Point cloud map construction and correction method and system based on tunnel construction map

Abstract

Disclosed in the present invention are a point cloud map construction and correction method and system based on a tunnel construction map. The method comprises: acquiring and pre-processing point cloud data, and combining point cloud key frames on the basis of initial poses corresponding thereto, so as to obtain a complete initial tunnel point cloud map; acquiring road sign points from the initial tunnel point cloud map; acquiring and using a geometric structure of a track in a tunnel construction map and an installation position of an on-board sensor in a train to calculate a pose trajectory of the on-board sensor in a tunnel; and using the pose trajectory to optimize the initial poses of the point cloud key frames on the basis of the positions of the road sign points in the tunnel construction map, so as to obtain a corrected tunnel point cloud map. The present invention can improve the accuracy and reliability of point cloud map construction without being limited and affected by an onsite environment.

Description

A method and system for constructing and correcting point cloud maps based on tunnel mapping. Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method and system for constructing and correcting point cloud maps based on tunnel mapping. Background Technology

[0002] With the acceleration of urbanization, the development and utilization of underground space has gradually become an important part of urban construction. Tunnels, as an effective underground transportation channel, are widely used in subways, light rail, and other fields. Among these, tunnel lidar point cloud maps play a crucial role in tunnel construction and the positioning of automated trains in tunnels. The methods for constructing and correcting tunnel point cloud maps directly affect the accuracy and reliability of the maps, and thus impact the positioning and obstacle detection of automated trains. Currently available point cloud map construction and correction methods struggle to handle the cumulative errors of odometer readings in tunnel scenarios lacking absolute observation. Therefore, an efficient and accurate point cloud map construction and correction method is needed to improve the accuracy and reliability of point cloud map construction. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for constructing and correcting point cloud maps based on tunnel mapping, so as to improve the accuracy and reliability of point cloud map construction without being limited or affected by the on-site environment.

[0004] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0005] A method for constructing and correcting point cloud maps based on tunnel mapping includes:

[0006] Acquire and preprocess point cloud data, and merge point cloud keyframes according to their corresponding initial poses to obtain a complete initial tunnel point cloud map.

[0007] Obtain road markers from the initial tunnel point cloud map;

[0008] By acquiring and utilizing the geometry of the track in the tunnel construction map and the installation positions of the on-board sensors in the train, the pose trajectory of the on-board sensors in the tunnel can be calculated.

[0009] Using the pose trajectory and based on the position of the landmark points in the tunnel map, the initial pose of the point cloud keyframe is optimized to obtain a corrected tunnel point cloud map.

[0010] Optionally, the acquisition and preprocessing of point cloud data includes: using the vehicle-mounted sensors to collect tunnel data.

[0011] The point cloud data is processed, and the point cloud data is filtered and distortion-reduced.

[0012] Optionally, the vehicle-mounted sensor includes a lidar.

[0013] Optionally, the initial pose corresponding to the point cloud keyframes is calculated using a laser odometry system, and all point cloud keyframes are merged using a point cloud library tool to obtain a complete initial tunnel point cloud map.

[0014] Optionally, a method for extracting prominent three-dimensional features of the tunnel surface is used to find landmarks in the initial tunnel point cloud map, the landmarks including kilometer markers and beacon points.

[0015] Optionally, the three-dimensional trajectory of the track in the tunnel can be constructed using the tunnel construction map, and then the pose trajectory of the on-board sensor moving in the tunnel can be calculated and converted according to the installation position of the on-board sensor in the train.

[0016] Optionally, by utilizing the pose trajectory of the vehicle-mounted sensor in the tunnel and the initial pose trajectory of the point cloud keyframe, and based on the location of the corresponding landmark point in the tunnel construction map, an objective equation is constructed to optimize the initial pose of the point cloud keyframe. Finally, the point cloud keyframes are merged to obtain the corrected tunnel point cloud map.

[0017] Optionally, the location of the corresponding road marker in the tunnel construction map is the absolute pose of the road marker relative to the vehicle-mounted sensor found in the tunnel construction map.

