Search

In the process of autonomous navigation of outdoor mobile robots, four modules are involved: perception, localization, planning, and controlling. The perception module utilizes sensors such as cameras and radars based on various principles to analyze the robot's surroundings in real-time. The localization module uses GPS (Global Positioning System), IMU (Inertial Measurement Unit), and prior maps for real-time positioning analysis. The planning module plans optimal path based on the outputs of the first two modules, and the controlling module directs the robot's chassis to move along the planned optimal path. For the localization module, the accuracy of GPS positioning results heavily depends on weather conditions and GPS signal receptions. Even if the positioning results of imu are integrated, the positioning accuracy still cannot meet the needs of robot navigation. Therefore, using prior maps for repositioning can compensate for this accuracy deficiency. The planning module also requires path planning based on prior maps. If the a priori map storage is large, it will lead to difficulties in usage, maintenance, and updates. Therefore, it is crucial to research lightweight navigation map mapping methods. In this paper, an automatically mapping method of lightweight navigation maps is proposed, combining cameras and LiDAR (Light Detection and Ranging), including semantic informations necessary for outdoor navigation positioning, such as pole-like objects and traffic signs for robot longitudinal positioning, and lane line elements for robot lateral positioning. This method automatic generates robot navigation maps in Lanelet2 format, providing support for subsequent positioning and path planning modules.

The calibration of the camera and LiDAR is one of the basis for the construction of the multi-sensor fusion mapping system. Planar features of walls and grounds in indoor environments provides effective constraints for multi-sensor calibration. In this paper, we proposed a new camera-LiDAR calibration method with the constraint of indoor spatial structure. Using the image and point cloud data collected by sensors, visual odometry and LiDAR odometry can be constructed to calculate the transformation between sensors. Based on visual odometry and LiDAR odometry, structural parameters in indoor environment are extracted from images and point cloud to constrain rotation estimation between sensors. In the method proposed in this paper, lines are extracted from the images acquired by the camera and used to estimate and track vanishing points. The direction estimated with vanishing points is used as a global constraint to optimize the rotation parameter estimation of the camera. The fitted planes from the point cloud acquired by the LiDAR are used to compute a set of orthogonal normal vectors corresponding to the ground and wall surfaces, which are used as global constraints to optimize the rotation parameter estimation of the LiDAR. The calibration method proposed in this paper is targetless and only constrained by the indoor spatial structure. The result shows that the proposed indoor spatial structure constraint calibration method can calibrate LiDAR and camera without generating cumulative errors during the rotation estimation process.

Indoor scene recognition remains a challenging problem in the fields of artificial intelligence and computer vision due to the complexity, similarity, and spatial variability of indoor scenes. The existing research is mainly based on 2D data, which lacks 3D information about the scene and cannot accurately identify scenes with a high frequency of changes in lighting, shading, layout, etc. Moreover, the existing research usually focuses on the global features of the scene, which cannot represent indoor scenes with cluttered objects and complex spatial layouts. To solve the above problems, this paper proposes a dual-task model for indoor scene recognition and semantic segmentation based on point cloud data. The model expands the data loading method by giving the dataset loader the ability to return multi-dimensional labels and then realizes the dual-task model of scene recognition and semantic segmentation by fine-tuning PointNet++, setting task state control parameters, and adding a common feature layer. Finally, in order to solve the problem that the similarity of indoor scenes leads to the wrong scene recognition results, the rules of scenes and elements are constructed to correct the scene recognition results. The experimental results showed that with the assistance of scene-element rules, the overall accuracy of scene recognition with the proposed method in this paper is 82.4%, and the overall accuracy of semantic segmentation is 98.9%, which is better than the comparison model in this paper and provides a new method for cognition of indoor scenes based on 3D point clouds.

Airborne Light Detection And Ranging (LiDAR) point clouds and images data fusion have been widely studied. However, with recent developments in photogrammetric technology, images can now provide dense image matching (DIM) point clouds. To make use of such DIM points, a sample selection framework is introduced. That is, first, the geometric features of LiDAR points and DIM points are extracted. Each feature per point is considered a sample. Then we extend the binary TrAdaboost classifier into a multi-class one to train all the samples. The classifier automatically assigns weights to the samples in the DIM points. The useful samples are assigned large weights and consequently impact the classification results largely and vice versa. As a result, the useful samples of the DIM points are kept to improve on the LiDAR points classification performance. Because only the samples are used, the registration between the DIM points and LiDAR points is not needed. Moreover, the DIM points capturing similar classes but not the same scene as the LiDAR points can also be used. By our framework, existing aerial images can be fully utilized. For testing the generation ability, the framework is applied in a super-voxel-based classification approach by replacing the points-based features with the super-voxel-based features. In the experiments, whether DIM points at the same places as those of LiDAR are used or not, the results after fusion show that, the LiDAR points classification performance has improved. Also, the better the quality of DIM points are, the better the classification performance achieves.

