reduces map quality. In this paper, we propose MapCleaner, an approach that can effectively remove from MapCleaner first estimates dense continuous terrain surface, based which then divided into noisy part below terrain, object above terrain. Next, specifically designed points identification algorithm performed to find objects. Experiments SemanticKITTI dataset. Results show proposed outperforms state-of-the-art approaches all five tested sequences. learning-free method has few parameters tune. It also successfully evaluated our own dataset collected with different type of LiDAR.