Normal Distribution Transform Research Articles

LIFNS: Design of a novel Lidar-IMU fusion navigation system for AGVs in smart factories

Traditional Automated Guided Vehicle (AGV) robots commonly employ 2D plane navigation systems. However, with the expansion of AGV application scenarios and the increasing complexity of structures, the limitations of existing 2D navigation systems have become more pronounced, rendering them inadequate for addressing these challenges. To tackle this issue, this paper proposes a novel navigation system suitable for AGV robots composed of fused Lidar and Inertial Measurement Unit (IMU), named as Lidar-IMU Fusion Navigation System (LIFNS). LIFNS primarily comprises mapping, localization, and fused path planning modules. In the mapping module, a 3D point cloud map for precise 3D localization and a 2D grid map for path planning are constructed using multiple-line Lidar and IMU. For the localization module, a fusion localization approach is introduced, combining IMU data with Normal Distributions Transform (NDT) point cloud registration through Unscented Kalman Filtering (UKF). Finally, global path planning is executed using the A* algorithm on the grid map, while local path planning utilizes the Timed-Elastic Band (TEB) algorithm. The effectiveness and universality of LIFNS are validated through simulation experiments and real-world deployment tests. The experimental results demonstrate that LIFNS achieves centimeter-level accuracy in both mapping and localization, effectively alleviating the issues of low precision and significant limitations present in traditional AGV robots. This positions LIFNS with promising applications in enclosed settings such as smart factories, industrial parks, and healthcare facilities.

Navigation system for orchard spraying robot based on 3D LiDAR SLAM with NDT_ICP point cloud registration

Autonomous navigation of orchard spraying robots can effectively improve operational efficiency and reduce workload. The conventional robot navigation solutions based on the Global Navigation Satellite System (GNSS) are susceptible to signal loss in orchards due to interference from the canopy of fruit trees. In this paper, an autonomous navigation scheme for orchard spraying robots based on multi-sensor devices and the three-dimensional (3D) LiDAR Simultaneous Localization and Mapping (SLAM) technology. The initial step of the scheme involved the construction of a 3D point cloud map of the orchard using the 3D LiDAR SLAM algorithm. Subsequently, a designed point cloud map transformation algorithm was used to convert the 3D point cloud map of the orchard into a two-dimensional (2D) grid map containing key feature information of the orchard. To enhance the mapping accuracy of the orchard point cloud, a point cloud registration algorithm combining Iterative Closet Point (ICP) and Normal Distributions Transform (NDT) was proposed. This algorithm initially utilized the NDT algorithm for coarse point cloud registration to obtain transformation parameters. Then, the ICP algorithm was applied in conjunction with the obtained parameters for fine registration. Regarding obstacle avoidance, a cooperative obstacle avoidance algorithm based on the Robot Operating System (ROS) multithreaded communication mechanism was introduced. Experimental trials conducted in a standardized spindle-shaped peach orchard validated the successful completion of the SLAM mapping trajectory, which was achieved through the utilization of the proposed NDT_ICP point cloud registration algorithm. The average absolute pose error between the mapped trajectory and the reference trajectory was 1.173 m, with a standard deviation of 0.498 m. The robot's positioning and navigation errors increased with higher speeds, with the largest positioning deviation observed at 1.2 m/s speed, with a maximum lateral positioning error of 7.04 cm. The robot's average lateral navigation error did not exceed 16 cm, and the average heading deviation was less than 8° at different speeds of 0.4, 0.8 and 1.2 m/s. Both the positioning and navigation accuracies of the robot were sufficient to meet the autonomous operational requirements of orchard spraying robots.

