Lidar Odometry Research Articles

Localization and Mapping for UGV in Dynamic Scenes with Dynamic Objects Eliminated

SLAM (Simultaneous Localization and Mapping) based on lidar is an important method for UGV (Unmanned Ground Vehicle) localization in real time under GNSS (Global Navigation Satellite System)-denied situations. However, dynamic objects in real-world scenarios affect odometry in SLAM and reduce localization accuracy. We propose a novel lidar SLAM algorithm based on LOAM (Lidar Odometry and Mapping), which is popular in this field. First, we applied elevation maps to label the ground point cloud. Then we extracted convex hulls in point clouds based on scanlines as materials for dynamic object clustering. We replaced these dynamic objects with background point cloud to avoid accuracy reduction. Finally, we extracted feature points from ground points and non-ground points, respectively, and matched these feature points frame-to-frame to estimate ground robot motion. We evaluated the proposed algorithm in dynamic industrial park roads, and it kept UGV maximum relative position error less than 3% and average relative position error less than 2%.

Point Wise or Feature Wise? A Benchmark Comparison of Publicly Available Lidar Odometry Algorithms in Urban Canyons

Robust and precise localization is essential for an autonomous system with navigation requirements. Lidar odometry (LO) has been extensively studied in the past decades to realize this goal. Satisfactory accuracy can be achieved in scenarios with abundant environmental features using existing LO algorithms. Unfortunately, the performance of the LO is significantly degraded in urban canyons with numerous dynamic objects and complex environmental structures. Meanwhile, it is still not clear from the existing literature which LO algorithms perform well in such challenging environments. To fill this gap, this article evaluates an array of popular and extensively studied LO pipelines using the data sets collected in urban canyons of Hong Kong. We present the results in terms of their positioning accuracy and computational efficiency. The three major factors dominating the performance of LO in urban canyons are concluded, including the ego-vehicle dynamic, moving objects, and the degree of urbanization. According to our experiment results, point wise accomplishes better accuracy in urban canyons while feature-wise achieves cost-efficiency and satisfactory positioning accuracy.

LiDAR Odometry and Mapping Based on Neighborhood Information Constraints for Rugged Terrain

The simultaneous localization and mapping (SLAM) method estimates vehicles’ pose and builds maps established on the collection of environmental information primarily through sensors such as LiDAR and cameras. Compared to the camera-based SLAM, the LiDAR-based SLAM is more geared to complicated environments and is not susceptible to weather and illumination, which has increasingly become a hot topic in autonomous driving. However, there has been relatively little research on the LiDAR-based SLAM algorithm in rugged scenes. The following two issues remain unsolved: on the one hand, the small overlap area of two adjacent point clouds results in insufficient valuable features that can be extracted; on the other hand, the conventional feature matching method does not take point cloud pitching into account, which frequently results in matching failure. Hence, a LiDAR SLAM algorithm based on neighborhood information constraints (LoNiC) for rugged terrain is proposed in this study. Firstly, we obtain the feature points with surface information using the distribution of the normal vector angles in the neighborhood and extract features with discrimination through the local surface information of the point cloud, to improve the describing ability of feature points in rugged scenes. Secondly, we provide a multi-scale constraint description based on point cloud curvature, normal vector angle, and Euclidean distance to enhance the algorithm’s discrimination of the differences between feature points and prevent mis-registration. Subsequently, in order to lessen the impact of the initial pose value on the precision of point cloud registration, we introduce the dynamic iteration factor to the registration process and modify the corresponding relationship of the matching point pairs by adjusting the distance and angle thresholds. Finally, the verification based on the KITTI and JLU campus datasets verifies that the proposed algorithm significantly improves the accuracy of mapping. Specifically in rugged scenes, the mean relative translation error is 0.0173%, and the mean relative rotation error is 2.8744°/m, reaching the current level of the state of the art (SOTA) method.

