A map construction method and device, computer equipment and readable storage medium

Abstract

The application relates to a map construction method and device, computer equipment and a readable storage medium. The method comprises the following steps: acquiring laser radar point clouds collected by different sensors, pipeline environment data, pipeline visual data and motion data during movement of a robot in a pipeline; in the case of reaching a current map construction moment, predicting an initial pose of the robot according to the motion data, and determining a refractive index compensation factor of the pipeline according to the pipeline environment data; determining initial map point clouds of a target period according to the initial pose, the refractive index compensation factor and the laser radar point clouds and the pipeline visual data in the target period; correcting the initial map point clouds according to a preset curvature threshold and the refractive index compensation factor to obtain a target local map; and updating an initial global map at a previous map construction moment according to the target local map to obtain a target global map. The method can improve the accuracy of the constructed map.

Description

Technical Field

[0001] This application relates to the field of robotics, and in particular to a map building method, apparatus, computer device, and readable storage medium. Background Technology

[0002] Underground pipelines serve as crucial infrastructure for urban energy transmission and industrial fluid transport, making their internal inspection and maintenance essential for ensuring operational safety. Traditional inspection methods primarily rely on manual labor or simple camera equipment, which is not only inefficient and labor-intensive but also suffers from poor accessibility in long-distance, narrow, or complexly curved pipelines, posing serious safety hazards.

[0003] With the development of robotics technology, robot-based pipeline inspection methods are being used more and more widely. However, due to the common problems inside pipelines such as low light, sparse texture, dust, and moisture interference, existing SLAM (Simultaneous Localization and Mapping) technologies suffer from cumulative errors and localization drift, resulting in low accuracy of the constructed maps. Summary of the Invention

[0004] Therefore, it is necessary to provide a map construction method, apparatus, computer equipment, and readable storage medium to address the aforementioned technical problems, thereby improving the accuracy of the constructed map.

[0005] Firstly, this application provides a map construction method, including:

[0006] During the robot's movement inside the pipeline, the system acquires lidar point cloud data, pipeline environment data, pipeline visual data, and robot motion data collected by different sensors mounted on the robot.

[0007] Given the current map building time, predict the robot's initial pose based on the motion data at the current map building time, and determine the pipe's refractive index compensation factor based on the pipe environment data at the current map building time.

[0008] Based on the initial pose, refractive index compensation factor, and lidar point cloud and pipeline visual data within the target time period, the initial map point cloud for the target time period is determined; where the target time period is the time period between the current map building time and the previous map building time.

[0009] Based on the preset curvature threshold and refractive index compensation factor, the initial map point cloud is corrected to obtain the target local map for the target time period;

[0010] The initial global map at the time of the previous map construction is updated based on the target local map to obtain the target global map at the time of the current map construction.

[0011] In one embodiment, the motion data includes inertial navigation measurement data and wheel rotation data; predicting the robot's initial pose based on the motion data at the current map construction time includes: determining the robot's initial motion parameters based on the wheel rotation data at the current map construction time, and correcting the initial motion parameters using a preset slip coefficient to obtain calibrated target motion parameters; using the inertial navigation measurement data as the predicted value and the target motion parameters as the observed value, predicting the robot's pose to obtain the robot's initial pose.

[0012] In one embodiment, the pipeline environment data includes temperature, humidity, and pressure; the refractive index compensation factor of the pipeline is determined based on the pipeline environment data at the current map construction time, including: determining the refractive index compensation factor of the pipeline based on the temperature, humidity, and pressure at the current map construction time.

[0013] In one embodiment, the motion data includes inertial navigation measurement data; determining the initial map point cloud for the target time period based on the initial pose, refractive index compensation factor, and LiDAR point cloud and pipeline visual data within the target time period includes: generating the robot's current trajectory sequence within the target time period based on the initial pose and pipeline visual data within the target time period; determining the robot's pose set within the target time period based on the pre-integration results of the current trajectory sequence and the inertial navigation measurement data within the target time period; using the pose set to perform distortion compensation on the LiDAR point cloud within the target time period to obtain a compensated point cloud, and performing intensity correction on the compensated point cloud based on the refractive index compensation factor to obtain a corrected point cloud; and filtering the corrected point cloud based on the radar echo data of each point in the corrected point cloud to obtain the initial map point cloud for the target time period.

[0014] In one embodiment, the initial map point cloud is corrected according to a preset curvature threshold and a refractive index compensation factor to obtain a target local map for the target time period. This includes: dividing the initial map point cloud into a first point cloud region and a second point cloud region according to the preset curvature threshold; wherein the curvature of the first point cloud region is less than the preset curvature threshold, and the curvature of the second point cloud region is not less than or greater than the preset curvature threshold; interpolating and fusing the boundary region between the first point cloud region and the second point cloud region to obtain an initial local map; and performing intensity correction on the initial local map according to the refractive index compensation factor to obtain a target local map for the target time period.

[0015] In one embodiment, updating the initial global map at the previous map construction time based on the target local map to obtain the target global map at the current map construction time includes: constructing an optimized local map at the current map construction time based on the current trajectory sequence, the target local map, and the motion data corresponding to each trajectory point in the current trajectory sequence; selecting a specified local map in the initial global map at the previous map construction time whose pipeline structure features match the pipeline structure features of the optimized local map; determining the overlap rate between the historical trajectory sequence corresponding to the specified local map and the current trajectory sequence; if the overlap rate is greater than a preset overlap rate threshold, correcting the specified local map based on the target local map, and correcting the initial global map based on the corrected specified local map to obtain the target global map at the current map construction time.

[0016] Secondly, this application also provides a map building apparatus, comprising:

[0017] The data acquisition module is used to acquire lidar point cloud data, pipeline environment data, pipeline visual data, and robot motion data collected by different sensors on the robot during the robot's movement inside the pipeline.