A Multimodal Fusion System for Object Identification in Point Clouds with Density and Coverage Differences

Data fusion, which involves integrating information from multiple sources to achieve a specific objective, is an essential area of contemporary scientific research. This article presents a multimodal fusion system for object identification in point clouds in a controlled environment. Several stages were implemented, including downsampling and denoising techniques, to prepare the data before fusion. Two denoising approaches were tested and compared: one based on neighborhood technique and the other using a median filter for each “x”, “y”, and “z” coordinate of each point. The downsampling techniques included Random, Grid Average, and Nonuniform Grid Sample. To achieve precise alignment of sensor data in a common coordinate system, registration techniques such as Iterative Closest Point (ICP), Coherent Point Drift (CPD), and Normal Distribution Transform (NDT) were employed. Despite facing limitations, variations in density, and differences in coverage among the point clouds generated by the sensors, the system successfully achieved an integrated and coherent representation of objects in the controlled environment. This accomplishment establishes a robust foundation for future research in the field of point cloud data fusion.

LeGO-LOAM-FN: An Improved Simultaneous Localization and Mapping Method Fusing LeGO-LOAM, Faster_GICP and NDT in Complex Orchard Environments.

To solve the problem of cumulative errors when robots build maps in complex orchard environments due to their large scene size, similar features, and unstable motion, this study proposes a loopback registration algorithm based on the fusion of Faster Generalized Iterative Closest Point (Faster_GICP) and Normal Distributions Transform (NDT). First, the algorithm creates a K-Dimensional tree (KD-Tree) structure to eliminate the dynamic obstacle point clouds. Then, the method uses a two-step point filter to reduce the number of feature points of the current frame used for matching and the number of data used for optimization. It also calculates the matching degree of normal distribution probability by meshing the point cloud, and optimizes the precision registration using the Hessian matrix method. In the complex orchard environment with multiple loopback events, the root mean square error and standard deviation of the trajectory of the LeGO-LOAM-FN algorithm are 0.45 m and 0.26 m which are 67% and 73% higher than those of the loopback registration algorithm in the Lightweight and Ground-Optimized LiDAR Odometry and Mapping on Variable Terrain (LeGO-LOAM), respectively. The study proves that this method effectively reduces the influence of the cumulative error, and provides technical support for intelligent operation in the orchard environment.

3DMNDT: 3D Multi-View Registration Method Based on the Normal Distributions Transform

The normal distributions transform (NDT) is an effective paradigm for point set registration. This method was initially designed for pair-wise registration and suffers from the accumulated error problem when directly applied to multi-view registration. Under the framework of point-to-cluster correspondence, this paper proposes a novel multi-view registration method named 3D multi-view registration based on the normal distributions transform (3DMNDT), which integrates the k-means clustering and Lie algebra optimizer to achieve multi-view registration. More specifically, the multi-view registration is cast into the maximum likelihood estimation problem. Firstly, k-means clustering is utilized to divide all data points into different clusters, where one normal distribution is computed to locally model the probability of measuring a data point in each cluster. Subsequently, the multi-view registration problem is formulated by the NDT-based likelihood function. To maximize this likelihood function, the Lie algebra optimizer is introduced and developed to optimize each rigid transformation sequentially. 3DMNDT implements data point clustering, NDT computing, and rigid transformation optimization alternately until the desired registration results are obtained. Experimental results tested on benchmark data sets illustrate that 3DMNDT can achieve state-of-the-art performance for multi-view registration. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This paper is motivated by solving the problem of registering multiple point sets. The normal distributions transform (NDT) is a well-known pair-wise registration method widely applied in the robotic domain. This paper extends the original NDT and proposes a novel registration method to simultaneously align more than two point sets. The multi-view registration is cast into the maximum likelihood estimation problem. Subsequently, the k-means clustering and Lie algebra optimizer are integrated to estimate registration parameters. Experimental results demonstrate its superior performance on the accuracy, efficiency, and robustness for multi-view registration of point sets.