A Hierarchical LiDAR Odometry via Maximum Likelihood Estimation With Tightly Associated Distributions

LiDAR odometry has gained popularity due to accurate depth measurement with the robustness to illuminations. However, existing distribution-based methods do not sufficiently exploit the information from source point cloud, which affects the odometry performance. In this paper, a novel distribution-to-distribution matching method is proposed based on maximum likelihood estimation to solve relative transformation, where source and target point sets are tightly jointed to represent the sampling distribution in the objective function. On this basis, a hierarchical 3D LiDAR odometry with the low-level scan-to-map matching and high-level fixed-lag smoothing is designed. With the decoupling strategy, the matching method is extended to a fixed-lag smoothing module and the heavy computation burden is overcome. Our smoothing module is universal, which can be attached to LiDAR odometry framework for performance improvement. The experiments on KITTI dataset, Newer College dataset, and large-scale KITTI-360 dataset verify the effectiveness of the proposed method.

LOCUS 2.0: Robust and Computationally Efficient Lidar Odometry for Real-Time 3D Mapping

Lidar odometry has attracted considerable attention as a robust localization method for autonomous robots operating in complex GNSS-denied environments. However, achieving reliable and efficient performance on heterogeneous platforms in large-scale environments remains an open challenge due to the limitations of onboard computation and memory resources needed for autonomous operation. In this work, we present LOCUS 2.0, a robust and computationally-efficient \lidar odometry system for real-time underground 3D mapping. LOCUS 2.0 includes a novel normals-based \morrell{Generalized Iterative Closest Point (GICP)} formulation that reduces the computation time of point cloud alignment, an adaptive voxel grid filter that maintains the desired computation load regardless of the environment's geometry, and a sliding-window map approach that bounds the memory consumption. The proposed approach is shown to be suitable to be deployed on heterogeneous robotic platforms involved in large-scale explorations under severe computation and memory constraints. We demonstrate LOCUS 2.0, a key element of the CoSTAR team's entry in the DARPA Subterranean Challenge, across various underground scenarios. We release LOCUS 2.0 as an open-source library and also release a \lidar-based odometry dataset in challenging and large-scale underground environments. The dataset features legged and wheeled platforms in multiple environments including fog, dust, darkness, and geometrically degenerate surroundings with a total of $11~h$ of operations and $16~km$ of distance traveled.

Map-Free Lidar Odometry (MFLO) Using a Range Flow Constraint and Point Patch Covariances

We present a lightweight real-time method to extract 3D ego-motion using a range flow constraint equation, point patch covariance, and a least squares solution. Our method exploits the structured data provided by range sensors, like rotating LiDARs, to attain 6 DOF odometry without building a map or scan-matching. To evaluate the performance of MFLO, a quadrotor was flown in various environments, and results indicate that MFLO matches and sometimes exceeds the performance of other LiDAR-based odometry methods while using fewer computational resources. In underground environments, MFLO captured 95.7% of the total vertical displacement for a 17.5 m translation upwards through a missile silo, the most of any other LiDAR algorithm evaluated in this study, and captured 92.8% of the total translation for a 42 m translation through an underground mine. In a motion capture lab, MFLO performed with only a 0.89–2.87% displacement percent error and 1.03–2.97% in final position error comparing to ground truth, making it the most consistent LiDAR odometry algorithm without mapping.

Generalized LOAM: LiDAR Odometry Estimation With Trainable Local Geometric Features

This letter presents a LiDAR odometry estimation framework called Generalized LOAM. Our proposed method is generalized in that it can seamlessly fuse various local geometric shapes around points to improve the position estimation accuracy compared to the conventional LiDAR odometry and mapping (LOAM) method. To utilize continuous geometric features for LiDAR odometry estimation, we incorporate tiny neural networks into a generalized iterative closest point (GICP) algorithm. These neural networks improve the data association metric and the matching cost function using local geometric features. Experiments with the KITTI benchmark demonstrate that our proposed method reduces relative trajectory errors compared to the other LiDAR odometry estimation methods.

FEVO-LOAM: Feature Extraction and Vertical Optimized Lidar Odometry and Mapping

Simultaneous Localization and Mapping (SLAM) is a significant research topic in robotics since it is one of the key technologies for robot automation. Although lidar-based SLAM methods have achieved promising performance, traditional lidar SLAM methods still produce large vertical errors. To address this issue, we propose a feature extraction and vertical optimized lidar odometry and mapping approach. Firstly, we optimize the feature extraction. Specifically, we propose a more accurate ground segmentation approach and a new curvature definition, which is used to extract more discriminative features. Additionally, we propose a lidar mapping approach, which adds new vertical residuals and pitch residuals to the objective function. Then a two-step Levenberg-Marquardt method is used to solve the pose transformation. Finally, we evaluate the proposed method in public datasets and real environments. Experiments show that compared with other state-of-the-art methods, our method achieves better accuracy and reduces the vertical error with a similar computational expense.