Automated 3D reconstruction of indoor environments from point clouds has been a topic of intensive research in recent years. Different methods developed for the generation of 3D indoor models have achieved promising results on different case studies. However, a comprehensive evaluation and comparison of the performance of these methods has not been available. This paper presents the preliminary results of the ISPRS benchmark on indoor modelling, an initiative of Working Group IV/5 to benchmark the performance of indoor modelling methods using a public dataset and a comprehensive evaluation framework. The performances of the different methods are compared through geometric quality evaluation of the reconstructed models in terms of completeness, correctness, and accuracy of wall elements. The results show that the reconstruction methods generally achieve high completeness but lower correctness for the reconstructed models while accuracies range from 0.5 cm to 6.7 cm.

Visual Simultaneous Localization and Mapping (SLAM) systems have been widely investigated in response to requirements, since the traditional positioning technology, such as Global Navigation Satellite System (GNSS), cannot accomplish tasks in restricted environments. However, traditional SLAM methods which are mostly based on point feature tracking, usually fail in harsh environments. Previous works have proven that insufficient feature points caused by missing textures, feature mismatches caused by too fast camera movements, and abrupt illumination changes will eventually cause state estimation to fail. And meanwhile, pedestrians are unavoidable, which introduces fake feature associations, thus violating the strict assumption that the unknown environment is static in SLAM. In order to ensure how our system copes with the huge challenges brought by these factors in a complex indoor environment, this paper proposes a semantic-assisted Visual Inertial Odometer (VIO) system towards low-textured scenes and highly dynamic environments. The trained U-net will be used to detect moving objects. Then all feature points in the dynamic object area need to be eliminated, so as to avoid moving objects to participate in the pose solution process and improve robustness in dynamic environments. Finally, the constraints of inertial measurement unit (IMU) are added for low-textured environments. To evaluate the performance of the proposed method, experiments were conducted on the EuRoC and TUM public dataset, and the results demonstrate that the performance of our approach is robust in complex indoor environments.

In this paper, we present a publicly available benchmark dataset on multisensorial indoor mapping and positioning (MiMAP), which is sponsored by ISPRS scientific initiatives. The benchmark dataset includes point clouds captured by an indoor mobile laser scanning system in indoor environments of various complexity. The benchmark aims to stimulate and promote research in the following three fields: (1) LiDAR-based Simultaneous Localization and Mapping (SLAM); (2) automated Building Information Model (BIM) feature extraction; and (3) multisensory indoor positioning. The MiMAP project provides a common framework for the evaluation and comparison of LiDAR-based SLAM, BIM feature extraction, and smartphone-based indoor positioning methods. This paper describes the multisensory setup, data acquisition process, data description, challenges, and evaluation metrics included in the MiMAP project.

Route planning and navigation in indoor space have become a hot topic recently. To accomplish this task, a map and a real-time detection system are needed. Due to Lidar systems’ high efficiency in data acquisition, Lidar sensors have become an indispensable part of an object detection system. In this paper, we use Lidar points to generate obstacle maps. The obstacle maps can be used as a reference for route planning and navigation. To identify single objects more precisely, a deep network combined PointNet with Markov Random Field (MRF) is designed in our work to classify Lidar points. Then, single objects are segmented by using the Euclidean clustering method. After that, the prior rules and derived criteria we summarized from large amount images are used to determine objects’ kind between Influence Movement Obstacles (IMO) and Non-Influence Movement Obstacles (N-IMO). Finally, objects are projected into a 2D plane to generate obstacle maps. To evaluate the performance of our method, experiments were performed on the S3DIS dataset of Stanford University. The results show that our method greatly improves the overall accuracy compared to the original PointNet model, and can generate high-quality obstacle maps.

