Iterative Closest Point Method Research Articles

Accuracy evaluation of dental CBCT and scanned model registration method based on pulp horn mapping surface: an in vitro proof-of-concept

Background and aim3D fusion model of cone-beam computed tomography (CBCT) and oral scanned data can be used for the accurate design of root canal access and guide plates in root canal therapy (RCT). However, the pose accuracy of the dental pulp and crown in data registration has not been investigated, which affects the precise implementation of clinical planning goals. We aimed to establish a novel registration method based on pulp horn mapping surface (PHMSR), to evaluate the accuracy of PHMSR versus traditional methods for crown-pulp registration of CBCT and oral scan data.Materials and methodsThis vitro study collected 8 groups of oral scanned and CBCT data in which the left mandibular teeth were not missing, No. 35 and No. 36 teeth were selected as the target teeth. The CBCT and scanned model were processed to generate equivalent point clouds. For the PHMSR method, the similarity between the feature directions of the pulp horn and the surface normal vectors of the crown were used to determine the mapping points in the CBCT point cloud that have a great influence on the pulp pose. The small surface with adjustable parameters is reconstructed near the mapping point of the crown, and the new matching point pairs between the point and the mapping surface are searched. The sparse iterative closest point (ICP) algorithm is used to solve the new matching point pairs. Then, in the C + + programming environment with a point cloud library (PCL), the PHMSR, the traditional sparse ICP, ICP, and coherent point drift (CPD) algorithms are used to register the point clouds under two different initial deviations. The root square mean error (RSME) of the crown, crown-pulp orientation deviation (CPOD), and position deviation (CPPD) were calculated to evaluate the registration accuracy. The significance between the groups was tested by a two-tailed paired t-test (p < 0.05).ResultsThe crown RSME values of the sparse ICP method (0.257), the ICP method (0.217), and the CPD method (0.209) were not significantly different from the PHMSR method (0.250). The CPOD and CPPD values of the sparse ICP method (4.089 and 0.133), the ICP method (1.787 and 0.700), and the CPD method (1.665 and 0.718) than for the PHMSR method, which suggests that the accuracy of crown-pulp registration is higher with the PHMSR method.ConclusionCompared with the traditional method, the PHMSR method has a smaller crown-pulp registration accuracy and a clinically acceptable deviation range, these results support the use of PHMSR method instead of the traditional method for clinical planning of root canal therapy.

Reliability of Artificial Intelligence-based Cone Beam Computed Tomography Integration with Digital Dental Images.

This study aimed to introduce cone-beam computed tomography (CBCT) digitization and integration of digital dental images (DDI) based on artificial intelligence (AI)-based registration (ABR) and to evaluate the reliability and reproducibility using this method compared with those of surface-based registration (SBR). This retrospective study consisted of CBCT images and DDI of 17 patients who had undergone computer-aided bimaxillary orthognathic surgery. The digitization of CBCT images and their integration with DDI were repeated using an AI-based program. CBCT images and DDI were integrated using a point-to-point registration. In contrast, with the SBR method, the three landmarks were identified manually on the CBCT and DDI, which were integrated with the iterative closest points method. After two repeated integrations of each method, the three-dimensional coordinate values of the first maxillary molars and central incisors and their differences were obtained. Intraclass coefficient (ICC) testing was performed to evaluate intra-observer reliability with each method's coordinates and compare their reliability between the ABR and SBR. The intra-observer reliability showed significant and almost perfect ICC in each method. There was no significance in the mean difference between the first and second registrations in each ABR and SBR and between both methods; however, their ranges were narrower with ABR than with the SBR method. This study shows that AI-based digitization and integration are reliable and reproducible.

A Comparative Analysis of UAV Photogrammetric Software Performance for Forest 3D Modeling: A Case Study Using AgiSoft Photoscan, PIX4DMapper, and DJI Terra.