[0018] Optionally, the optimized point cloud keyframes can be merged using a laser odometry system to obtain the corrected tunnel point cloud map.

[0019] Optionally, the objective equation is a pose objective equation for landmark points and point cloud keyframes constructed using the pose graph optimization method, and the expression of the objective equation is as follows:

[0020]

[0021] Among them, residual It can be defined as:

[0022]

[0023] In the formula, This represents the pose matrix of the nth landmark point in the tunnel at the corresponding position on the lidar trajectory. This represents the keyframe pose matrix that allows observation of the nth landmark point. The superscript N indicates that there are a total of N landmark points. The function represents the mapping of the pose matrix form T logarithmically to the corresponding Lie algebra form, which corresponds to a six-dimensional vector; This is an information matrix representing the constraint factors of tunnel landmark points. This matrix can be adjusted according to the accuracy of the tunnel construction map. It is typically set as a diagonal matrix, with the diagonal elements representing the weights of the corresponding residuals. Represented as:

[0024]

[0025] in, Corresponding to the first three-dimensional translation components, Corresponding rotational component.

[0026] On the other hand, the present invention also provides a point cloud map construction and correction system based on tunnel construction maps, comprising: an initial point cloud map construction module, used to acquire and preprocess point cloud data, and merge point cloud keyframes according to corresponding initial poses to obtain a complete initial tunnel point cloud map; a landmark point extraction module, used to acquire landmark points in the initial tunnel point cloud map; a tunnel construction map track modeling module, used to acquire and utilize the geometric structure of the track in the tunnel construction map and the installation position of the on-board sensor in the train to calculate the pose trajectory of the on-board sensor in the tunnel; and a point cloud map correction module, used to optimize the initial pose of the point cloud keyframes using the pose trajectory and according to the position of the landmark points in the tunnel construction map to obtain a corrected tunnel point cloud map.

[0027] In another aspect, the present invention also provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the method described above.

[0028] In other respects, the present invention also provides a readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0029] Compared with the prior art, the present invention has at least the following technical effects:

[0030] This invention utilizes tunnel construction maps to map the prominent three-dimensional tunnel landmarks to the precise coordinates of the landmarks in the tunnel construction maps. It introduces global information to optimize the tunnel point cloud map, eliminating accumulated errors in the point cloud map construction process. This achieves improved accuracy and reliability of point cloud map construction without being limited or affected by the site environment.

[0031] This invention takes into account the difference between the direction of the track trajectory and the direction of the train lidar. It transforms the train track trajectory in the tunnel construction map into the motion trajectory of the on-board lidar based on geometric relationships, which is then used for point cloud map correction to improve the accuracy of point cloud map construction and correction.

[0032] This invention does not introduce additional sensor information, which reduces the consumption of real-time computing resources.

[0033] This invention requires no other auxiliary equipment, which can reduce equipment and installation costs. Attached Figure Description

[0034] Figure 1 is a flowchart illustrating a point cloud map construction and correction method based on tunnel mapping according to an embodiment of the present invention.

[0035] Figure 2 is a structural block diagram of a point cloud map construction and correction system based on tunnel mapping provided in an embodiment of the present invention;

[0036] Figure 3 is a schematic diagram of initial point cloud map construction and landmark extraction provided in an embodiment of the present invention;

[0037] Figure 4 is a schematic diagram of the tunnel construction map track center trajectory to the train lidar trajectory provided in an embodiment of the present invention;

[0038] Figure 5 is a schematic diagram of the keyframe pose optimization corresponding to the tunnel construction map and the initial point cloud map landmark points provided in an embodiment of the present invention. Detailed Implementation

[0039] The following detailed description, in conjunction with the accompanying drawings and specific embodiments, provides a further detailed explanation of the point cloud map construction and correction method and system based on tunnel mapping proposed in this invention. The advantages and features of this invention will become clearer from the following description. It should be noted that the accompanying drawings are in a very simplified form and use non-precise scales, used only to facilitate and clearly illustrate the purpose of the embodiments of this invention. Please refer to the accompanying drawings to make the objectives, features, and advantages of this invention more apparent and understandable. It should be understood that the structures, scales, sizes, etc., depicted in the accompanying drawings are only used to complement the content disclosed in the specification, for those skilled in the art to understand and read, and are not intended to limit the implementation conditions of this invention. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportional relationships, or adjustments to the size, without affecting the effects and objectives achieved by this invention, should still fall within the scope of the technical content disclosed in this invention.