[0018] The pose prediction module is used to predict the robot's initial pose based on the motion data at the current map building time.

[0019] The compensation factor determination module is used to determine the refractive index compensation factor of the pipeline based on the pipeline environment data at the current map construction time.

[0020] The point cloud determination module is used to determine the initial map point cloud for the target time period based on the initial pose, refractive index compensation factor, and LiDAR point cloud and pipeline visual data within the target time period; where the target time period is the time period between the current map building time and the previous map building time.

[0021] The map building module is used to correct the initial map point cloud according to the preset curvature threshold and refractive index compensation factor to obtain the target local map for the target time period.

[0022] The map update module is used to update the initial global map at the time of the previous map construction based on the target local map, so as to obtain the target global map at the time of the current map construction.

[0023] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the various method embodiments provided in the first aspect above.

[0024] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the various method embodiments provided in the first aspect above.

[0025] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the various method embodiments provided in the first aspect above.

[0026] The aforementioned map building method, apparatus, computer equipment, and readable storage medium utilize various sensors mounted on a robot to acquire different types of data, such as LiDAR data, pipeline environment data, pipeline visual data, and robot motion data, during the robot's movement within the pipeline. At each map building moment, the robot's initial pose is predicted based on the motion data at that moment, and the pipeline's refractive index compensation factor is determined based on the pipeline environment data at that moment. Then, using the time interval between the current and previous map building moments as the target time interval, the initial map point cloud for the target time interval is determined based on the initial pose, refractive index compensation factor, LiDAR point cloud, and pipeline visual data within the target time interval. This initial map point cloud is then corrected according to a preset curvature threshold and the refractive index compensation factor to obtain a target local map for the target time interval. Finally, the initial global map from the previous map building moment is updated based on the target local map to obtain the target global map for the current map building moment. Thus, the map building for the current map building moment is completed. In this way, on the one hand, by fusing multi-source data collected by different sensors on the robot, high-precision map construction can be achieved in complex pipeline environments, improving the accuracy of the constructed local map; on the other hand, by updating the historical global map with the local map at each map construction time, the positioning drift of the global map can be reduced, improving the accuracy of the pipeline global map, and thus enhancing the detection accuracy of pipeline inspection based on the constructed map. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This application provides an application environment diagram for the map construction method provided in some embodiments of this application.

[0029] Figure 2A flowchart illustrating a map construction method provided in some embodiments of this application;

[0030] Figure 3 A flowchart illustrating the prediction of initial pose provided for some embodiments of this application;

[0031] Figure 4 A schematic diagram illustrating the process of determining an initial map point cloud provided in some embodiments of this application;

[0032] Figure 5 A flowchart illustrating the process of determining a target local map provided in some embodiments of this application;

[0033] Figure 6 A schematic diagram illustrating the process of updating the initial global map provided in some embodiments of this application;

[0034] Figure 7 A flowchart illustrating a map construction method provided in other embodiments of this application;

[0035] Figure 8 Structural block diagrams of a map building apparatus provided in some embodiments of this application;

[0036] Figure 9 Internal structural diagrams of a computer device provided in some embodiments of this application;

[0037] Figure 10 Internal structural diagrams of a computer device provided for other embodiments of this application. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0039] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments or any combination of multiple embodiments.

[0040] With the development of robotics technology, robot-based pipeline inspection methods are being used more and more widely. However, due to the common problems inside pipelines such as low light, sparse texture, dust, and moisture interference, existing SLAM (Simultaneous Localization and Mapping) technologies suffer from cumulative errors and localization drift, resulting in low accuracy of the constructed maps.

[0041] To address the aforementioned technical issues, in an exemplary embodiment, a map construction method is provided. This method can be applied to the robot's own controller (processor) or to the robot's backend server. The backend server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services, hereinafter referred to as a server.

[0042] Based on the above embodiments, in an exemplary embodiment, the map construction method provided in this application can be applied to, for example, Figure 1 In the application environment shown, robot 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located in the cloud or on other network servers. The data storage system can record various types of data collected by different sensors mounted on the robot, local maps built at different map-building times, global maps, etc. Server 104 then retrieves the aforementioned data from the data storage system to execute a map-building method provided in this embodiment.

[0043] Based on the above embodiments, in an exemplary embodiment, such as Figure 2 As shown, a map building method is provided. Taking the application of this method to the robot's own controller (processor) as an example, it can include the following steps:

[0044] S201: During the robot's movement within the pipeline, it acquires LiDAR point cloud data, pipeline environment data, pipeline visual data, and robot motion data collected by different sensors mounted on the robot.

[0045] Typically, various sensors can be mounted on the robot, such as lidar, IMU (Inertial Measurement Unit), wheel speedometer, image acquisition equipment, and environmental sensors (such as thermometers, hygrometers, and pressure sensors), to collect data during the robot's movement within the pipeline, thereby obtaining lidar point clouds, pipeline visual data, pipeline visual data, and robot motion data.

[0046] Optionally, the robot's motion data includes inertial navigation measurement data acquired by the IMU and wheel rotation data acquired by the wheel speedometer. The robot may also be equipped with other sensors to collect motion data, such as an odometer to collect displacement data; no specific limitations are imposed on this.

[0047] Optionally, the image acquisition device includes a global shutter camera and / or an event camera for acquiring pipe visual data. The pipe visual data acquired by the global shutter camera is pipe image data, while the pipe visual data acquired by the event camera is a pipe event stream, which is a time-series data set based on asynchronous visual sensing principles, triggered by pixel-level brightness changes, and composed of discrete events as basic units. The robot may also be equipped with other sensors to acquire pipe visual data, such as a depth camera; no specific limitations are imposed on this.

[0048] Optionally, the robot is equipped with a lidar, which can generate lidar point clouds based on the emitted radar signals and the received echo signals.