Three-dimensional (3D) modeling of trees has many applications in various areas, such as forest and urban planning, forest health monitoring, and carbon sequestration, to name a few. Unmanned Aerial Vehicle (UAV) photogrammetry has recently emerged as a low cost, rapid, and accurate method for 3D modeling of urban and forest trees replacing the costly traditional methods such as plot measurements and surveying. There are numerous commercial and open-source software programs available, each processing UAV data differently to generate forest 3D modeling and photogrammetric products, including point clouds, Digital Surface Models (DSMs), Canopy Height Models (CHMs), and orthophotos in forest areas. The objective of this study is to compare the three widely-used commercial software packages, namely, AgiSoft Photoscan (Metashape) V 1.7.3, PIX4DMapper (Pix4D) V 4.4.12, and DJI Terra V 3.7.6 for processing UAV data over forest areas from three perspectives: point cloud density and reconstruction quality, computational time, DSM assessment for height accuracy (z) and ability of tree detection on DSM. Three datasets, captured by UAVs on the same day at three different flight altitudes, were used in this study. The first, second, and third datasets were collected at altitudes of 60 m, 100 m, and 120 m, respectively over a forested area in Tully, New York. While the first and third datasets were taken horizontally, the second dataset was taken 20 degrees off-nadir to investigate the impact of oblique images. Results show that Pix4D and AgiSoft generate 2.5 times denser point clouds than DJI Terra. However, reconstruction quality evaluation using the Iterative Closest Point method (ICP) shows DJI Terra has fewer gaps in the point cloud and performed better than AgiSoft and Pix4D in generating a point cloud of trees, power lines and poles despite producing a fewer number of points. In other words, the outperformance in key points detection and an improved matching algorithm are key factors in generating improved final products. The computational time comparison demonstrates that the processing time for AgiSoft and DJI Terra is roughly half that of Pix4D. Furthermore, DSM elevation profiles demonstrate that the estimated height variations between the three software range from 0.5 m to 2.5 m. DJI Terra's estimated heights are generally greater than those of AgiSoft and Pix4D. Furthermore, DJI Terra outperforms AgiSoft and Pix4D for modeling the height contour of trees, buildings, and power lines and poles, followed by AgiSoft and Pix4D. Finally, in terms of the ability of tree detection, DJI Terra outperforms AgiSoft and Pix4D in generating a comprehensive DSM as a result of fewer gaps in the point cloud. Consequently, it stands out as the preferred choice for tree detection applications. The results of this paper can help 3D model users to have confidence in the reliability of the generated 3D models by comprehending the accuracy of the employed software.

Three-Dimensional Quantitative Assessment of Pedicle Screw Accuracy in Clinical Utilization of a New Robotic System in Spine Surgery: A Multicenter Study.

The objective of this study was to evaluate the accuracy of pedicle screw placement in patients undergoing percutaneous pedicle screw fixation with robotic guidance, using a newly developed 3-dimensional quantitative measurement system. The study also aimed to assess the clinical feasibility of the robotic system in the field of spinal surgery. A total of 113 patients underwent pedicle screw insertion using the CUVIS-spine pedicle screw guide system (CUREXO Inc.). Intraoperative O-arm images were obtained, and screw insertion pathways were planned accordingly. Image registration was performed using paired-point registration and iterative closest point methods. The accuracy of the robotic-guided pedicle screw insertion was assessed using 3-dimensional offset calculation and the Gertzbein-Robbins system (GRS). A total of 448 screws were inserted in the 113 patients. The image registration success rate was 95.16%. The average error of entry offset was 2.86 mm, target offset was 2.48 mm, depth offset was 1.99 mm, and angular offset was 3.07°. According to the GRS grading system, 88.39% of the screws were classified as grade A, 9.60% as grade B, 1.56% as grade C, 0.22% as grade D, and 0.22% as grade E. Clinically acceptable screws (GRS grade A or B) accounted for 97.54% of the total, with no reported neurologic complications. Our study demonstrated that pedicle screw insertion using the novel robot-assisted navigation method is both accurate and safe. Further prospective studies are necessary to explore the potential benefits of this robot-assisted technique in comparison to conventional approaches.

Development of Notch-Free, Pre-Bent Rod Applicable for Posterior Corrective Surgery of Thoracolumbar/Lumbar Adolescent Idiopathic Scoliosis.