Rocks are one of the major Martian surface features and yield significant information about the relevant geology process and the life exploration. However, autonomous Martian rock detection is still a challenging task due to the appearance similar to the background, the view and illumination change. Therefore, this paper presents a gradient-region constrained level set method based on mars rover image for automatic Martian rock extraction. In our method, the evolution function of level set consists of the internal energy term for guaranteeing the deviation of the level set function from a signed distance function and the external energy term, where the gradient-based information is integrated with the locally adaptive region-based information, for robustly driving the motion of the zero-level set toward the object boundaries even in images with ununiform grey scale. The resulting evolution of the level set function is based on the minimisation of the overall energy functional using the standard gradient descent method. As a result, those detected Martian surface regions that are most likely to yield valuable scientific discoveries will be further analysed based on two-dimensional shape characterisation. To evaluate the performance of the proposed method, experiments were performed on mars rover image under various terrain and illumination conditions. Results demonstrate that the proposed method is robust and efficient for automatically detecting both small-scale and large-scale rocks on Martian surfaces.

Since Global Navigation Satellite System may be unavailable in complex dynamic environments, visual SLAM systems have gained importance in robotics and its applications in recent years. The SLAM system based on point feature tracking shows strong robustness in many scenarios. Nevertheless, point features over images might be limited in quantity or not well distributed in low-textured scenes, which makes the behaviour of these approaches deteriorate. Compared with point features, line features as higher-dimensional features can provide more environmental information in complex scenes. As a matter of fact, line segments are usually sufficient in any human-made environment, which suggests that scene characteristics remarkably affect the performance of point-line feature based visual SLAM systems. Therefore, this paper develops a scene-assisted point-line feature based visual SLAM method for autonomous flight in unknown indoor environments. First, ORB point features and Line Segment Detector (LSD)-based line features are extracted and matched respectively to build two types of projection models. Second, in order to effectively combine point and line features, a Convolutional Neural Network (CNN)-based model is pre-trained based on the scene characteristics for weighting their associated projection errors. Finally, camera motion is estimated through non-linear minimization of the weighted projection errors between the correspondent observed features and those projected from previous frames. To evaluate the performance of the proposed method, experiments were conducted on the public EuRoc dataset. Experimental results indicate that the proposed method outperforms the conventional point-line feature based visual SLAM method in localization accuracy, especially in low-textured scenes.

This paper presents the design of the benchmark dataset on multisensory indoor mapping and position (MIMAP) which is sponsored by ISPRS scientific initiatives. The benchmark dataset including point clouds captured by indoor mobile laser scanning system (IMLS) in indoor environments of various complexity. The benchmark aims to stimulate and promote research in the following three fields: (1) SLAM-based indoor point cloud generation; (2) automated BIM feature extraction from point clouds, with an emphasis on the elements, such as floors, walls, ceilings, doors, windows, stairs, lamps, switches, air outlets, that are involved in building management and navigation tasks ; and (3) low-cost multisensory indoor positioning, focusing on the smartphone platform solution. MIMAP provides a common framework for the evaluation and comparison of LiDAR-based SLAM, BIM feature extraction, and smartphone indoor positioning methods.

Indoor scenes have the characteristics of abundant semantic categories, illumination changes, occlusions and overlaps among objects, which poses great challenges for indoor semantic segmentation. Therefore, we in this paper develop a method based on higher-order Markov random field model for indoor semantic segmentation from RGB-D images. Instead of directly using RGB-D images, we first train and perform RefineNet model only using RGB information for generating the high-level semantic information. Then, the spatial location relationship from depth channel and the spectral information from color channels are integrated as a prior for a marker-controlled watershed algorithm to obtain the robust and accurate visual homogenous regions. Finally, higher-order Markov random field model encodes the short-range context among the adjacent pixels and the long-range context within each visual homogenous region for refining the semantic segmentations. To evaluate the effectiveness and robustness of the proposed method, experiments were conducted on the public SUN RGB-D dataset. Experimental results indicate that compared with using RGB information alone, the proposed method remarkably improves the semantic segmentation results, especially at object boundaries.

Despite recent progress in the development of methods for automated reconstruction of indoor models, a comparative performance evaluation of these methods is not available due to the lack of publicly available benchmark datasets and a common evaluation framework. The ISPRS Benchmark on Indoor Modelling is an effort to enable comparison and benchmarking of indoor modelling methods by providing a benchmark dataset and a comprehensive evaluation framework. In this paper, we propose a framework for the evaluation of indoor modelling methods, and discuss various quality aspects of the reconstruction methods as well as the reconstructed models. We discuss the challenges in quantitative quality evaluation of indoor models through comparison with a reference model, and propose suitable measures and methods for comparing an automatically reconstructed indoor model with a reference.