[0049] Optionally, the robot is equipped with a near-infrared illumination (NIRIllumination) device to provide a light source for the aforementioned image acquisition equipment, so as to accurately acquire visual data of the pipeline.

[0050] S202, when the current map building time is reached, predict the robot's initial pose based on the motion data at the current map building time.

[0051] Optionally, the robot performs map building periodically as it moves within the pipe. Therefore, a map building cycle can be preset. Each time interval corresponding to this cycle marks a map building moment. For example, if the robot's map building cycle is 0.01 seconds, then a map building moment occurs every 0.01 seconds. This map building cycle moment is then considered the current map building moment.

[0052] Optionally, during the robot's movement within the pipe, the map-building time can be determined based on the robot's operating state. For example, if the robot's speed change is detected to be greater than a speed change threshold, the current time can be determined as the current map-building time, indicating that the robot's motion state has changed drastically, such as the robot turning within the pipe. Alternatively, if the rate of change of the texture features in the pipe image data is detected to be greater than a feature change rate threshold, the current time can be determined as the current map-building time, indicating that the robot has moved into a new environment.

[0053] Based on this, when the current map building time is detected, the robot's pose at the current map building time can be predicted based on the motion data of the current map building time collected by the different sensors, so as to obtain the robot's initial pose.

[0054] Optionally, a Kalman filter is used to determine the robot's initial pose based on the aforementioned motion data. For example, the robot's prior pose at the current map-building time is determined based on the initial pose predicted at the previous map-building time and the control parameters at the previous map-building time. Then, based on the robot's actual pose at the previous map-building time, the robot's motion data observations at the current map-building time are determined. Finally, the prior pose is optimized based on the robot's motion data at the current map-building time and the aforementioned motion data observations to obtain the robot's initial pose at the current map-building time.

[0055] Optionally, since the data acquisition cycles of the different sensors mounted on the robot differ from the defined cycle (or the non-periodic definition method) of the map building time, at the current map building time, the most recently acquired data from the different sensors mounted on the robot can be obtained as the data acquired at the current map building time. Time synchronization can then be performed on these various acquired data sets to align the different sensors. Optionally, time synchronization can be performed using the following formula:

[0056]

[0057] in, This refers to the trace length on a PCB (Printed Circuit Board). Where is the dielectric constant. To reduce the delay in digital-to-analog conversion (ensuring microsecond-level time synchronization accuracy). At the speed of light, This is the timestamp of the original data collection time. This is the timestamp of the data collection time after time synchronization.

[0058] Furthermore, considering that sensors such as IMUs, LiDAR, and image acquisition devices use different coordinate systems when acquiring data, the coordinate systems of these different sensors can be unified during initial pose prediction to calibrate the robot's extrinsic parameters. Optionally, the set of extrinsic parameters after coordinate system unification can be obtained by executing a helical trajectory in the calibration field (i.e., calibrating specific parameters through controllable helical motion), as shown in the following equation:

[0059]

[0060] in, Represents the set of extrinsic parameters. This indicates the transformation from the image acquisition device's coordinate system to the IMU's coordinate system. For the transformation of the lidar sensor to the IMU coordinate system, This is the transformation from the wheel velocity meter coordinate system to the robot base coordinate system.

[0061] Optionally, the radius of the spiral trajectory gradually increases and covers a certain range in both the horizontal and vertical directions to ensure full coverage of the sensor in terms of spatial attitude and position.

[0062] S203, Determine the refractive index compensation factor of the pipeline based on the pipeline environment data at the current map construction time.

[0063] The so-called refractive index compensation factor (RICF) is a dimensionless correction coefficient introduced in measurement scenarios based on optical / electromagnetic principles to correct measurement errors caused by the refractive index of the medium (such as gas, liquid, or solid material) along the measurement path deviating from the standard value.

[0064] Correspondingly, the refractive index compensation factor of a pipeline is a dimensionless dynamic correction coefficient introduced in pipeline-related measurement scenarios based on optical / electromagnetic principles to correct the deviation of the measurement results caused by the refractive index deviating from the standard value due to the pipeline's own material, the medium transported / contained inside the pipe, and environmental conditions.

[0065] In one optional embodiment, if the aforementioned pipeline environmental data includes temperature, humidity, and pressure, then step S103 may include determining the pipeline's refractive index compensation factor based on the temperature, humidity, and pressure at the current map construction time. Optionally, the pipeline's refractive index compensation factor can be determined using the following formula:

[0066]

[0067] in, The refractive index compensation factor for the pipeline (abbreviated as) ), The temperature at the moment the current map was built. The humidity at the moment the map is constructed. The pressure (e.g., atmospheric pressure) at the moment the map is constructed. and These are preset coefficients.

[0068] S204. Based on the initial pose, refractive index compensation factor, and lidar point cloud and pipeline visual data within the target time period, determine the initial map point cloud for the target time period.

[0069] The target time period is the time between the current map construction time and the previous map construction time.

[0070] The local map built at each map-building moment is a map of the path traversed by the robot during the time period between the previous map-building moment and the current map-building moment. During the robot's movement, the LiDAR performs laser scanning at preset laser scanning times to build a LiDAR point cloud, and the image acquisition device also collects pipeline visual data at the acquisition frequency. Thus, at the current map-building moment, the LiDAR point cloud and pipeline visual data within the target time period between the current map-building moment and the previous map-building moment can be determined. Based on the initial pose, refractive index compensation factor, and the LiDAR point cloud and pipeline visual data within the target time period, the initial map point cloud for the target time period is determined.

[0071] S205, based on the preset curvature threshold and refractive index compensation factor, corrects the initial map point cloud to obtain the target local map for the target time period.

[0072] Considering that the continuous movement of the robot can easily cause distortion of the LiDAR point cloud during the generation process, the initial map point cloud can be corrected first according to the preset curvature threshold and refractive index compensation factor to obtain the corrected point cloud. The corrected point cloud is then output in the form of a structured map to obtain the target local map for the target time period.