NDT RC: Normal Distribution Transform Occupancy 3D Mapping With Recentering

The Normal Distribution Transform Occupancy Map (NDT OM) is a mapping algorithm able to represent a dynamic 3D environment. The resulting map has fixed boundaries, thus a robot with unbounded displacement might fall outside of the map due to memory limitation. In this paper, a recentering algorithm called NDT RC is proposed to avoid this issue. NDT RC extends the use of NDT OM for vehicles with unbounded displacements. NDT RC provides a seamless translation of the map as the robot gets far from the center of the previous map. The influence of NDT RC on the precision of the estimated trajectory of the robot, or odometry, is examined on two publicly available datasets, the KITTI and Ford datasets. An analysis of the sensitivity of the NDT RC to its tuning parameters is carried out using the Ford dataset, while the KITTI dataset is used to measure the influence of the density of the input point cloud. The results show that the proposed recentering strategy improves the accuracy of the odometry calculated by registering the latest lidar scan on the generated map compared to other NDT based approaches (NDT OM, NDT OM Fusion, SE-NDT). In particular, the proposed method, which does not perform loop closure, reduces the mean absolute translation error by 16% and the runtime by 88% compared to the NDT OM Fusion on the Ford dataset.

Advanced 3D Navigation System for AGV in Complex Smart Factory Environments

The advancement of Industry 4.0 has significantly propelled the widespread application of automated guided vehicle (AGV) systems within smart factories. As the structural diversity and complexity of smart factories escalate, the conventional two-dimensional plan-based navigation systems with fixed routes have become inadequate. Addressing this challenge, we devised a novel mobile robot navigation system encompassing foundational control, map construction positioning, and autonomous navigation functionalities. Initially, employing point cloud matching algorithms facilitated the construction of a three-dimensional point cloud map within indoor environments, subsequently converted into a navigational two-dimensional grid map. Simultaneously, the utilization of a multi-threaded normal distribution transform (NDT) algorithm enabled precise robot localization in three-dimensional settings. Leveraging grid maps and the robot’s inherent localization data, the A* algorithm was utilized for global path planning. Moreover, building upon the global path, the timed elastic band (TEB) algorithm was employed to establish a kinematic model, crucial for local obstacle avoidance planning. This research substantiated its findings through simulated experiments and real vehicle deployments: Mobile robots scanned environmental data via laser radar and constructing point clouds and grid maps. This facilitated centimeter-level localization and successful circumvention of static obstacles, while simultaneously charting optimal paths to bypass dynamic hindrances. The devised navigation system demonstrated commendable autonomous navigation capabilities. Experimental evidence showcased satisfactory accuracy in practical applications, with positioning errors of 3.6 cm along the x-axis, 3.3 cm along the y-axis, and 4.3° in orientation. This innovation stands to substantially alleviate the low navigation precision and sluggishness encountered by AGV vehicles within intricate smart factory environments, promising a favorable prospect for practical applications.

Experimental Study of Seamless Switch Between GNSS- and LiDAR-Based Self-Localization

A self-localization method that can seamlessly switch positions and attitudes estimated using normal distributions transform (NDT) scan matching and a real-time kinematic global navigation satellite system (GNSS) is successfully developed. One of the issues encountered in this method is the sharing of global coordinates among the different estimation methods. Therefore, the three-dimensional environmental maps utilized in the NDT scan matching are created based on the planar Cartesian coordinate system used in the GNSS to obtain accurate information regarding the location, shape, and size of the actual terrain and geographic features. Consequently, seamlessly switching between different methods enables mobile robots to stably obtain accurate estimated positions and attitudes. An autonomous driving experiment is conducted using this self-localization method in the Tsukuba Challenge 2022, and the mobile robot completed a designated course involving more than 2 km in an urban area.

An Underwater Distributed SLAM Approach Based on Improved GMRBnB Framework

Multi-vehicle collaborative mapping proves more efficient in constructing maps in unfamiliar underwater environments in comparison to single-vehicle methods. One of the pivotal hurdles of Simultaneous Localization and Mapping (SLAM) with multiple underwater vehicles is map registration. Due to the inadequate characteristics of the underwater grid maps, matching map features poses a challenge, and outliers between maps add to the complexity. We propose an algorithm to solve this problem. This approach employs the Gaussian Mixture Robust Branch and Bound (GMRBnB) algorithm with an interior point filtering technique. Feature point extraction, registration using the GMRBnB algorithm, inlier extraction based on density, and registration of the inlier are performed to obtain a more precise transformation matrix. The results of the simulation and experiments demonstrate that this technique heightens outlier tolerance and reinforces map registration accuracy. The proposed approach surpasses Iterative Closest Point (ICP) and Normal Distributions Transform (NDT) methods with respect to map registration quality.