Automated generation of 3D indoor models from point cloud data has been a topic of intensive research in recent years. While results on various datasets have been reported in literature, a comparison of the performance of different methods has not been possible due to the lack of benchmark datasets and a common evaluation framework. The ISPRS benchmark on indoor modelling aims to address this issue by providing a public benchmark dataset and an evaluation framework for performance comparison of indoor modelling methods. In this paper, we present the benchmark dataset comprising several point clouds of indoor environments captured by different sensors. We also discuss the evaluation and comparison of indoor modelling methods based on manually created reference models and appropriate quality evaluation criteria. The benchmark dataset is available for download at: http://www2.isprs.org/commissions/comm4/wg5/benchmark-on-indoor-modelling.html.

With the rapid development of LBS (Location-based Service), the demand for commercialization of indoor location has been increasing, but its technology is not perfect. Currently, the accuracy of indoor location, the complexity of the algorithm, and the cost of positioning are hard to be simultaneously considered and it is still restricting the determination and application of mainstream positioning technology. Therefore, this paper proposes a method of knowledge-based optimization of indoor location based on low energy Bluetooth. The main steps include: 1) The establishment and application of a priori and posterior knowledge base. 2) Primary selection of signal source. 3) Elimination of positioning gross error. 4) Accumulation of positioning knowledge. The experimental results show that the proposed algorithm can eliminate the signal source of outliers and improve the accuracy of single point positioning in the simulation data. The proposed scheme is a dynamic knowledge accumulation rather than a single positioning process. The scheme adopts cheap equipment and provides a new idea for the theory and method of indoor positioning. Moreover, the performance of the high accuracy positioning results in the simulation data shows that the scheme has a certain application value in the commercial promotion.

Based on the frame of ORB-SLAM in this paper the transformation parameters between adjacent Kinect image frames are computed using ORB keypoints, from which priori information matrix and information vector are calculated. The motion update of multi-feature extended information filter is then realized. According to the point cloud data formed by depth image, ICP algorithm was used to extract the point features of the point cloud data in the scene and built an observation model while calculating a-posteriori information matrix and information vector, and weakening the influences caused by the error accumulation in the positioning process. Furthermore, this paper applied ORB-SLAM frame to realize autonomous positioning in real time in interior unknown environment. In the end, Lidar was used to get data in the scene in order to estimate positioning accuracy put forward in this paper.

With increasing attention for the indoor environment and the development of low-cost RGB-D sensors, indoor RGB-D images are easily acquired. However, scene semantic segmentation is still an open area, which restricts indoor applications. The depth information can help to distinguish the regions which are difficult to be segmented out from the RGB images with similar color or texture in the indoor scenes. How to utilize the depth information is the key problem of semantic segmentation for RGB-D images. In this paper, we propose an Encode-Decoder Fully Convolutional Networks for RGB-D image classification. We use Multiple Kernel Maximum Mean Discrepancy (MK-MMD) as a distance measure to find common and special features of RGB and D images in the network to enhance performance of classification automatically. To explore better methods of applying MMD, we designed two strategies; the first calculates MMD for each feature map, and the other calculates MMD for whole batch features. Based on the result of classification, we use the full connect CRFs for the semantic segmentation. The experimental results show that our method can achieve a good performance on indoor RGB-D image semantic segmentation.

This paper realizes automatically the navigating elements defined by indoorGML data standard – door, stairway and wall. The data used is indoor 3D point cloud collected by Kinect v2 launched in 2011 through the means of ORB-SLAM. By contrast, it is cheaper and more convenient than lidar, but the point clouds also have the problem of noise, registration error and large data volume. Hence, we adopt a shape descriptor – histogram of distances between two randomly chosen points, proposed by Osada and merges with other descriptor – in conjunction with random forest classifier to recognize the navigation elements (door, stairway and wall) from Kinect point clouds. This research acquires navigation elements and their 3-d location information from each single data frame through segmentation of point clouds, boundary extraction, feature calculation and classification. Finally, this paper utilizes the acquired navigation elements and their information to generate the state data of the indoor navigation module automatically. The experimental results demonstrate a high recognition accuracy of the proposed method.

The lunar terrain can show its collision and volcanic history. The lunar surface roughness can give a deep indication of the effects of lunar surface magma, sedimentation and uplift. This paper aims to get different information from the roughness through different data sources. Besides introducing the classical Root-mean-square height method and Morphological Surface Roughness (MSR) algorithm, this paper takes the area of the Jurassic mountain uplift in the Sinus Iridum and the Plato Crater area as experimental areas. And then make the comparison and contrast of the lunar roughness derived from LRO's DEM and CE-2 DOM. The experimental results show that the roughness obtained by the traditional roughness calculation method reflect the ups and downs of the topography, while the results obtained by morphological surface roughness algorithm show the smoothness of the lunar surface. So, we can first use the surface fluctuation situation derived from RMSH to select the landing area range which ensures the lands are gentle. Then the morphological results determine whether the landing area is suitable for the detector walking and observing. The results obtained at two different scales provide a more complete evaluation system for selecting the landing site of the lunar probe.