Efficient 3D Lidar Odometry Based on Planar Patches.

In this paper we present a new way to compute the odometry of a 3D lidar in real-time. Due to the significant relation between these sensors and the rapidly increasing sector of autonomous vehicles, 3D lidars have improved in recent years, with modern models producing data in the form of range images. We take advantage of this ordered format to efficiently estimate the trajectory of the sensor as it moves in 3D space. The proposed method creates and leverages a flatness image in order to exploit the information found in flat surfaces of the scene. This allows for an efficient selection of planar patches from a first range image. Then, from a second image, keypoints related to said patches are extracted. This way, our proposal computes the ego-motion by imposing a coplanarity constraint between pairs <point, plane> whose correspondences are iteratively updated. The proposed algorithm is tested and compared with state-of-the-art ICP algorithms. Experiments show that our proposal, running on a single thread, can run 5× faster than a multi-threaded implementation of GICP, while providing a more accurate localization. A second version of the algorithm is also presented, which reduces the drift even further while needing less than half of the computation time of GICP. Both configurations of the algorithm run at frame rates common for most 3D lidars, 10 and 20 Hz on a standard CPU.

Efficient and Accurate Tightly-Coupled Visual-Lidar SLAM

We investigate a novel way to integrate visual SLAM and lidar SLAM. Instead of enhancing visual odometry via lidar depths or using visual odometry as the motion initial guess of lidar odometry, we propose tightly-coupled visual-lidar SLAM (TVL-SLAM), in which the visual and lidar frontend are run independently and which incorporates all of the visual and lidar measurements in the backend optimizations. To achieve large-scale bundle adjustments in TVL-SLAM, we focus on accurate and efficient lidar residual compression. The visual-lidar SLAM system implemented in this work is based on the open-source ORB-SLAM2 and a lidar SLAM method with average performance, whereas the resulting visual-lidar SLAM clearly outperforms existing visual/lidar SLAM approaches, achieving 0.52% error on KITTI training sequences and 0.56% error on testing sequences.

LIDAR-Inertial Real-Time State Estimator with Rod-Shaped and Planar Feature

State estimation and mapping based on Light Detection and Ranging (LIDAR) are important for autonomous systems. Point cloud registration is a crucial module affecting the accuracy and real-time performance of LIDAR simultaneous localization and mapping (SLAM). In this paper, a novel point cloud feature selection for LIDAR-inertial tightly coupled systems is proposed. In the front-end, a point cloud registration is carried out after marking rod-shaped and planar feature information which is different from the existing LIDAR and inertial measurement unit (IMU) integration scheme. This preprocessing method subsequently reduces the outliers. IMU pre-integration outputs high-frequency result and is used to provide the initial value for LIDAR solution. In the scan-to-map module, a computationally efficient graph optimization framework is applied. Moreover, the LIDAR odometry further constrains the IMU states. In the back-end, the optimization based on sliding-window incorporates the LIDAR-inertial measurement and loop closure global constraints to reduce the cumulative error. Combining the front-end and back-end, we propose the low drift and high real-time LIDAR-inertial positioning system. Furthermore, we conducted an exhaustive comparison in open data sequences and real-word experiments. The proposed system outperforms much higher positioning accuracy than the state-of-the-art methods in various scenarios. Compared with the LIO-SAM, the absolute trajectory error (ATE) average RMSE (Root Mean Square Error) in this study increases by 64.45% in M2DGR street dataset (street_01, 04, 07, 10) and 24.85% in our actual scene datasets. In the most time-consuming mapping module of each system, our system runtime can also be significantly reduced due to the front-end preprocessing and back-end graph model.