Research on applying scan matching algorithm to improve velocity measurement accuracy of single-line ranging lidar

There has been increased demand for safe and efficient speed measurement of moving objects in engineering and practical applications, particularly in scenarios involving the measurement of the movement speed of objects at close range. Thus, this study designed a single-line scanning pulsed lidar system. In addition, algorithms were employed to enhance the measurement accuracy of the single-line ranging pulsed lidar. In this study, the lidar was installed on a guideway-integrated testing platform to obtain point cloud data, which was processed using the normal distributions transform (NDT), iterative closest points (ICP), and correlative scan matching (CSM) algorithms. The experimental results demonstrated that the NDT algorithm outperformed the ICP and CSM algorithms in terms of velocity measurement accuracy and algorithm robustness. Moreover, the velocity measurement results obtained from the NDT algorithm were closer to the actual track speed, with a maximum velocity deviation of lesser than 0.021 m/s compared to the actual guideway speed.

GNSS/IMU/LiDAR fusion for vehicle localization in urban driving environments within a consensus framework

The continuous, reliable, and accurate acquisition of position information is a fundamental requirement for autonomous vehicles. It underpins perception, decision-making, and planning modules in automated driving systems. However, the performance of multi-modal sensors, including the Global Navigation Satellite System (GNSS), Inertial Measurement Unit (IMU), and LiDAR, typically found in autonomous vehicles, is easily influenced by complex environmental factors encountered in urban driving conditions. This creates challenges for existing localization algorithms. More specifically, while GNSS performs accurately in open-sky areas, it is susceptible to the multi-path effect. Although the IMU-based Inertial Navigation System (INS) provides continuous position data, it suffers from cumulative errors. LiDAR-based map-matching is highly effective when sufficient point cloud features are present but fall short in plain environments. To address these individual sensor deficiencies, we propose the application of a consensus framework for GNSS/IMU/LiDAR fusion, and it is able to be expanded for multi-modal sensor fusion. Our approach seeks to take advantage of the redundancy inherent in these sensor systems. Our method begins with the derivation of the INS mechanization and its error dynamics. Subsequently, based on the Consensus Kalman Filter (CKF), we establish a GNSS measurement node and a Normal Distribution Transformation (NDT) map-matching measurement node. These are designed to estimate position errors in the INS. To counter sensor measurement failure or degradation, we introduce an adaptation mechanism for the CKF. This ensures position accuracy and continuity by normalizing measurement quality across different measurement nodes. Finally, to validate our consensus-based localization framework, we conducted extensive real-world vehicle tests in urban driving conditions. The results have confirmed the superior accuracy and robustness of our approach, even in the face of individual sensor failure. In our experiments, the proposed method reduces the localization mean absolute error by around 17% compared to INS/map-matching fusion, and around 59% compared to INS/GNSS fusion while map-matching fails. While the GNSS and map-matching measurements contain large noises, our framework achieves at least approximately a 30% improvement compared to traditional methods.