[0040] As described in the background section, the main methods for constructing and correcting point cloud maps within tunnels currently include the following:

[0041] One solution involves incorporating information from IMUs (Inertial Measurement Units) and odometers. IMUs can provide information such as angular velocity and acceleration, while odometers provide train displacement information. By fusing IMU and odometer information with LiDAR data, the accuracy of point cloud map construction within tunnels can be improved.

[0042] Another solution is to introduce auxiliary equipment such as UWB (Ultra Wide Band) and reflectors. These auxiliary methods can provide additional information to assist in the construction of point cloud maps and improve the accuracy of point cloud mapping.

[0043] However, the above methods have some problems in tunnel scenarios, mainly including the following:

[0044] First, in tunnel scenarios, absolute observation is lacking. While introducing IMU and odometry information can improve the accuracy of point cloud map construction, its accuracy and stability are still limited by sensor noise and errors. Furthermore, the cumulative errors of the IMU and odometry can lead to cumulative errors in the point cloud map construction process, thus affecting subsequent point cloud localization and obstacle detection.

[0045] In the absence of global information in tunnels, auxiliary methods such as reflective pillars can only enrich the point cloud features of the tunnel scene, but cannot eliminate the cumulative errors in the mapping process. Furthermore, auxiliary methods such as UWB or reflective pillars will greatly increase the external deployment cost, and the implementation conditions are easily limited and affected by the site environment.

[0046] In summary, existing methods for constructing and correcting tunnel point cloud maps for rail transit suffer from problems such as difficulty in eliminating accumulated errors, low accuracy, high cost of additional equipment, and poor versatility.

[0047] In view of this, as shown in Figure 1, this embodiment provides a point cloud map construction and correction method based on tunnel mapping, including: acquiring and preprocessing point cloud data, and merging point cloud keyframes according to their corresponding initial poses to obtain a complete initial tunnel point cloud map.

[0048] In this embodiment, each frame of point cloud data extracted periodically from the point cloud data at preset time intervals is used as the point cloud keyframe.

[0049] In this embodiment, the point cloud registration method is used to calculate the point cloud keyframes to obtain the initial pose of all point cloud keyframes.

[0050] Obtain landmarks from the initial tunnel point cloud map.

[0051] By acquiring and utilizing the geometry of the track in the tunnel construction map and the installation positions of the onboard sensors in the train, the pose trajectory of the onboard sensors in the tunnel can be calculated.

[0052] Using the pose trajectory and based on the position of the landmark points in the tunnel map, the initial pose of the point cloud keyframe is optimized to obtain a corrected tunnel point cloud map.

[0053] This embodiment utilizes tunnel construction maps to map the prominent 3D tunnel landmarks in the tunnel to the precise coordinates of the landmarks in the tunnel construction maps. It introduces global information to optimize the tunnel point cloud map, eliminating accumulated errors in the point cloud map construction process. This improves the accuracy and reliability of point cloud map construction without being limited or affected by the site environment.

[0054] Please continue to refer to Figure 1. In this embodiment, the acquisition and preprocessing of point cloud data includes: using the vehicle-mounted sensor to collect the point cloud data in the tunnel, and performing filtering and distortion removal processing on the point cloud data.

[0055] Please continue referring to Figure 1. In this embodiment, the vehicle-mounted sensor includes a lidar. This embodiment does not introduce other additional sensor information, which reduces the consumption of real-time computing resources.

[0056] Please continue referring to Figure 1. In this embodiment, the initial pose corresponding to the point cloud keyframes is calculated using laser odometry. All point cloud keyframes are then merged using a Point Cloud Library (PCL) tool to obtain a complete initial tunnel point cloud map. It is understood that laser odometry is a commonly used method in laser SLAM (Simultaneous Localization and Mapping). It typically uses the Iterative Closest Point (ICP) algorithm or Normal Distribution Transform (NDT) to register two point cloud frames and obtain the transformation relationship between them.