Adolescent idiopathic scoliosis (AIS), the most common pediatric musculoskeletal disorder, causes a three-dimensional spine deformity. Lenke type 5 AIS is defined as a structural thoracolumbar/lumbar curve with nonstructural thoracic curves. Although a rod curvature will affect clinical outcomes, intraoperative contouring of the straight rod depends on the surgeon's knowledge and experience. This study aimed to determine the optimum rod geometries to provide a pre-bent rod system for posterior spinal surgery in patients with Lenke type 5 AIS. These pre-bent rods will be beneficial for achieving proper postoperative outcomes without rod contouring based on surgeon experience. We investigated 20 rod geometries traced in posterior spinal reconstruction in patients with Lenke type 5 AIS. The differences between the center point clouds in each cluster were evaluated using the iterative closest point (ICP) method with modification. Before the evaluation using the ICP method, the point clouds were divided into four clusters based on the rod length using a hierarchical cluster analysis. Because the differences in the values derived from the ICP method were <5 mm for each length-based cluster, four representative rod shapes were generated from the length-based clusters. We identified four optimized rod shapes that will reduce operation time, leading to a decreased patient and surgeon burden.

Accuracy analysis for machine tool spindles considering full parallel connections and form errors based on skin model shapes

Abstract The rotation accuracy of a machine tool spindle is essential for ensuring the machining precision. Due to the existence of manufacturing and assembly errors, the rotation accuracy of the spindle will be inevitably impacted and degraded. Therefore, to reduce the influence of the errors and improve the work performance, this paper focuses on accuracy analysis for the spindle and a novel optimization-oriented skin model shape method to tackle this highly complex problem. First, a structural analysis of the spindle is carried out to elaborate the intractable full parallel collections in the assembly. Then, based on the iterative closest point method, the deviation propagation of the spindle considering complex full parallel collections is transformed into an optimization problem, in which the skin model shapes and small displacement torsor are utilized to represent the form and pose errors of the part, respectively. By solving the optimization problem, assembly accuracy analysis for the spindle in terms of full parallel connections and form errors is accordingly achieved. On this basis, the tolerance analysis model of the spindle is also comprehensively established by employing the corresponding error simulation. Finally, measurement experiments are conducted to validate the effectiveness of the proposed method. The experiments show the predicted rotation runout and tolerance magnitude are close to the testing results, therefore indicating the proposed method can provide effective accuracy analysis for spindles.

Accuracy and inter-cloud precision of low-cost mobile LiDAR technology in estimating soil disturbance in forest operations

Forest operations can cause long-term soil disturbance, leading to environmental and economic losses. Mobile LiDAR technology has become increasingly popular in forest management for mapping and monitoring disturbances. Low-cost mobile LiDAR technology, in particular, has attracted significant attention due to its potential cost-effectiveness, ease of use, and ability to capture high-resolution data. The LiDAR technology, which is integrated in the iPhone 13–14 Pro Max series, has the potential to provide high accuracy and precision data at a low cost, but there are still questions on how this will perform in comparison to professional scanners. In this study, an iPhone 13 Pro Max equipped with SiteScape and 3D Scanner apps, and the GeoSlam Zeb Revo scanner were used to collect and generate point cloud datasets for comparison in four plots showing variability in soil disturbance and local topography. The data obtained from the LiDAR devices were analyzed in CloudCompare using the Iterative Closest Point (ICP) and Least Square Plane (LSP) methods of cloud-to-cloud comparisons (C2C) to estimate the accuracy and intercloud precision of the LiDAR technology. The results showed that the low-cost mobile LiDAR technology was able to provide accurate and precise data for estimating soil disturbance using both the ICP and LSP methods. Taking as a reference the point clouds collected with the Zeb Revo scanner, the accuracy of data derived with SiteScape and 3D Scanner apps varied from RMS = 0.016 to 0.035 m, and from RMS = 0.017 to 0.025 m, respectively. This was comparable to the precision or repeatability of the professional LiDAR instrument, Zeb Revo (RMS = 0.019–0.023 m). The intercloud precision of the data generated with SiteScape and 3D Scanner apps varied from RMS = 0.015 to 0.017 m and from RMS = 0.012 to 0.014 m, respectively, and were comparable to the precision of Zeb Revo measurements (RMS = 0.019–0.023 m). Overall, the use of low-cost mobile LiDAR technology fits well to the requirements to map and monitor soil disturbances and it provides a cost-effective and efficient way to gather high resolution data, which can assist the sustainable forest management practices.