In the detection of the moon, the visible and near-infrared reflectance data of the lunar material are important information sources for lunar chemical substances and mineral inversion. The Interferometer Imaging Spectrometer (IIM) aboard the Chang'E-1 lunar orbiter is the first multispectral imaging spectrometer for Chinese lunar missions. In this paper, we use the mosaic image of global moon acquired by the Wide-angle Camera (WAC) of the Lunar Reconnaissance Orbiter Camera (LROC) to realize the accurate registration of Chang'E-1 IIM hyperspectral images. Due to the lack of GCPs, the emphasis of this work is to find a huge number of homologous points. The method proposed in this paper is to obtain several homologous points by manually matching, and then we utilize those points to calculate the initial homography matrix of LROC-WAC image and IIM image. This matrix is used to predict the area on IIM image where homologous points may be located, and the locations of the homologous points are determined by the orientation correlation in frequency domain. Finally we save the parts of homologous points which satisfied the conversion relationship of initial homography matrix to calculate homography matrix again. We use this iterative way to obtain a more accurate location of the homologous points. In this process, we take into account that the geometric deformations of different regions on IIM image are quite different. Therefore, we added image threshold segmentation based on the initial homography matrix in the experiment, and completed the above work of finding the homologous points on the segmented images. The final realization of registration accuracy of IIM images are in 1–2 pixels (RMSE). This provides a reliable data assurance for the subsequent study of using IIM images to inverse the lunar elements.

Impact craters are among the most noticeable geo-morphological features on the planetary surface and yield significant information on terrain evolution and the history of the solar system. Thus, the recognition of lunar impact craters is an important branch of modern planetary studies. To address problems associated with the insufficient and inaccurate detection of lunar impact craters, this paper extends the strategy that integrates multi-source data and proposes a Bayesian Network (BN) framework for the automatic recognition of impact craters that is based on CCD stereo camera images and associated Digital Elevation Model (DEM) data. The method uses the SVM model to fit the probability distribution of the impact craters in the feature space. SVM model, whose output is used as the intermediate posterior probability, is embedded in the Bayesian network as a node, and the final posterior probability is obtained by integration under the Bayesian network. We validated our proposed framework with both CCD stereo camera images acquired by the Chang’e-2 satellite and DEM data acquired by Lunar Reconnaissance Orbiter (LRO). Experimental results demonstrate that the proposed framework can provide a very high level of accuracy in the recognition phase. Moreover, the results showed a significant improvement in the detection rate, particularly for the detection of sub-kilometer craters, compared with previous approaches.

Crater-detection approaches can be divided into four categories: manual recognition, shape-profile fitting algorithms, machine-learning methods and geological information-based analysis using terrain and spectral data. The mainstream method is Shape-profile fitting algorithms. Many scholars throughout the world use the illumination gradient information to fit standard circles by least square method. Although this method has achieved good results, it is difficult to identify the craters with poor "visibility", complex structure and composition. Moreover, the accuracy of recognition is difficult to be improved due to the multiple solutions and noise interference. Aiming at the problem, we propose a method for the automatic extraction of impact craters based on spectral characteristics of the moon rocks and minerals: 1) Under the condition of sunlight, the impact craters are extracted from MI by condition matching and the positions as well as diameters of the craters are obtained. 2) Regolith is spilled while lunar is impacted and one of the elements of lunar regolith is iron. Therefore, incorrectly extracted impact craters can be removed by judging whether the crater contains "non iron" element. 3) Craters which are extracted correctly, are divided into two types: simple type and complex type according to their diameters. 4) Get the information of titanium and match the titanium distribution of the complex craters with normal distribution curve, then calculate the goodness of fit and set the threshold. The complex craters can be divided into two types: normal distribution curve type of titanium and non normal distribution curve type of titanium. We validated our proposed method with MI acquired by SELENE. Experimental results demonstrate that the proposed method has good performance in the test area.