A Fast Registration Method for MEMS LiDAR Point Cloud Based on Self-Adaptive Segmentation

The Micro-Electro-Mechanical System (MEMS) LiDAR point cloud in autonomous vehicles has a large deflection range, which results in slow registration speed and poor applicability. To maximize speed, an improved Normal Distribution Transform (NDT) method that integrates point cloud density features has been proposed. First, the point cloud is reduced using a modified voxel filter and a pass-through filter. Next, the Intrinsic Shape Signature (ISS) algorithm is utilized to analyze the point cloud features and extract key points; the Four-Point Congruent Set (4PCS) algorithm is then employed to calculate the initial pose under the constraints of the key point set to complete the coarse registration. Finally, the self-adaptive segmentation model is constructed by using a K-D tree to obtain the density features of key points, and the NDT algorithm is combined with this model to form an SSM-NDT algorithm, which is used for fine registration. Each algorithm was compared on the autonomous vehicle dataset PandaSet and actual collected datasets. The results show that the novel method increases the speed by at least 60% and takes into account good registration accuracy and strong anti-interference.

Survey of Point Cloud Registration Methods and New Statistical Approach

The use of a 3D range scanning device for autonomous object description or unknown environment mapping leads to the necessity of improving computer methods based on identical point pairs from different point clouds (so-called registration problem). The registration problem and three-dimensional transformation of coordinates still require further research. The paper attempts to guide the reader through the vast field of existing registration methods so that he can choose the appropriate approach for his particular problem. Furthermore, the article contains a regression method that enables the estimation of the covariance matrix of the transformation parameters and the calculation of the uncertainty of the estimated points. This makes it possible to extend existing registration methods with uncertainty estimates and to improve knowledge about the performed registration. The paper’s primary purpose is to present a survey of known methods and basic estimation theory concepts for the point cloud registration problem. The focus will be on the guiding principles of the estimation theory: ICP algorithm; Normal Distribution Transform; Feature-based registration; Iterative dual correspondences; Probabilistic iterative correspondence method; Point-based registration; Quadratic patches; Likelihood-field matching; Conditional random fields; Branch-and-bound registration; PointReg. The secondary purpose of this article is to show an innovative statistical model for this transformation problem. The new theory needs known covariance matrices of identical point coordinates. An unknown rotation matrix and shift vector have been estimated using a nonlinear regression model with nonlinear constraints. The paper ends with a relevant numerical example.

Active Navigation System for a Rubber-Tapping Robot Based on Trunk Detection

To address the practical navigation issues of rubber-tapping robots, this paper proposes an active navigation system guided by trunk detection for a rubber-tapping robot. A tightly coupled sliding-window-based factor graph method is proposed for pose tracking, which introduces normal distribution transform (NDT) measurement factors, inertial measurement unit (IMU) pre-integration factors, and prior factors generated by sliding window marginalization. To actively pursue goals in navigation, a distance-adaptive Euclidean clustering method is utilized in conjunction with cylinder fitting and composite criteria screening to identify tree trunks. Additionally, a hybrid map navigation approach involving 3D point cloud map localization and 2D grid map planning is proposed to apply these methods to the robot. Experiments show that our pose-tracking approach obtains generally better performance in accuracy and robustness compared to existing methods. The precision of our trunk detection method is 93% and the recall is 87%. A practical validation is completed in robot rubber-tapping tasks of a real rubber plantation. The proposed method can guide the rubber-tapping robot in complex forest environments and improve efficiency.

Efficient and Accurate Vehicle Localization Based on LiDAR Place Recognition

An efficient and accurate LiDAR place recognition methodology is proposed for vehicle localization. First, the Iris-LOAM is proposed to overcome the disadvantages of low accuracy of loop-closure detection and low efficiency of map construction in the existing LOAM-series methods. The method integrates the LiDAR-Iris global descriptor and Normal Distribution Transform (NDT) registration method into the loop-closure detection module of LiDAR Odometry and Mapping (LOAM), thereby improving the accuracy and efficiency of map construction. For the shortcomings of low map loading and matching efficiency, the Random Sample Consensus method is used to remove the ground point cloud information. The Voxel Grid method is used to down sample the loaded map. Finally, the NDT method is adopted for point cloud map matching to obtain the position information. Show that the Iris-LOAM has higher efficiency than the SC-LeGO-LOAM. The average time of point cloud map matching is reduced by 39.5%. The place recognition can be executed to achieve accuracy vehicle localization.