GR-LO: A specific lidar odometry system optimized with ground and road edges

For the specific environment, we propose a specific lidar odometry system optimized with ground and road edges, GR-LO, for unmanned ground vehicle (UGV) navigation in the Global Positioning System (GPS) coordinate system. Many high-performance lidar odometry (LO) and lidar-based simultaneous localization and mapping (SLAM) systems have been proposed. However, when we need to convert the pose of lidar from the lidar coordinate to the GPS coordinate system for navigation, there is usually a nonnegligible error. One of the reasons for the error is that the transformation between the two coordinate systems is not calibrated correctly. Our method is proposed to reduce this error by extracting prior surrounding features in the specific environment. We have tested the proposed method with our datasets. The results show that by utilizing the prior features, the proposed method can reduce the absolute pose errors by more than 50% compared to the adopted lidar odometry method.

A 3D LiDAR odometry for UGVs using coarse-to-fine deep scene flow estimation

Light detection and ranging (LiDAR) odometry plays a crucial role in autonomous mobile robots and unmanned ground vehicles (UGVs). This paper presents a deep learning–based odometry system using two successive three-dimensional (3D) point clouds to estimate their scene flow and then predict their relative pose. The network consumes continuous 3D point clouds directly and outputs their scene flow and uncertain mask in a coarse-to-fine fashion. A pose estimation layer without trainable parameters is designed to compute the pose with the scene flow. We also introduce a scan-to-map optimization algorithm to enhance the robustness and accuracy of the system. Our experiments on the KITTI odometry data set and our campus data set demonstrate the effectiveness of the proposed deep learning–based point cloud odometry.

Simultaneous Localization of Rail Vehicles and Mapping of Surroundings With LiDAR-Inertial-GNSS Integration

Accurate rail vehicle positioning is crucial for railroad operational safety. Modern light detection and ranging (LiDAR) simultaneously localization and mapping (SLAM) systems have delivered excellent results in real-world scenarios. However, it still lacks well investigation for rail vehicle applications. In this paper, we propose to achieve real-time accurate and robust positioning and mapping for rail vehicles utilizing LiDAR SLAM. Our framework tightly couples one non-repetitive scanning LiDAR with IMU, wheel odometer, and global navigation satellite system (GNSS) into pose estimation and simultaneous global map generation. As frontend, the IMU/odometer preintegration data de-skews the denoised point clouds and produces initial guess for LiDAR odometry. Besides, we leverage the plane constraints from extracted rail tracks and the height descriptor to further improve the system accuracy. As backend, a sliding window based factor graph is constructed to jointly optimize multi-modal information. To ensure a globally-consistent and less blurry mapping result, we develop a two-stage mapping method to register the submaps to the global. The proposed method is extensively evaluated on real-world datasets of numerous scenarios, including general-speed and high-speed ones, both freight traffic and passenger traffic is covered. The results show that our system delivers meter-level localization accuracy even in large or degenerated environments.

A Tight Filtering and Smoothing Fusion Method With Feature Tracking for LiDAR Odometry

LiDAR odometry (LO) has gained popularity in recent years because of its growing applications. Existing solutions include filtering and smoothing, where the former optimizes the latest pose with high efficiency and the latter optimizes multiple poses with strong consistency. Still, it is promising to fuse filtering and smoothing in LO. In this paper, we propose a novel tight filtering and smoothing fusion method. Firstly, two nested sliding windows are maintained and they share estimated poses and features to balance the accuracy and efficiency. The large outer window is used to build a large local map for high-quality scan-to-map matching in the filtering module, and multiple poses in the small inner window are optimized in the smoothing module to improve local consistency and efficiency. Furthermore, sparse directed geometry points (DGPs) are tracked across window keyframes to produce feature-wise multiple residuals for poses optimization in both filtering and smoothing modules. Besides, the strategies of ratio-based refined pose optimization and efficient DGP feature triangulation are developed for pose refinement and local map update, respectively. The experiments on KITTI and Newer College (NC) datasets demonstrate the effectiveness of the proposed method.