[0057] Please continue to refer to Figure 1. In this embodiment, the method of extracting the three-dimensional features of the tunnel surface is used to find the landmark points in the initial tunnel point cloud map. The landmark points include kilometer markers and beacon points.

[0058] Please continue to refer to Figure 1. In this embodiment, the three-dimensional trajectory of the track in the tunnel is constructed using the tunnel construction map. Then, based on the installation position of the vehicle-mounted sensor in the train, the pose trajectory of the vehicle-mounted sensor moving in the tunnel is calculated and converted.

[0059] Please continue to refer to Figure 1. In this embodiment, the initial pose trajectory of the point cloud keyframes and the pose trajectory of the vehicle-mounted sensor in the tunnel are used. Based on the location of the corresponding road sign points in the tunnel construction map, an objective equation is constructed to optimize the initial pose of the point cloud keyframes. Finally, the point cloud keyframes are merged to obtain the corrected tunnel point cloud map.

[0060] Please continue to refer to Figure 1. In this embodiment, the location of the road sign point in the tunnel construction map is the absolute pose of the road sign point relative to the vehicle sensor found in the tunnel construction map.

[0061] Please continue to refer to Figure 1. In this embodiment, the optimized point cloud keyframes are merged using a laser odometry to obtain the corrected tunnel point cloud map.

[0062] Please continue to refer to Figure 1. In this embodiment, the objective equation is the pose objective equation for landmark points and point cloud keyframes constructed using the pose graph optimization method:

[0063]

[0064] Among them, residual It can be defined as:

[0065]

[0066] In the formula, This represents the pose matrix of the nth landmark point in the tunnel at the corresponding position on the lidar trajectory. This represents the keyframe pose matrix that allows observation of the nth landmark point. The superscript N indicates that there are a total of N landmark points. The function represents the mapping of the pose matrix form T logarithmically to the corresponding Lie algebra form, which corresponds to a six-dimensional vector; This is an information matrix representing the constraint factors of tunnel landmark points. This matrix can be adjusted according to the accuracy of the tunnel construction map. It is typically set as a diagonal matrix, and the diagonal elements are used to represent the weights of the corresponding residuals. Represented as:

[0067]

[0068] in, Corresponding to the first three-dimensional translation components, Corresponding rotational component.

[0069] This embodiment takes into account the difference between the direction of the track trajectory and the direction of the lidar. The train track trajectory in the tunnel construction map is transformed into the motion trajectory of the vehicle-mounted lidar according to the geometric relationship, which is used for point cloud map correction to improve the accuracy of point cloud map construction and correction.

[0070] As shown in Figure 2, on the other hand, the present invention also provides a point cloud map construction and correction system based on tunnel mapping, comprising:

[0071] The initial point cloud map construction module S01 is used to acquire and preprocess point cloud data, and merge point cloud keyframes according to the corresponding initial pose to obtain a complete initial tunnel point cloud map.

[0072] The road sign extraction module S02 is used to obtain road signs from the initial tunnel point cloud map;

[0073] The tunnel construction map track modeling module S03 is used to obtain and utilize the geometry of the track in the tunnel construction map and the installation position of the on-board sensors in the train to calculate the pose trajectory of the on-board sensors in the tunnel.

[0074] The point cloud map correction module S04 is used to optimize the initial pose of the point cloud keyframe using the pose trajectory and according to the position of the landmark point in the tunnel map, so as to obtain the corrected tunnel point cloud map.

[0075] This embodiment requires no other auxiliary equipment, which can reduce equipment and installation costs.

[0076] To better understand the above embodiments, a more specific embodiment will be described below:

[0077] As shown in Figure 3, raw tunnel point cloud data was collected by a lidar installed in front of the train. The distance between adjacent stations is about 1000m. A frame of raw point cloud data was extracted at intervals of about 3m as a key frame. The content shown in Figure 3 is several consecutive key frames and feature points in an adjacent station. The key frames were filtered to remove invalid points, remove distortion, and extract features.