[0073] S206, Update the initial global map at the time of the previous map construction based on the target local map to obtain the target global map at the time of the current map construction.

[0074] Since the local map built at each map building moment is a map of the path the robot traversed during the time period between the previous map building moment and the current map building moment, after obtaining the target local map for the target time period, the initial global map (also known as the historical global map) at the previous map building moment is updated based on the target local map to obtain the target global map at the current map building moment.

[0075] Optionally, extract the pipe structure features (such as bending radius, flange, valve structure, support rhythm, etc.) of the target local map, and select a specified local map from the initial global map at the previous map construction time that matches the pipe structure features of the target local map. Replace the specified local map with the target local map to obtain the target global map at the current map construction time.

[0076] Optionally, the target local map and the initial global map at the time of the previous map construction can be sequentially stitched together to obtain the target global map at the time of the current map construction.

[0077] In the aforementioned map construction method, various sensors are mounted on the robot to acquire different types of data during the robot's movement within the pipe, such as LiDAR data, pipe environment data, pipe visual data, and robot motion data. At each map construction moment, the robot's initial pose is predicted based on the motion data at that moment, and the pipe's refractive index compensation factor is determined based on the pipe environment data at that moment. Next, using the time interval between the current and previous map construction moments as the target time interval, the initial map point cloud for the target time interval is determined based on the initial pose, refractive index compensation factor, LiDAR point cloud, and pipe visual data within the target time interval. This initial map point cloud is then corrected according to a preset curvature threshold and the refractive index compensation factor to obtain the target local map for the target time interval. Finally, the initial global map from the previous map construction moment is updated based on the target local map to obtain the target global map for the current map construction moment. Thus, the map construction for the current map construction moment is completed. In this way, on the one hand, by fusing multi-source data collected by different sensors on the robot, high-precision map construction can be achieved in complex pipeline environments, improving the accuracy of the constructed local map; on the other hand, by updating the historical global map with the local map at each map construction time, the positioning drift of the global map can be reduced, improving the accuracy of the pipeline global map, and thus enhancing the detection accuracy of pipeline inspection based on the constructed map.

[0078] Based on the above embodiments, in an exemplary embodiment, the prediction of the initial pose in S202 is further refined. Optionally, the motion data includes inertial navigation measurement data and wheel rotation data, then as follows... Figure 3 As shown, the following steps may be included:

[0079] S301: Based on the wheel rotation data at the current map construction time, determine the robot's initial motion parameters, and use a preset slip coefficient to correct the initial motion parameters to obtain the calibrated target motion parameters.

[0080] In this embodiment, the robot is equipped with an IMU and a wheel speedometer. The inertial navigation measurement data collected by the IMU and the wheel rotation data collected by the wheel speedometer constitute the robot's motion data. The wheel rotation data includes wheel speed and wheel steering angle. Therefore, the initial motion parameters of the robot at the current map construction moment can be determined based on the wheel rotation data. Optionally, the initial motion parameters include the initial linear velocity. Furthermore, considering the drift in wheel rotation data caused by wheel slippage due to factors such as wetness inside the pipe, a preset slip coefficient can be used to correct the initial motion parameters to obtain the calibrated target motion parameters, ensuring the accuracy of the final created map. Optionally, the preset slip coefficient is determined based on a preset slip ratio, where the preset slip ratio is shown in the following formula:

[0081]

[0082] in, To preset the sliding ratio, The linear velocity is determined based on wheel rotation data. This refers to the actual linear velocity. Optionally, the actual linear velocity can be determined in various ways, such as by using high-precision velocity sensors like laser Doppler velocimeters to collect linear velocity data, or by using an IMU and GPS (Global Positioning System), etc. There are no specific limitations on this method.

[0083] Optionally, the calibrated target motion parameters are shown in the following formula:

[0084] ,

[0085] in, For the linear velocity in the calibrated target motion parameters, For the angular velocity in the calibrated target motion parameters, The angular velocity of the robot's left wheel in the initial motion parameters is... The angular velocity of the robot's right wheel in the initial motion parameters. The wheelbase of the robot's left and right wheels. The preset slip coefficient is determined based on a preset slip ratio (e.g., 1 / (1+ )).

[0086] S302 uses inertial navigation measurement data as the predicted value and target motion parameters as the observed value to predict the robot's pose and obtain the robot's initial pose.

[0087] Since the inertial navigation measurement data includes the robot's three-axis angular velocity and three-axis acceleration, the robot's pose and motion data can be predicted based on the inertial navigation measurement data to obtain the predicted pose and predicted motion data. Then, the predicted motion data is compared with the target motion parameters determined above, and the predicted pose is corrected by comparing the obtained residual data to obtain the robot's initial pose.

[0088] Optionally, the initial pose is determined using a Kalman filter. Specifically, given the predicted pose and predicted motion data, the noise covariance matrix (process noise covariance matrix and measurement noise covariance matrix) of the Kalman filter is updated. Then, based on the residual data and the updated noise covariance matrix, the predicted pose is corrected to obtain the robot's initial pose. The initial pose is shown below:

[0089]

[0090] in, This is the initial pose. For inertial navigation measurement data, For the target motion parameters, This is a function for state transitions and observation updates in the EKF (Extended Kalman Filter).

[0091] Optionally, the process noise covariance matrix is ​​determined through experimental statistics, while the measurement noise covariance matrix is ​​set using the calibration data and measured error distribution provided by the sensor manufacturer at the time of sensor delivery.

[0092] Based on the above embodiments, in an exemplary embodiment, the determination of the initial map point cloud in S204 is further refined. Optionally, if the motion data includes inertial navigation measurement data, then as follows... Figure 4 As shown, the following steps may be included:

[0093] S401, based on the initial pose and the pipeline visual data within the target time period, generates the robot's current trajectory sequence within the target time period.