Research and Application of the Obstacle Avoidance System for High-Speed Railway Tunnel Lining Inspection Train Based on Integrated 3D LiDAR and 2D Camera Machine Vision Technology

This study presents an innovative, intelligent obstacle avoidance module intended to significantly enhance the collision prevention capabilities of the robotic arm mechanism onboard a high-speed rail tunnel lining inspection train. The proposed module employs a fusion of ORB-SLAM3 and Normal Distribution Transform (NDT) point cloud registration techniques to achieve real-time point cloud densification, ensuring reliable detection of small-volume targets. By leveraging spatial filtering, cluster computation, and feature extraction, precise obstacle localization information is further obtained. A fusion of multi-modal data is achieved by jointly calibrating 3D LiDAR and camera images. Upon validation through field testing, it is demonstrated that the module can effectively detect obstacles with a minimum diameter of 0.5 cm, with an average deviation controlled within a 1–2 cm range and a safety margin of 3 cm, effectively preventing collisions. Compared to traditional obstacle avoidance sensors, this module provides information across more dimensions, offering robust support for the construction of powerful automated tunnel inspection control systems and digital twin lifecycle analysis techniques for railway tunnels.

A Benchmark for Multi-Modal LiDAR SLAM with Ground Truth in GNSS-Denied Environments

LiDAR-based simultaneous localization and mapping (SLAM) approaches have obtained considerable success in autonomous robotic systems. This is in part owing to the high accuracy of robust SLAM algorithms and the emergence of new and lower-cost LiDAR products. This study benchmarks the current state-of-the-art LiDAR SLAM algorithms with a multi-modal LiDAR sensor setup, showcasing diverse scanning modalities (spinning and solid state) and sensing technologies, and LiDAR cameras, mounted on a mobile sensing and computing platform. We extend our previous multi-modal multi-LiDAR dataset with additional sequences and new sources of ground truth data. Specifically, we propose a new multi-modal multi-LiDAR SLAM-assisted and ICP-based sensor fusion method for generating ground truth maps. With these maps, we then match real-time point cloud data using a normal distributions transform (NDT) method to obtain the ground truth with a full six-degrees-of-freedom (DOF) pose estimation. These novel ground truth data leverage high-resolution spinning and solid-state LiDARs. We also include new open road sequences with GNSS-RTK data and additional indoor sequences with motion capture (MOCAP) ground truth, complementing the previous forest sequences with MOCAP data. We perform an analysis of the positioning accuracy achieved, comprising ten unique configurations generated by pairing five distinct LiDAR sensors with five SLAM algorithms, to critically compare and assess their respective performance characteristics. We also report the resource utilization in four different computational platforms and a total of five settings (Intel and Jetson ARM CPUs). Our experimental results show that the current state-of-the-art LiDAR SLAM algorithms perform very differently for different types of sensors. More results, code, and the dataset can be found at GitHub.

Research on orchard navigation method based on fusion of 3D SLAM and point cloud positioning.

Accurate navigation is crucial in the construction of intelligent orchards, and the need for vehicle navigation accuracy becomes even more important as production is refined. However, traditional navigation methods based on global navigation satellite system (GNSS) and 2D light detection and ranging (LiDAR) can be unreliable in complex scenarios with little sensory information due to tree canopy occlusion. To solve these issues, this paper proposes a 3D LiDAR-based navigation method for trellis orchards. With the use of 3D LiDAR with a 3D simultaneous localization and mapping (SLAM) algorithm, orchard point cloud information is collected and filtered using the Point Cloud Library (PCL) to extract trellis point clouds as matching targets. In terms of positioning, the real-time position is determined through a reliable method of fusing multiple sensors for positioning, which involves transforming the real-time kinematics (RTK) information into the initial position and doing a normal distribution transformation between the current frame point cloud and the scaffold reference point cloud to match the point cloud position. For path planning, the required vector map is manually planned in the orchard point cloud to specify the path of the roadway, and finally, navigation is achieved through pure path tracking. Field tests have shown that the accuracy of the normal distributions transform (NDT) SLAM method can reach 5cm in each rank with a coefficient of variation that is less than 2%. Additionally, the navigation system has a high positioning heading accuracy with a deviation within 1° and a standard deviation of less than 0.6° when moving along the path point cloud at a speed of 1.0m/s in a Y-trellis pear orchard. The lateral positioning deviation was also controlled within 5cm with a standard deviation of less than 2cm. This navigation system has a high level of accuracy and can be customized to specific tasks, making it widely applicable in trellis orchards with autonomous navigation pesticide sprayers.