[0078] The initial pose of all point cloud keyframes was obtained using a point cloud registration method based on the iterative nearest point method, which employs laser odometry.

[0079] The calculation formula is:

[0080]

[0081] in, Representing the source cloud, Represents the target point cloud; This represents the optimal estimated pose result.

[0082] The initial poses of all point cloud keyframes are calculated using the above formula. This is represented using the triangles in Figure 3. This represents the rotation matrix of the point cloud in the nth frame relative to the initial point cloud coordinate system; This represents the translation vector of the point cloud in frame n relative to the initial point cloud coordinate system.

[0083] Based on the initial poses of all point cloud keyframes, approximately three hundred point cloud keyframes are stitched together according to their corresponding initial poses to obtain an initial tunnel point cloud map with complete tunnel feature structure.

[0084] Understandably, the point cloud keyframe data here is taken from the original frame point cloud data. Since the amount of original frame point cloud data of the LiDAR is too large, 10 frames of point cloud can be received per second, and each frame of point cloud has 20,000+ points, a fixed distance is used to select one frame of point cloud data as the point cloud keyframe. This can ensure that the device's calculation time is short and the map point cloud is evenly distributed.

[0085] Then, continue searching the initial tunnel point cloud map for prominent kilometer markers and beacon points (road sign locations) within the tunnel. n As shown in the pentagram in Figure 3.

[0086] It is understandable that the kilometer markers are located on the side of the tunnel wall, while the beacon points are located in the middle of the track. Only the kilometer markers and beacons have clear location markings on the tunnel construction map and can be directly observed and identified in the initial tunnel point cloud map.

[0087] And continue to find the keyframe K in the initial tunnel point cloud map that corresponds to the observed features. n (That is, the keyframe point cloud of beacons or kilometer markers can be observed).

[0088] Understandably, the features here are those that beacons and kilometer markers can be directly observed in the initial tunnel point cloud map. For example, a beacon is located in the middle of the track sleepers and protrudes, making it directly observable in the initial tunnel point cloud map. A kilometer marker is a square metal plate with kilometer numbers on the side of the tunnel wall, which is a highly reflective target in the initial tunnel point cloud map, making it visually observable. The specific kilometer numbers can also be seen in the corresponding video, thus allowing for precise location on the tunnel construction map.

[0089] Furthermore, assuming two adjacent stations are 1 km apart, there are approximately 300 point cloud keyframes. Beacons and kilometer markers are relatively sparse, with about ten kilometer markers and about 20 beacons. Not all point cloud keyframes contain observable beacons or kilometer markers. We need to find the keyframes that contain observable one type of landmark (i.e., keyframe K). n Only when the pose of such keyframes is obtained can it be used as an optimization variable in the subsequent tunnel point cloud map optimization process.

[0090] Next, use a tool capable of viewing 3D point clouds (such as the pcl_viewer tool, which allows selecting beacon points or kilometer markers in the point cloud and obtaining their coordinates in the keyframe point cloud coordinate system) to view the keyframe K of the point cloud. n Middle kilometer markers and beacon points F n The transformation matrix T to the origin of the point cloud coordinates in this frame Fn->Kn Used for converting the center coordinate trajectory of the lidar in subsequent tunnel construction maps.

[0091] As shown in Figure 4, the track path between adjacent stations can be calculated from the side and top views of the tunnel construction drawing, forming a three-dimensional track in three-dimensional space. Since the lidar is located at the head of the train, its direction of motion does not coincide with the direction of the track path in the tunnel. Therefore, based on its installation position at the head of the train, the new trajectory of the lidar in the tunnel as the train moves forward needs to be calculated using the geometric relationships shown in Figure 4.

[0092] It is understandable that the geometric relationship here can be obtained from the positional relationship between the lidar installation position and the center of the train chassis in Figure 3. The center trajectory of the track can be obtained from the tunnel construction map, and the actual movement trajectory of the lidar can be obtained from the schematic diagram of the positional relationship in Figure 4.