[0094] Optionally, the pipeline visual data includes pipeline image data and pipeline event stream. Appropriate pipeline visual data can be selected based on the lighting conditions within the pipeline to generate the current trajectory sequence. Specifically, when the lighting within the pipeline is sufficient (e.g., the ambient brightness collected by the robot's onboard brightness sensor exceeds a brightness threshold), sparse feature points from the pipeline image data (such as ORB (Oriented Fast and Rotated BRIEF) descriptors) are used to generate the robot's current trajectory sequence within the target time period. Conversely, when the lighting within the pipeline is insufficient (e.g., the ambient brightness collected by the robot's onboard brightness sensor is below a brightness threshold), the robot's onboard event camera is triggered to increase the temporal resolution features through the pipeline event stream, thereby generating the robot's current trajectory sequence within the target time period and improving robustness.

[0095] S402, based on the pre-integration results of the current trajectory sequence and the inertial navigation measurement data within the target time period, determine the robot's pose set within the target time period.

[0096] The so-called pre-integration result represents the relative motion increment of the inertial navigation measurement data from the previous map building time to the current map building time. This relative motion increment includes the rotation pre-integration component. (i.e., rotational increment, representing relative attitude change), velocity pre-integral component (i.e., velocity increment, characterizing the change in relative velocity), displacement pre-integral component (i.e., the increment of position change, representing relative displacement), where, When the previous map was built, This represents the current map construction time. Then, using the aforementioned pre-integration results as constraints, we can solve for the optimization objective to obtain the robot's pose set within the target time period.

[0097] Optionally, the optimization objective is as follows:

[0098]

[0099] in, The visual residuals of the pipeline visual data within the target time period. The IMU residuals of the inertial navigation measurement data within the target time period. This represents the geometric residual based on the pipe diameter constraint.

[0100] Thus, the above optimization objective integrates visual residuals, IMU residuals, and geometric residuals based on pipe diameter constraints. Furthermore, the weight coefficients of the above three types of residuals are calibrated experimentally and adaptively adjusted in combination with online residual statistics to maintain balance in different scenarios.

[0101] S403 uses pose set to perform distortion compensation on the lidar point cloud within the target time period to obtain the compensated point cloud, and then performs intensity correction on the compensated point cloud according to the refractive index compensation factor to obtain the corrected point cloud.

[0102] By using edge interpolation compensation technology, the LiDAR pose at each laser scanning moment of the LiDAR within the target time period is calculated based on each pose in the above pose set. Then, the distortion compensation of the LiDAR point cloud within the target time period is performed based on the above LiDAR pose to correct the LiDAR point cloud and obtain the compensated point cloud.

[0103] Optionally, distortion compensation can be performed on the lidar point cloud within the target time period using the following formula to obtain a spatiotemporally consistent compensated point cloud:

[0104]

[0105] in, To compensate for the laser scanning time in the point cloud The corresponding point cloud coordinates, Laser scanning time The pose of the lidar. The laser scanning time in the lidar point cloud within the target time period. The corresponding point cloud coordinates.

[0106] Furthermore, based on the aforementioned refractive index compensation factor and considering the aerosol scattering effect, the intensity values ​​of the compensated point cloud can be adaptively corrected to obtain the corrected point cloud. Optionally, the intensity values ​​of the compensated point cloud can be adaptively corrected using the following formula:

[0107]

[0108] in, To correct the intensity values ​​of the point cloud, To compensate for the intensity value of the point cloud, This is the straight-line distance from the laser radar's emission point to the target surface reached by the radar. This is the empirical scattering coefficient. The corrected intensity value better reflects actual physical properties and helps distinguish between pipe wall material and surface defects.

[0109] S404: Based on the radar echo data of each point in the corrected point cloud, the corrected point cloud is filtered to obtain the initial map point cloud for the target time period.

[0110] Based on the characteristics of lidar, each point in the corrected point cloud can have multiple radar echo data. Therefore, the initial map point cloud can be obtained by filtering the radar echo data of each point in the corrected point cloud.

[0111] Optionally, the corrected point cloud can be filtered using the following confidence-weighted selection formula:

[0112]

[0113] in, The corrected point cloud The filtering results at each point No. The first point Intensity weights for each echo data point For the first The first point The echo distance corresponding to each echo data point This is the attenuation coefficient.

[0114] In this embodiment, not only can the original point cloud of the lidar point cloud for the target time period be adaptively corrected by the environment, but the multi-echo return characteristics of each point in the lidar point cloud are also utilized to perform weighted selection of different echo signals of the same point to ensure the balanced preservation of high-reflection and weak-reflection points. Among them, a selection mechanism based on weight threshold is introduced in the filtering stage, which effectively eliminates false points caused by reflection from metal walls or foreign objects, while retaining key points that contribute to the geometric structure.

[0115] Based on the above embodiments, in an exemplary embodiment, the determination of the target local map in S205 is further refined. Optionally, such as Figure 5 As shown, the following steps may be included:

[0116] S501, based on a preset curvature threshold, divides the initial map point cloud into a first point cloud region and a second point cloud region.

[0117] Among them, the curvature of the first point cloud region is less than the preset curvature threshold, and the curvature of the second point cloud region is not less than or greater than the preset curvature threshold.

[0118] The curvature of each region in the initial map point cloud is determined, and regions with curvature less than a preset curvature threshold are classified as the first point cloud region, while regions with curvature not less than or greater than the preset curvature threshold are classified as the second point cloud region.