Because the Infra-Red (IR) Kinect sensor only provides accurate depths up to 5 m for a limited field of view (60°), the problem of registration error accumulation becomes inevitable in indoor mapping. Therefore, in this paper, a global registration method is proposed based on augmented extended Information Filter (AEIF). The point cloud registration is regarded as a stochastic system so that AEIF is used to produces the accurate estimates of rigid transformation parameters through eliminating the error accumulation suffered by the pair-wise registration. Moreover, because the indoor scene normally contains planar primitives, they can be employed to control the registration of multiple scans. Therefore, the planar primitives are first fitted based on optimized BaySAC algorithm and simplification algorithm preserving the feature points. Besides the constraint of corresponding points, we then derive the plane normal vector constraint as an additional observation model of AEIF to optimize the registration parameters between each pair of adjacent scans. The proposed approach is tested on point clouds acquired by a Kinect camera from an indoor environment. The experimental results show that our proposed algorithm is proven to be capable of improving the accuracy of multiple scans aligning by 90%.

This paper presents an algorithm for the automatic registration of terrestrial point clouds by match selection using an efficiently conditional sampling method -- threshold-independent BaySAC (BAYes SAmpling Consensus) and employs the error metric of average point-to-surface residual to reduce the random measurement error and then approach the real registration error. BaySAC and other basic sampling algorithms usually need to artificially determine a threshold by which inlier points are identified, which leads to a threshold-dependent verification process. Therefore, we applied the LMedS method to construct the cost function that is used to determine the optimum model to reduce the influence of human factors and improve the robustness of the model estimate. Point-to-point and point-to-surface error metrics are most commonly used. However, point-to-point error in general consists of at least two components, random measurement error and systematic error as a result of a remaining error in the found rigid body transformation. Thus we employ the measure of the average point-to-surface residual to evaluate the registration accuracy. The proposed approaches, together with a traditional RANSAC approach, are tested on four data sets acquired by three different scanners in terms of their computational efficiency and quality of the final registration. The registration results show the st.dev of the average point-to-surface residuals is reduced from 1.4 cm (plain RANSAC) to 0.5 cm (threshold-independent BaySAC). The results also show that, compared to the performance of RANSAC, our BaySAC strategies lead to less iterations and cheaper computational cost when the hypothesis set is contaminated with more outliers.

The research project ‘Moon Mapping’ has been established in 2014 between the Italian and Chinese Governments to promote cooperation and exchange between undergraduate students from both countries. The operational phase of the project started in early 2015, and will end in 2017, for a total length of three years. The main aim is to train new scholars to be able to work on different kinds of remotely-sensed data collected over the Moon surface by the Chinese space missions Chang’E-1/2. The project coordination has been assigned to the Italian Space Agency for the Italian side and to the Center of Space Exploration, China Ministry of Education, for the Chinese side. Several Chinese universities and Italian national research institutes and universities have been officially involved in this project. Six main research topics have been identified: (1) map of the solar wind ion; (2) geomorphological map of the Moon; (3) data preprocessing of Chang’E-1 mission; (4) map of element distribution; (5) establishment of 3D digital visualization system; and (6) compilation and publication of a tutorial on joint lunar mapping.

The deformation monitoring of subway tunnel is of extraordinary necessity. Therefore, a method for deformation monitoring based on terrestrial point clouds is proposed in this paper. First, the traditional adjacent stations registration is replaced by sectioncontrolled registration, so that the common control points can be used by each station and thus the error accumulation avoided within a section. Afterwards, the central axis of the subway tunnel is determined through RANSAC (Random Sample Consensus) algorithm and curve fitting. Although with very high resolution, laser points are still discrete and thus the vertical section is computed via the quadric fitting of the vicinity of interest, instead of the fitting of the whole model of a subway tunnel, which is determined by the intersection line rotated about the central axis of tunnel within a vertical plane. The extraction of the vertical section is then optimized using RANSAC for the purpose of filtering out noises. Based on the extracted vertical sections, the volume of tunnel deformation is estimated by the comparison between vertical sections extracted at the same position from different epochs of point clouds. Furthermore, the continuously extracted vertical sections are deployed to evaluate the convergent tendency of the tunnel. The proposed algorithms are verified using real datasets in terms of accuracy and computation efficiency. The experimental result of fitting accuracy analysis shows the maximum deviation between interpolated point and real point is 1.5 mm, and the minimum one is 0.1 mm; the convergent tendency of the tunnel was detected by the comparison of adjacent fitting radius. The maximum error is 6 mm, while the minimum one is 1 mm. The computation cost of vertical section abstraction is within 3 seconds/section, which proves high efficiency..