RGB-D 데이터를 이용한 비전 기반 로봇 간 상대 위치 및 자세 추정 기법 연구

In this paper, a vision-based robot-to-robot relative pose estimation method is proposed for markerless wheeled mobile robots using an RGB-D camera. The proposed method comprises three parts: (i) object detection using a machine vision technique with RGB data, (ii) relative position estimation averaging over the point cloud with outlier rejection, and (iii) relative orientation estimation using point cloud scan matching through iterative closest point (ICP) methods. The neighboring or target robots are detected using YOLOv3-tiny so that the position of the robot can be determined in pixel coordinates. To estimate the 3D position of the target robot in the camera coordinates of the ego-robot, the object detection results are used along with the depth information and the precalibrated pinhole camera model. To estimate the orientation of the target robot, a point cloud is generated using the depth image, downsampled, and then used for scan matching by using ICP methods. For experimental verification, the proposed relative pose estimation method was implemented as a robot operating system package and tested on a small-scale differential-driving wheeled mobile robot, TurtleBot3; the estimation results were compared with the ground truth in real-world environments.

VOXEL-BASED POINT CLOUD LOCALIZATION FOR SMART SPACES MANAGEMENT

Abstract. This paper proposes a voxel-based approach for creating a digital twin of an urban environment that is capable of efficiently managing smart spaces. The paper explains the registration and localization procedure of the point cloud dataset, which uses the KISS ICP for scan point cloud combination and the RANSAC method for the initial alignment of the combined point cloud. The mobile mapping point cloud using Riegl VMX-250 serves as the reference map, and Velodyne scans are used for localization purposes. The point-to-plane iterative closest-point method is then employed to refine the alignment. The paper evaluates the efficacy of the proposed method by calculating the errors between the estimated and ground truth positions. The results indicate that the voxel-based approach is capable of accurately estimating the position of the sensor platform, which are applicable for various use cases. A specific use case in the context is smart parking space management, which is described and initial visualization results are shown.

TARGETLESS REGISTRATION METHODS BETWEEN UAV LIDAR AND WEARABLE MMS POINT CLOUDS

Abstract. Fixed-wing Unmanned Aerial Vehicles (UAV) and wearable or portable Mobile Mapping Systems (MMS) are two widely used platforms for point cloud acquisition with Light Detection And Ranging (LiDAR) sensors. The two platforms acquire from distant viewpoints and produce complementary point clouds, one describing predominantly horizontal surfaces and the other primarily vertical. Thus, the registration of the two data is not straightforward. This paper presents a test of targetless registration between a UAV LiDAR point cloud and terrestrial MMS surveys. The case study is a vegetated hilly landscape characterized by the presence of a structure of interest; the UAV acquisition allows the entire area to be acquired from above, while the terrestrial MMS acquisitions will enable the construction of interest to be detailed. The paper describes the survey phase with both techniques. It focuses on processing and registration strategies to fuse the two data together.Our approach is based on the ICP (Iterative Closest Point) method by exploiting the data processing algorithms available in the Heron Desktop post-processing software for handling data acquired with the Heron Backpack MMS instrument. Two co-registration methods are compared. Both ways use the UAV point cloud as a reference and derive the registration of the terrestrial MMS data by finding ICP matches between the ground acquisition and the reference cloud exploiting only a few areas of overlap. The two methods are detailed in the paper, and both allow us to complete the co-registration task.