[0119] Optionally, the preset curvature threshold is 0.005. Optionally, a two-grid map is used to construct the first point cloud region, while a three-dimensional sparse voxel hashing SDF (Signed Distance Function) model is used to model the second point cloud region. The curvature threshold is determined based on experimental statistics. A suitable curvature threshold can effectively distinguish between smooth pipe wall regions and complex structural regions within the pipeline. Constructing a two-dimensional map for the first point cloud region has lower computational overhead and is suitable for real-time operation during long-distance inspections. Constructing a three-dimensional map for the second point cloud region significantly enhances the ability to represent fine structures and defects, thus achieving a balance between efficiency and accuracy in map construction.

[0120] S502, interpolation and fusion are performed on the boundary area between the first point cloud region and the second point cloud region to obtain an initial local map.

[0121] The boundaries of the first and second point cloud regions are determined to identify their intersection. Then, by interpolating and fusing this intersection, a smoothly transitioning initial local map is obtained. Optionally, different map-building methods are used to construct maps for the first and second point cloud regions, and the intersection of these different maps is interpolated and fused to obtain the initial local map. This ensures a smooth transition between the different maps corresponding to the first and second point cloud regions in the initial local map, avoiding geometric abrupt changes.

[0122] Optionally, the aforementioned intersection region is defined as a region with a preset width (e.g., 2 meters) at the center of the boundary between the first point cloud region and the second point cloud region. Further, optionally, the boundary region between the first point cloud region and the second point cloud region is interpolated and fused using the following formula:

[0123]

[0124] in, This is the map of the boundary area after interpolation and fusion. This refers to the map within the intersection region obtained by constructing a map of the first point cloud region. This refers to the map within the intersection region obtained by constructing a map of the second point cloud region. These are preset weighting coefficients.

[0125] S503, based on the refractive index compensation factor, performs intensity correction on the initial local map to obtain the target local map for the target time period.

[0126] Optionally, the initial map point cloud is subjected to intensity correction based on the refractive index compensation factor to obtain the correction result. Then, the initial local map is subjected to intensity correction using the correction result to obtain the local map point cloud for the target time period.

[0127] Optionally, the aforementioned target local map can be divided into multiple sub-map tiles and saved through compression and breakpoint resume mechanisms to support offline operation and indirect backhaul of map construction. This ensures that even in underground pipeline environments with long distances and no continuous communication links, the robot can still solidify and securely store the mapping results segment by segment.

[0128] Based on the above embodiments, in an exemplary embodiment, the update of the initial global map in S106 is further refined. Optionally, such as Figure 6 As shown, the following steps may be included:

[0129] S601, construct an optimized local map at the current map construction time based on the current trajectory sequence, the target local map, and the motion data corresponding to each trajectory point in the current trajectory sequence.

[0130] For each trajectory point in the current trajectory sequence, the map position corresponding to the trajectory point is determined in the aforementioned target local map, and the motion data and pipeline visual data corresponding to the trajectory point are determined from the acquired motion data and pipeline visual data. Then, for any two adjacent trajectory points in the current trajectory sequence, the difference between the map positions of the two trajectory points and the difference between the motion data and pipeline visual data corresponding to the two adjacent trajectory points are determined, and the residual between the above two differences is determined. Thus, the trajectory point with the smallest residual sum is selected from the aforementioned current trajectory sequence, and an optimized local map at the current map construction time is constructed using the selected trajectory points.

[0131] Optionally, based on the current trajectory sequence and target local map, as well as the motion data and pipeline visual data within the target time period, a factor graph optimization model is constructed, and the optimized local map at the current map construction time is obtained by solving the following optimization objective:

[0132]

[0133] in, The motion residuals of the wheel rotation data within the target time period. The visual residuals of the pipeline visual data within the target time period. The IMU residuals of the inertial navigation measurement data within the target time period. This represents the geometric residual based on the pipe diameter constraint.

[0134] S602, in the initial global map at the previous map construction time, select a specified local map whose pipeline structure features match the pipeline structure features of the optimized local map.

[0135] Extract the pipeline structure features (such as bending radius, flanges, valve structure, support rhythm, etc.) from the optimized local map, and select specified local maps from the initial global map at the time of the previous map construction that match the pipeline structure features of the optimized local map. Optionally, select specified local maps whose pipeline structure features are more similar to the pipeline structure features of the optimized local map than a similarity threshold.

[0136] S603, determine the overlap rate between the historical trajectory sequence and the current trajectory sequence corresponding to the specified local map.

[0137] Typically, when a robot traverses the same section of a pipe multiple times, the resulting sequences of movement trajectories should have a high overlap rate. Therefore, when a specific local map is obtained, the historical trajectory sequence corresponding to the specified local map can be determined from the robot's historical trajectory sequence, and the overlap rate between the historical trajectory sequence and the current trajectory sequence can be determined.

[0138] Optionally, steps S602-S603 above can be implemented through loop closure detection, i.e., extracting the structural features of the pipeline and combining them with a global appearance description based on a 3D (Three-Dimensional) scene Bag of Words (BoW) to perform loop closure detection. The similarity function is shown in the following equation:

[0139]

[0140] in, For similarity, when S exceeds the threshold And the ratio of prior residuals to a posteriori residuals At that time, it is confirmed that a closed loop has been detected, and a closed loop factor is generated. (That is, specifying the local map as the closure factor to correct the initial global map). and To integrate weights, This is the residual discrimination threshold, used to avoid erroneous loop closure; For a priori residuals, The former refers to the residual between the target local point cloud and the specified local point cloud, while the latter refers to the residual between the optimized local map and the specified local point cloud.

[0141] S604, if the overlap rate is greater than the preset overlap rate threshold, the specified local map is corrected according to the target local map, and the initial global map is corrected according to the corrected specified local map to obtain the target global map at the current map construction time.

[0142] As mentioned earlier, when the overlap rate determined above is greater than the preset overlap rate threshold, the drift of the constructed target local map can be considered small. Therefore, the specified local map can be directly corrected based on the target local map, and the initial global map can be corrected based on the corrected specified local map to obtain the target global map at the current map construction time, thereby achieving low-drift SLAM in long-distance pipeline environments. Optionally, the specified local map and the initial global map can be corrected using a sliding window method, wherein the correction frequency of the specified local map is higher than that of the initial global map.