[0093] Calculation formula:

[0094]

[0095] in, These are the pose points on the motion trajectory of the LiDAR at the front of the vehicle. This represents the position and pose point of the track center trajectory in the tunnel construction diagram. This represents the positional relationship between the lidar installation location and the center of the train's chassis.

[0096] As shown in Figure 5, based on the precise marking positions of road markers in the tunnels between stations in the tunnel construction map, the precise position F of the road markers on the track path can be calculated. nT Then, the transformation matrix T calculated in Figure 3 is used. Fn->Kn The precise location F of the landmarks in the track path n T Re-convert to the actual pose K of the corresponding lidar in the orbital path n Tunnel .

[0097] The specific calculation formula is as follows:

[0098]

[0099] Based on the keyframe pose K of the landmark point that can be observed in both of the above processes, n and actual pose K n Tunnel Construct the objective equation:

[0100]

[0101] Among them, residual It can be defined as:

[0102]

[0103] In the formula, This represents the pose matrix of the nth landmark point in the tunnel at the corresponding position on the lidar trajectory. This represents the keyframe pose matrix that allows observation of the nth landmark point. The superscript N indicates that there are a total of N landmark points. The function represents the mapping of the pose matrix form T logarithmically to the corresponding Lie algebra form, which corresponds to a six-dimensional vector; The information matrix representing the constraint factors of tunnel landmarks can be adjusted according to the accuracy of the tunnel construction map. It can usually be set as a diagonal matrix, and the diagonal elements are used to represent the weights of the corresponding residuals.

[0104] Solve the optimization objective equation (for example, using Ceres Solver (a nonlinear optimization library developed by Google) to obtain the keyframe poses of the optimized landmarks, and then re-merge them to obtain the corrected point cloud map.

[0105] In other aspects, the present invention also provides an electronic device including a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the method described above.

[0106] In other respects, the present invention also provides a readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0107] In summary, the method provided in this embodiment includes: First, point cloud data within the tunnel is collected using sensors such as LiDAR. After preprocessing operations such as filtering and distortion correction, key frames of the point cloud are merged using a laser odometry system based on their corresponding initial poses to obtain a complete initial tunnel point cloud map. The corresponding kilometer markers and beacon points in the initial tunnel point cloud map are then located. Simultaneously, a three-dimensional trajectory of the track within the tunnel is constructed using the tunnel construction map. Then, based on the location of the LiDAR installed on the train, the pose trajectory of the LiDAR moving within the tunnel is calculated and converted. Using the calculated LiDAR pose trajectory within the tunnel and the initial pose trajectory of the key frames, an objective equation is constructed based on the corresponding landmark positions in the tunnel construction map to optimize the initial pose of the key frames. Finally, the key frame point clouds are merged to obtain a corrected point cloud map, achieving the goal of eliminating accumulated errors from the laser odometry system and improving the accuracy and reliability of the point cloud map.

[0108] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0109] It should be noted that the apparatus and methods disclosed in the embodiments herein can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments herein. In this regard, each block in a flowchart or block diagram may represent a module, program, or part of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system to perform the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0110] In addition, the functional modules in the various embodiments of this article can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0111] Although the present invention has been described in detail through the preferred embodiments above, it should be understood that the above description should not be considered as a limitation of the present invention. Various modifications and substitutions to the present invention will be apparent to those skilled in the art after reading the above description. Therefore, the scope of protection of the present invention should be defined by the appended claims.

Claims

1. A method for constructing and correcting point cloud maps based on tunnel mapping, characterized in that, include: Acquire and preprocess point cloud data, and merge point cloud keyframes according to their corresponding initial poses to obtain a complete initial tunnel point cloud map. Obtain road markers from the initial tunnel point cloud map; By acquiring and utilizing the geometry of the track in the tunnel construction map and the installation positions of the on-board sensors in the train, the pose trajectory of the on-board sensors in the tunnel can be calculated. Using the pose trajectory and based on the position of the landmark points in the tunnel map, the initial pose of the point cloud keyframe is optimized to obtain a corrected tunnel point cloud map.