Long-term monitoring method for tunnel structure transformation using a 3D light detection and ranging equipped in a mobile robot

This study proposes a long-term health monitoring method for tunnel structure transformation using a 3D light detection and ranging (LiDAR) equipped in a mobile robot. The proposed method comprised two phases: 3D point cloud maps (PCMs) generation and health indicators (HIs) extraction at the same location for the long-term structure transformation inspection. First, the 3D point cloud data (PCD) and six-degree-of-freedom acceleration and gyration of a mobile robot were measured during periodic inspection of tunnel structures using the 3D LiDAR and an inertial measurement unit. This information was combined to generate a 3D PCM by utilizing tightly coupled LiDAR inertial odometry. This procedure could be repeated to inspect tunnel structure transformation at patrol inspection in a long term. 3D PCMs measured during period inspection were then registered to extract a 3D PCD at the location of the first measurement. Hence, an iterative closest point method with edge features was executed to register periodic PCD measurements. Second, 3D PCD in predefined locations were extracted at every measurement. These PCD were used to extract features that are highly correlated to the inner wall and ground of the tunnel structure to obtain five HIs representing their health conditions. These HIs were then employed to evaluate the structural health of the tunnel. Field experiments were conducted to demonstrate the effectiveness of the proposed method in buildings and tunnels. The proposed method outperformed other 3D PCM methods, especially in structures with few features such as tunnels and corridors. A systematic analysis of the experimental results also revealed that the proposed method ensures the accuracy and robustness of HIs extraction at the same location, confirming the feasibility of long-term inspection. The proposed inspection method for tunnel structures and embedded mobile robot system provides a cost-effective and autonomous solution for infrastructure health monitoring.

A Symmetry-Based Superposition Method for Planning and Surgical Outcome Assessment

Computer-aided surgical planning has been widely used to increase the safety and predictability of surgery. The validation of the target of surgical planning to surgical outcomes on a patient-specific model is an important issue. The aim of this research was to develop a robust superposition method to assess the deviation of planning and outcome by using the symmetrical characteristic of the affected target. The optimal symmetry plane (OSP) of an object is usually used to evaluate the degree of symmetry of an object. We proposed a refined OSP-based contouring method to transfer a complex three-dimensional superposition operation into two dimensions. We compared the typical iterative closest point (ICP) algorithm with the refined OSP-based contouring method and examined the differences between them. The results using the OSP-based method were much better than the traditional method. As for processing time, the OSP-based contouring method was 11 times faster than the ICP method overall. The proposed method was not affected by the metallic artifacts from medical imaging or geometric changes due to surgical intervention. This technique can be applied for post-operative assessment, such as quantifying the differences between surgical targets and outcomes as well as performing long-term medical follow-up.

Adaptive weighted robust iterative closest point

The Iterative Closest Point (ICP) algorithm is one of the most important methods for rigid registration between point sets. However, its performance begins to degenerate with the point data are overly contaminated by noise, outliers, and missing data. In this paper, we propose an adaptive weighted robust ICP method (AW-RICP). A sparse weight vector can be automatically learned by adaptive neighbors assigning process. The sparseness of the weight vector can be achieved by fitting the number of selected samples. The adaptive weight vector leads to the robustness of AW-RICP by eliminating the negative effect caused by point cloud pairs with largest registration errors. Further, the retained point pairs are assigned suitable weights to suppress the noises and outliers. The new error metric can effectively improve the robustness of ICP method. Additionally, we propose an efficient iterative solution to optimize our problem. Experiments on both synthetic and real data sets show that the proposed method can achieve superior performance with other state-of-the-art methods.

Fast and Accurate Registration of Terrestrial Point Clouds Using a Planar Approximation of Roof Features

3D reconstructed models are becoming more diffused daily, especially in the Cultural Heritage field. These geometric models are typically obtained from elaborating a 3D point cloud. A significant limit in using these methods is the realignment of different point clouds acquired from different acquisitions, particularly for those whose dimensions are millions of points. Although several methodologies have tried to propose a solution for this necessity, none of these seems to solve definitively the problems related to the realignment of large point clouds. This paper presents a new and innovative procedure for the fine registration of large point clouds. The method performs an alignment by using planar approximations of roof features, taking the roof’s extension into account. It looks particularly suitable for the alignment of large point clouds acquired in urban and archaeological environments. The proposed methodology is compared in terms of accuracy and time with a standard photogrammetric reconstruction based on Ground Control Points (GCPs) and other ones, aligned by the Iterative Closest Point method (ICP) and markers. The results evidence the excellent performance of the methodology, which could represent an alternative for aligning extensive photogrammetric reconstructions without the use of GCPs.