[0143] Based on the above embodiments, in an exemplary embodiment, such as Figure 7 As shown, the map construction method may include the following steps:

[0144] S701 acquires LiDAR point cloud data, pipeline environment data, pipeline visual data, and robot motion data collected by different sensors on the robot as it moves within the pipeline.

[0145] S702 determines the robot's initial motion parameters based on the wheel rotation data at the current map construction time, and corrects the initial motion parameters using a preset slip coefficient to obtain the calibrated target motion parameters; wherein, the motion data includes inertial navigation measurement data and wheel rotation data.

[0146] S703 uses inertial navigation measurement data as predicted values ​​and target motion parameters as observed values ​​to predict the robot's pose, obtain the robot's initial pose, and determine the refractive index compensation factor of the pipeline based on the temperature, humidity, and pressure at the time of map construction; the pipeline environmental data includes temperature, humidity, and pressure.

[0147] S704: Based on the initial pose and the pipeline visual data within the target time period, generate the robot's current trajectory sequence within the target time period, and determine the robot's pose set within the target time period based on the pre-integration results of the current trajectory sequence and the inertial navigation measurement data within the target time period.

[0148] The S705 uses pose ensembles to perform distortion compensation on the lidar point cloud within the target time period, obtaining the compensated point cloud. Then, it performs intensity correction on the compensated point cloud according to the refractive index compensation factor, obtaining the corrected point cloud.

[0149] S706, based on the radar echo data of each point in the corrected point cloud, the corrected point cloud is filtered to obtain the initial map point cloud for the target time period, and the initial map point cloud is divided into a first point cloud region and a second point cloud region according to a preset curvature threshold; wherein the curvature of the first point cloud region is less than the preset curvature threshold, and the curvature of the second point cloud region is not less than or greater than the preset curvature threshold.

[0150] S707, interpolation and fusion are performed on the boundary area between the first point cloud region and the second point cloud region to obtain an initial local map, and intensity correction is performed on the initial local map according to the refractive index compensation factor to obtain the target local map for the target time period.

[0151] S708: Based on the current trajectory sequence, the target local map, and the motion data corresponding to each trajectory point in the current trajectory sequence, construct an optimized local map at the current map construction time.

[0152] S709, In ​​the initial global map at the previous map construction time, select a specified local map whose pipeline structure features match the pipeline structure features of the optimized local map, and determine the overlap rate between the historical trajectory sequence and the current trajectory sequence corresponding to the specified local map.

[0153] S710: If the overlap rate is greater than the preset overlap rate threshold, the specified local map is corrected according to the target local map, and the initial global map is corrected according to the corrected specified local map to obtain the target global map at the current map construction time.

[0154] The specific implementation methods of S701-S710 are the same as those in the above method embodiments, and will not be repeated here.

[0155] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0156] Based on the same inventive concept, this application also provides a map building apparatus for implementing the map building method described above. The solution provided by this apparatus is similar to the solution described in the above method; therefore, the specific limitations in one or more map building apparatus embodiments provided below can be found in the limitations of the map building method described above, and will not be repeated here.

[0157] In one exemplary embodiment, such as Figure 8 As shown, a map building apparatus is provided, including: a data acquisition module 810, a pose prediction module 820, a compensation factor determination module 830, a point cloud determination module 840, a map building module 850, and a map update module 860, wherein:

[0158] The data acquisition module 810 is used to acquire lidar point cloud, pipeline environment data, pipeline visual data and robot motion data collected by different sensors on the robot during the robot's movement in the pipeline.

[0159] The pose prediction module 820 is used to predict the robot's initial pose based on the motion data at the current map building time.

[0160] The compensation factor determination module 830 is used to determine the refractive index compensation factor of the pipeline based on the pipeline environment data at the current map construction time.

[0161] The point cloud determination module 840 is used to determine the initial map point cloud for the target time period based on the initial pose, refractive index compensation factor, and LiDAR point cloud and pipeline visual data within the target time period; wherein, the target time period is the time period between the current map construction time and the previous map construction time.

[0162] The map building module 850 is used to correct the initial map point cloud according to the preset curvature threshold and refractive index compensation factor to obtain the target local map for the target time period.

[0163] The map update module 860 is used to update the initial global map at the time of the previous map construction based on the target local map, so as to obtain the target global map at the time of the current map construction.

[0164] In an exemplary embodiment, the motion data includes inertial navigation measurement data and wheel rotation data; the pose prediction module 820 is specifically used to: determine the robot's initial motion parameters based on the wheel rotation data at the current map construction time, and correct the initial motion parameters using a preset slip coefficient to obtain the calibrated target motion parameters; and predict the robot's pose using the inertial navigation measurement data as the predicted value and the target motion parameters as the observed value to obtain the robot's initial pose.

[0165] In an exemplary embodiment, the pipeline environmental data includes temperature, humidity, and pressure; the compensation factor determination module 830 is specifically used to determine the refractive index compensation factor of the pipeline based on the temperature, humidity, and pressure at the current map construction time.

[0166] In an exemplary embodiment, the point cloud determination module 840 is specifically configured to: generate the robot's current trajectory sequence within the target time period based on the initial pose and the pipeline visual data within the target time period; determine the robot's pose set within the target time period based on the pre-integration results of the current trajectory sequence and the inertial navigation measurement data within the target time period; use the pose set to perform distortion compensation on the lidar point cloud within the target time period to obtain a compensated point cloud, and perform intensity correction on the compensated point cloud based on the refractive index compensation factor to obtain a corrected point cloud; and perform filtering processing on the corrected point cloud based on the radar echo data of each point in the corrected point cloud to obtain the initial map point cloud for the target time period.