Benchmark of multi-view Terrestrial Laser Scanning Point Cloud data registration algorithms

This study addresses multi-view Terrestrial Laser Scanning Point Cloud data registration methods. Multiple rigid point cloud data registration is mandatory for aligning all scans into a common reference frame and it is still considered a challenge looking from a large-scale surveys point of view. The goal of this work is to support the development of cutting-edge registration methods in geoscience and mobile robotics domains. This work evaluates 3 data sets of total 20 scenes available in the literature. This paper provides a novel open-source framework for multi-view Terrestrial Laser Scanning Point Cloud data registration benchmarks. The goal was to verify experimentally which registration variant can improve the open-source data looking from the quantitative and qualitative points of view. In particular, the following scanners provided measurement data: Z+F TLS Imager 5006i, Z+F TLS Imager 5010C, Leica ScanStation C5, Leica ScanStation C10, Leica P40 and Riegl VZ-400. The benchmark shows an impact of the metric e.g. point to point, point to projection onto a plane, plane to plane etc..., rotation matrix parameterization (Tait–Bryan, quaternion, Rodrigues) and other implementation variations (e.g. multi-view Normal Distributions Transform, Pose Graph SLAM approach) onto the multi-view data registration accuracy and performance. An open-source project is created and it can be used for improving existing data sets reported in the literature, it is the added value of the presented research. The combination of metrics, rotation matrix parameterization and optimization algorithms creates hundreds of possible approaches. It is shown that chosen metric is a dominant factor in data registration. The rotation parameterization and other degrees of freedom of proposed variants are rather negligible compared with chosen metric. Most of the proposed approaches improve registered reference data provided by other researchers. Only for 2 from 20 scenes it was not possible to provide significant improvement. The largest improvements are evident for large-scale scenes. The project is available and maintained at https://github.com/MapsHD/HDMapping.

ASL-SLAM: A LiDAR SLAM With Activity Semantics-Based Loop Closure

A practical back-end module with loop closure detection is very useful and important for a LiDAR simultaneous localization and mapping (SLAM) system to perform high-precision positioning and mapping tasks. However, most existing loop closure detection methods are based on images or point clouds, and these methods may produce errors when the structure or texture is similar. To overcome this problem, we propose a complete LiDAR SLAM system, including a front-end odometry module based on normal distribution transform (NDT)-LOAM and a back-end optimization module with loop closure based on activity semantics. Through the analysis and calculation of inertial measurement unit (IMU) data from SLAM platforms such as unmanned ground vehicles (UGVs), the activity semantics of turning and passing over a speed bump are detected based on the peak <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${z}$ </tex-math></inline-formula> -axis angular velocity and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${z}$ </tex-math></inline-formula> -axis acceleration, respectively. Then, according to this activity semantics information and its unique and definite attributes, we establish correct loop closure detection using rough geometric detection, activity semantics matching, and point cloud rematching for validation. Finally, graph optimization theory is utilized to reduce the global cumulative error, improve the global trajectory accuracy and map consistency, and obtain the final global motion trajectory and point cloud map. We collected a dataset for evaluation, which contains indoor data, outdoor data, and indoor–outdoor integration data, and we also evaluated our method on the KITTI dataset. The experimental results for different scenes show that the addition of activity semantics can effectively help loop closure detection and improve LiDAR SLAM system performance.