LiDAR Ground Segmentation and Modeling for Mobile Robots in Unstructured Terrain

Abstract

We propose a new method for fine registering multiple point clouds simultaneously. The approach is characterized by being dense, therefore point clouds are not reduced to pre-selected features in advance. Furthermore, the approach is robust against small overlaps and dynamic objects, since no direct correspondences are assumed between point clouds. Instead, all points are merged into a global point cloud, whose scattering is then iteratively reduced. This is achieved by dividing the global point cloud into uniform grid cells whose contents are subsequently modeled by normal distributions. We show that the proposed approach can be used in a sliding window continuous trajectory optimization combined with IMU measurements to obtain a highly accurate and robust LiDAR inertial odometry estimation. Furthermore, we show that the proposed approach is also suitable for large scale keyframe optimization to increase accuracy. We provide the source code and some experimental data on https://github.com/davidskdds/DMSA_LiDAR_SLAM.git.

SLAM for Indoor Mapping of Wide Area Construction Environments

“…Figure 5b) provides a relatively sparse but precise representation of the environment. With DMSA SLAM (Skuddis and Haala, 2024) we were able to successfully process the data from the LiDAR sensor together with the IMU data after a few adaptations to the initially published pipeline/parameters. The alignment of the environment representation created on the basis of Visual SLAM showed that the resulting map was down-scaled by approximately 5% compared to the reference.…”

“…Newer approaches also include mechanisms such as Scan Context (Kim and Kim, 2018) or LoGG3D-Net (Vidanapathirana et al, 2022) to detect the revisiting of already mapped locations (loop closures) and to improve consistency through pose graph optimization (Behley and Stachniss, 2018), (Ramezani et al, 2022). In some modern methods, the points are still reduced to edge and planar features during preprocessing (Shan et al, 2020), while in others the points are combined into disk-like surface elements (Behley and Stachniss, 2018), (Ramezani et al, 2022) or in so-called dense approaches simply (downsampled) points are used (Dellenbach et al, 2022), (Xu et al, 2022), (Vizzo et al, 2023), (Skuddis and Haala, 2024). In modern LiDAR SLAM systems IMU data are used to increase robustness in situations with fast movements and to reduce the orientational drift through gravity estimations (Ramezani et al, 2022), (Shan et al, 2020), (Xu et al, 2022), (Skuddis and Haala, 2024).…”

“…In some modern methods, the points are still reduced to edge and planar features during preprocessing (Shan et al, 2020), while in others the points are combined into disk-like surface elements (Behley and Stachniss, 2018), (Ramezani et al, 2022) or in so-called dense approaches simply (downsampled) points are used (Dellenbach et al, 2022), (Xu et al, 2022), (Vizzo et al, 2023), (Skuddis and Haala, 2024). In modern LiDAR SLAM systems IMU data are used to increase robustness in situations with fast movements and to reduce the orientational drift through gravity estimations (Ramezani et al, 2022), (Shan et al, 2020), (Xu et al, 2022), (Skuddis and Haala, 2024). While in IMU-free LIDAR SLAM approaches it is common to model the trajectory within one scan as linear motion (Dellenbach et al, 2022), (Neuhaus et al, 2019), (Vizzo et al, 2023), IMU supported LiDAR inertial SLAM approaches enable higher resolution trajectory representations, which can result in improved accuracy , (Shan et al, 2020), (Xu et al, 2022), (Skuddis and Haala, 2024).…”

“…In modern LiDAR SLAM systems IMU data are used to increase robustness in situations with fast movements and to reduce the orientational drift through gravity estimations (Ramezani et al, 2022), (Shan et al, 2020), (Xu et al, 2022), (Skuddis and Haala, 2024). While in IMU-free LIDAR SLAM approaches it is common to model the trajectory within one scan as linear motion (Dellenbach et al, 2022), (Neuhaus et al, 2019), (Vizzo et al, 2023), IMU supported LiDAR inertial SLAM approaches enable higher resolution trajectory representations, which can result in improved accuracy , (Shan et al, 2020), (Xu et al, 2022), (Skuddis and Haala, 2024). In recent works, for LiDAR SLAM Neural Network representations of the environment have been proposed to obtain more consistent maps (Wiesmann et al, 2023), (Zhong et al, 2023), (Deng et al, 2023).…”

“…Figure 3 gives an overview of the processing steps. For details we refer to (Skuddis and Haala, 2024).…”