Stein ICP for Uncertainty Estimation in Point Cloud Matching

Quantification of uncertainty in point cloud matching is critical in many tasks such as pose estimation, sensor fusion, and grasping. Iterative closest point (ICP) is a commonly used pose estimation algorithm which provides a point estimate of the transformation between two point clouds. There are many sources of uncertainty in this process that may arise due to sensor noise, ambiguous environment, initial condition, and occlusion. However, for safety critical problems such as autonomous driving, a point estimate of the pose transformation is not sufficient as it does not provide information about the multiple solutions. Current probabilistic ICP methods usually do not capture all sources of uncertainty and may provide unreliable transformation estimates which can have a detrimental effect in state estimation or decision making tasks that use this information. In this work we propose a new algorithm to align two point clouds that can precisely estimate the uncertainty of ICP’s transformation parameters. We develop a Stein variational inference framework with gradient based optimization of ICP’s cost function. The method provides a non-parametric estimate of the transformation, can model complex multi-modal distributions, and can be effectively parallelized on a GPU. Experiments using 3D kinect data as well as sparse indoor/outdoor LiDAR data show that our method is capable of efficiently producing accurate pose uncertainty estimates.

Registration method for the analysis of mooring chain links degraded by corrosion and wear

The observation of the three-dimensional geometrical evolution of mooring chain links exposed to the marine environment can be achieved by using 3D scanners. Successive surveys make it possible to precisely identify the location, shapes and quantities of material loss by corrosion or fretting in interlink contact areas. However, the comparison is only possible after an alignment of the obtained scans, because they are not expressed in the same reference marks. The usual point cloud alignment algorithms not being adapted to the expected accuracy, the carried out work implements a new algorithm based on the iterative closest point method (ICP). The developed iterative process of alignment, moves the point cloud corresponding to the aged link onto the new link, after detecting and deleting the points located in the areas of localized material losses (pitting corrosion or wear), and after adding an extra layer on the remaining surfaces, of a value corresponding to the loss of material by a generalised corrosion assumed to be constant. The tests carried out on perfect meshes from CAD constructions show a convergence of the alignment process towards a sufficient precision with regard to the work objectives.

Accuracy evaluation of surface registration algorithm using normal distribution transform in stereotactic body radiotherapy/radiosurgery: A phantom study

PurposeTo evaluate a feasibility of normal distribution transform (NDT) algorithm compared with the iterative closest point (ICP) method as a useful surface registration in stereotactic body radiotherapy (SBRT)/stereotactic radiosurgery (SRS).MethodsPoint cloud images using the 3D triangulation technology were obtained from a depth camera‐based optical imaging (OSI) system equipped in a radiosurgery room. Two surface registration algorithms, NDT and ICP, were used to measure and compare the discrepancy values between the reference and the current surfaces during the positioning of the patient. The performance evaluation was investigated by calculating the registration error and root‐mean‐square (RMS) values for the surface model, reposition, and target accuracy, which were analyzed statistically using a paired t‐test.ResultsFor surface model accuracy, the average of the registration error and RMS values were measured as 3.56 ± 2.20 mm and 6.98 ± 1.89 mm for ICP method, and 1.76 ± 1.32 mm and 3.58 ± 1.30 mm for NDT method (p < 0.05). For reposition accuracy, the average registration error and RMS values were calculated as 1.41 ± 0.98 mm and 2.53 ± 1.64 mm using ICP method, and 0.92 ± 0.61 mm and 1.75 ± 0.80 mm using NDT method (p = 0.005). The overall target accuracy using the NDT method reduced the average of the reposition error and overall RMS value by 0.71 and 1.32 mm, respectively, compared to the ICP method (p = 0.03).ConclusionsWe found that the surface registration algorithm based on NDT method provides more reliable accuracy in the values of surface model, reposition, and target accuracies than the classic ICP method. The NDT method in OSI systems offers reasonable accuracy in SBRT/SRS.