RO-LOAM: 3D Reference Object-based Trajectory and Map Optimization in LiDAR Odometry and Mapping

We propose an extension to the LiDAR Odometry and Mapping framework (LOAM) that enables reference object-based trajectory and map optimization. Our approach assumes that the location and geometry of a large reference object are known, e.g., as a CAD model from Building Information Modeling (BIM) or a previously captured dense point cloud model. We do not expect the reference object to be present in every LiDAR scan. Our approach uses the poses of the LOAM algorithm as an initial guess to refine them with scan-to-model alignment. To evaluate if the alignment was accurate, an EKF-based motion prior filtering step is employed. Subsequently, the past trajectory is optimized by adding the model-aligned pose as a pose graph constraint and the map of the LOAM algorithm is corrected to improve future localization and mapping. We evaluate our approach with data captured in a visual airplane inspection scenario inside an aircraft hangar. A 3D LiDAR sensor is mounted via a gimbal on an Unmanned Aerial Vehicle (UAV) and is continuously actuated. We compare the localization accuracy of the LOAM and R-LOAM algorithms when enabling or disabling our proposed reference object-based trajectory and map optimization extension. For three recorded datasets, enabling the proposed extension yields a reduction in Absolute Pose Error compared to conventional LOAM and R-LOAM, while being able to run online. This reduces drift and improves map quality.

Efficient and Probabilistic Adaptive Voxel Mapping for Accurate Online LiDAR Odometry

This letter proposes an efficient and probabilistic adaptive voxel mapping method for LiDAR odometry. The map is a collection of voxels; each contains one plane feature that enables the probabilistic representation of the environment and accurate registration of a new LiDAR scan. We further analyze the need for coarse-to-fine voxel mapping and then use a novel voxel map organized by a Hash table and octrees to build and update the map efficiently. We apply the proposed voxel map to an iterated extended Kalman filter and construct a maximum a posteriori probability problem for pose estimation. Experiments on the open KITTI dataset show the high accuracy and efficiency of our method compared to other state-of-the-art methods. Experiments on indoor and unstructured outdoor environments with solid-state LiDAR and non-repetitive scanning LiDAR further verify the adaptability of our mapping method to different environments and LiDAR scanning patterns (see our attached video <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> ). Our codes and dataset are open-sourced on Github <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>

Multi-Uncertainty Captured Multi-Robot Lidar Odometry and Mapping Framework for Large-Scale Environments

Multi-robot simultaneous localization and mapping (MR-SLAM) is of great importance for enhancing the efficiency of large-scale environment exploration. Despite remarkable advances in schemes for cooperation, there is a critical lack of approaches to handle multiple uncertainties inherent to MR-SLAM in large-scale environments. This paper proposes a multi-uncertainty captured multi-robot lidar odometry and mapping (MUC-LOAM) framework, to quantify and utilize the uncertainties of feature points and robot mutual poses in large-scale environments. A proposed hybrid weighting strategy for pose update is integrated into MUC-LOAM to handle feature uncertainty from distance changing and dynamic objects. A devised Bayesian Neural Network (BNN) is proposed to capture mutual pose uncertainty. Then the covariance propagation of quaternions to Euler angles conversion is leveraged to filter out unreliable mutual poses. Another covariance propagation through coordinate transformations in nonlinear optimization improves the accuracy of map merging. The feasibility and enhanced robustness of the proposed framework for large-scale exploration are validated on both public datasets and real-world experiments.

Fast Sparse LiDAR Odometry Using Self-Supervised Feature Selection on Intensity Images

Fast Sparse LiDAR Odometry Using Self-Supervised Feature Selection on Intensity Images

Marked-LIEO: Visual Marker-Aided LiDAR/IMU/Encoder Integrated Odometry.

In this paper, we propose a visual marker-aided LiDAR/IMU/encoder integrated odometry, Marked-LIEO, to achieve pose estimation of mobile robots in an indoor long corridor environment. In the first stage, we design the pre-integration model of encoder and IMU respectively to realize the pose estimation combined with the pose estimation from the second stage providing prediction for the LiDAR odometry. In the second stage, we design low-frequency visual marker odometry, which is optimized jointly with LiDAR odometry to obtain the final pose estimation. In view of the wheel slipping and LiDAR degradation problems, we design an algorithm that can make the optimization weight of encoder odometry and LiDAR odometry adjust adaptively according to yaw angle and LiDAR degradation distance respectively. Finally, we realize the multi-sensor fusion localization through joint optimization of an encoder, IMU, LiDAR, and camera measurement information. Aiming at the problems of GNSS information loss and LiDAR degradation in indoor corridor environment, this method introduces the state prediction information of encoder and IMU and the absolute observation information of visual marker to achieve the accurate pose of indoor corridor environment, which has been verified by experiments in Gazebo simulation environment and real environment.