2. The point cloud map construction and correction method based on tunnel mapping as described in claim 1, characterized in that, The acquisition and preprocessing of point cloud data includes: using the vehicle-mounted sensor to collect the point cloud data in the tunnel, and performing filtering and distortion removal processing on the point cloud data.

3. The point cloud map construction and correction method based on tunnel mapping as described in claim 1, characterized in that, The vehicle-mounted sensors include lidar.

4. The point cloud map construction and correction method based on tunnel mapping as described in claim 3, characterized in that, The initial pose corresponding to the point cloud keyframes is calculated using laser odometry, and all point cloud keyframes are merged using a point cloud library tool to obtain a complete initial tunnel point cloud map.

5. The point cloud map construction and correction method based on tunnel mapping as described in claim 1, characterized in that, By using the method of extracting prominent three-dimensional features of the tunnel surface, the landmark points in the initial tunnel point cloud map are found. The landmark points include kilometer markers and beacon points.

6. The point cloud map construction and correction method based on tunnel mapping as described in claim 1, characterized in that, The three-dimensional trajectory of the track in the tunnel is constructed using the tunnel construction map. Then, based on the installation position of the on-board sensor in the train, the pose trajectory of the on-board sensor moving in the tunnel is calculated and converted.

7. The point cloud map construction and correction method based on tunnel mapping as described in claim 1, characterized in that, By utilizing the pose trajectory of the vehicle-mounted sensor in the tunnel and the initial pose trajectory of the point cloud keyframes, and based on the corresponding landmark positions in the tunnel construction map, an objective equation is constructed to optimize the initial pose of the point cloud keyframes. Finally, the point cloud keyframes are merged to obtain the corrected tunnel point cloud map.

8. The point cloud map construction and correction method based on tunnel mapping as described in claim 7, characterized in that, The location of the corresponding road sign in the tunnel construction map is to find the road sign in the tunnel construction map. Absolute pose of onboard sensors.

9. The point cloud map construction and correction method based on tunnel mapping as described in claim 7, characterized in that, The optimized point cloud keyframes are merged using a laser odometry system to obtain the corrected tunnel point cloud map.

10. The point cloud map construction and correction method based on tunnel mapping as described in claim 7, characterized in that, The objective equation is the pose objective equation for landmark points and point cloud keyframes constructed using the pose graph optimization method. The expression of the objective equation is as follows: Among them, residual It can be defined as: In the formula, This represents the pose matrix of the nth landmark point in the tunnel at the corresponding position on the lidar trajectory. This represents the keyframe pose matrix that allows observation of the nth landmark point. The superscript N indicates that there are a total of N landmark points. The function represents the mapping of the pose matrix form T logarithmically to the corresponding Lie algebra form, which corresponds to a six-dimensional vector; This is an information matrix representing the constraint factors of tunnel landmark points. This matrix can be adjusted according to the accuracy of the tunnel construction map. It is typically set as a diagonal matrix, with the diagonal elements representing the weights of the corresponding residuals. Represented as: in, Corresponding to the first three-dimensional translation components, Corresponding rotational component.

11. A point cloud map construction and correction system based on tunnel mapping, characterized in that, include: The initial point cloud map construction module is used to acquire and preprocess point cloud data, and merge point cloud keyframes according to the corresponding initial pose to obtain a complete initial tunnel point cloud map. The landmark extraction module is used to obtain landmarks from the initial tunnel point cloud map; The tunnel construction map track modeling module is used to obtain and utilize the geometry of the track in the tunnel construction map and the installation position of the on-board sensors in the train to calculate the pose trajectory of the on-board sensors in the tunnel. The point cloud map correction module is used to optimize the initial pose of the point cloud keyframes by utilizing the pose trajectory and the position of the landmark points in the tunnel map, so as to obtain the corrected tunnel point cloud map.

12. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, which, when executed by the processor, implements the method of any one of claims 1 to 10.

13. A readable storage medium, characterized in that, The readable storage medium stores a computer program, which, when executed by a processor, implements the method of any one of claims 1 to 10.

Images