Robust affine registration method using line/surface normals and correntropy criterion

The problem of matching point clouds is an efficient way of registration, which is significant for many research fields including computer vision, machine learning, and robotics. There may be linear or non-linear transformation between point clouds, but determining the affine relation is more challenging among linear cases. Various methods have been presented to overcome this problem in the literature and one of them is the affine variant of the iterative closest point (ICP) algorithm. However, traditional affine ICP variants are highly sensitive to effects such as noises, deformations, and outliers; the least-square metric is substituted with the correntropy criterion to increase the robustness of ICPs to such effects. Correntropy-based robust affine ICPs available in the literature use point-to-point metric to estimate transformation between point clouds. Conversely, in this study, a line/surface normal that examines point-to-curve or point-to-plane distances is employed together with the correntropy criterion for affine point cloud registration problems. First, the maximum correntropy criterion measure is built for line/surface normal conditions. Then, the closed-form solution that maximizes the similarity between point sets is achieved for 2D registration and extended for 3D registration. Finally, the application procedure of the developed robust affine ICP method is given and its registration performance is examined through extensive experiments on 2D and 3D point sets. The results achieved highlight that our method can align point clouds more robustly and precisely than the state-of-the-art methods in the literature, while the registration time of the process remains at reasonable levels.

Memory-Augmented Point Cloud Registration Network for Bucket Pose Estimation of the Intelligent Mining Excavator

Bucket pose estimation is a key technology for the intelligent mining excavator to realize automatic excavation and loading. The existing estimation methods mainly include two categories. The sensor measurement methods install sensors to detect the bucket pose, which suffer from large cumulative error and short service life caused by strong vibrations. The traditional registration methods estimate the bucket pose by registering its point clouds with the iterative closest point method, which easily fall into a local optimum. Recently, point cloud registration networks have been studied for object pose estimation based on balanced training samples. However, since the bucket speed changes obviously during excavation, the number of bucket samples at different poses is imbalanced. As a result, point cloud registration networks will easily forget the bucket features at some poses which have small samples, also called nondominant pose features, during learning. This will reduce the accuracy of bucket pose estimation. To handle the forgetting problem caused by the sample imbalance, we propose a novel memory-augmented registration network (MARNet) for bucket pose estimation. The MARNet consists of an encoder, a dual-pose-memory module (DPMM), and a decoder. The encoder extracts the global pose features from the input point clouds. The DPMM learns and memorizes the prototypical pose features covering all the samples by using the memory units at the training phase. Then, nondominant pose features are augmented by retrieving the stored prototypical pose features at the testing phase. At last, the decoder calculates the final poses. Our method can effectively alleviate the forgetting problem and strengthen the generalization by using the DPMM. The experimental results demonstrate that our method increases the estimation accuracy of rotation and translation by at least 1.49° and 0.021 m, respectively, for the bucket poses with small samples. This contributes to the development of intelligent mining excavators.

A lidar-only SLAM algorithm for marine vessels and autonomous surface vehicles

Research into autonomous surface vehicles is noticeably limited in regards to the functionality of the vehicles themselves. Specifically, testing and evaluation typically occurs at speeds considerably lower than what is allowed in an operational setting. For a vessel to be able to take advantage of higher speeds, there must be a robust and tested method for determining localisation and navigation. With an emphasis of development for small vessels with higher impulse capabilities, working in confined and restricted environments, the decision was made to develop a method of navigation that relied solely upon lightweight sensors. For this, a single light ranging sensor was utilised to develop both simultaneous localisation and mapping for the vessel, using the normal distribution transform and iterative closest point methods. Evaluation of the algorithm accuracy as the vessel moved above speeds greater than two metres per second was conducted, and it was feasibly evaluated that there was no observable drift of mapping in horizontal planes, however, there was a accumulated drift in the vertical plane and a transient response in localisation deviation as the vessel changed impulse through the two metre per second window.