[0167] In an exemplary embodiment, the map construction module 850 is specifically used to: divide the initial map point cloud into a first point cloud region and a second point cloud region according to a preset curvature threshold; wherein the curvature of the first point cloud region is less than the preset curvature threshold, and the curvature of the second point cloud region is not less than or greater than the preset curvature threshold; perform interpolation fusion on the boundary region between the first point cloud region and the second point cloud region to obtain an initial local map; and perform intensity correction on the initial local map according to the refractive index compensation factor to obtain a target local map for the target time period.

[0168] In an exemplary embodiment, the map update module 860 is specifically configured to: construct an optimized local map at the current map construction time based on the current trajectory sequence, the target local map, and the motion data corresponding to each trajectory point in the current trajectory sequence; in the initial global map at the previous map construction time, select a specified local map whose pipeline structure features match the pipeline structure features of the optimized local map; determine the overlap rate between the historical trajectory sequence corresponding to the specified local map and the current trajectory sequence; if the overlap rate is greater than a preset overlap rate threshold, correct the specified local map based on the target local map, and correct the initial global map based on the corrected specified local map to obtain the target global map at the current map construction time.

[0169] Each module in the aforementioned map building device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0170] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores various data collected by different sensors on the robot, local maps built at different times, global maps, and other data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a map building method.

[0171] In one exemplary embodiment, a computer device is provided, which may be a robot, and its internal structure diagram may be as follows: Figure 10 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a map-building method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0172] Those skilled in the art will understand that Figure 9 and Figure 10The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0173] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the various method embodiments of the map construction method described above.

[0174] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the various method embodiments of the map construction method described above.

[0175] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the various method embodiments of the map construction method described above.

[0176] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0177] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0178] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A map construction method, characterized in that, The method includes: During the robot's movement inside the pipeline, the system acquires lidar point cloud data, pipeline environment data, pipeline visual data, and robot motion data collected by different sensors mounted on the robot. Given the current map building time, the robot's initial pose is predicted based on the motion data at the current map building time, and the refractive index compensation factor of the pipeline is determined based on the pipeline environment data at the current map building time. Based on the initial pose, the refractive index compensation factor, and the lidar point cloud and pipeline visual data within the target time period, the initial map point cloud for the target time period is determined; wherein, the target time period is the time period between the current map construction time and the previous map construction time. According to a preset curvature threshold, the initial map point cloud is divided into a first point cloud region and a second point cloud region; wherein the curvature of the first point cloud region is less than the preset curvature threshold, and the curvature of the second point cloud region is not less than the preset curvature threshold. Interpolation and fusion are performed on the boundary region between the first point cloud region and the second point cloud region to obtain an initial local map; wherein, the boundary region is a region of a preset width centered on the boundary center of the first point cloud region and the second point cloud region; Based on the refractive index compensation factor, the initial local map is intensity corrected to obtain the target local map for the target time period; The initial global map at the time of the previous map construction is updated based on the target local map to obtain the target global map at the time of the current map construction.

2. The method according to claim 1, characterized in that, The motion data includes inertial navigation measurement data and wheel rotation data; predicting the robot's initial pose based on the motion data at the current map construction time includes: Based on the wheel rotation data at the current map construction time, the initial motion parameters of the robot are determined, and the initial motion parameters are corrected using a preset slip coefficient to obtain the calibrated target motion parameters. Using the inertial navigation measurement data as the predicted value and the target motion parameters as the observed value, the robot's pose is predicted to obtain the robot's initial pose.

3. The method according to claim 1, characterized in that, The pipeline environmental data includes temperature, humidity, and pressure; determining the refractive index compensation factor of the pipeline based on the pipeline environmental data at the current map construction time includes: The refractive index compensation factor of the pipeline is determined based on the temperature, humidity, and pressure at the time of map construction.

4. The method according to claim 1, characterized in that, The motion data includes inertial navigation measurement data; determining the initial map point cloud for the target time period based on the initial pose, the refractive index compensation factor, and the lidar point cloud and pipeline visual data within the target time period includes: Based on the initial pose and the pipeline visual data within the target time period, the current trajectory sequence of the robot within the target time period is generated; Based on the pre-integration results of the current trajectory sequence and the inertial navigation measurement data within the target time period, the robot's pose set within the target time period is determined; Using the pose set, distortion compensation is performed on the lidar point cloud within the target time period to obtain the compensated point cloud, and the intensity of the compensated point cloud is corrected according to the refractive index compensation factor to obtain the corrected point cloud. Based on the radar echo data of each point in the corrected point cloud, the corrected point cloud is filtered to obtain the initial map point cloud for the target time period.

5. The method according to claim 1, characterized in that, The step of updating the initial global map at the previous map construction time based on the target local map to obtain the target global map at the current map construction time includes: Based on the current trajectory sequence, the target local map, and the motion data corresponding to each trajectory point in the current trajectory sequence, an optimized local map is constructed at the current map construction time. In the initial global map at the previous map construction time, a specified local map is selected that matches the pipeline structure features of the optimized local map. Determine the overlap rate between the historical trajectory sequence corresponding to the specified local map and the current trajectory sequence; If the overlap rate is greater than a preset overlap rate threshold, the specified local map is corrected according to the target local map, and the initial global map is corrected according to the corrected specified local map to obtain the target global map at the current map construction time.

6. The method according to claim 1, characterized in that, The step of performing intensity correction on the initial local map based on the refractive index compensation factor to obtain the target local map for the target time period includes: Based on the refractive index compensation factor, the initial map point cloud is subjected to intensity correction to obtain a correction result; and based on the correction result, the initial local map is subjected to intensity correction to obtain the local map point cloud for the target time period, which serves as the target local map for the target time period.

7. A map building device, characterized in that, The device includes: The data acquisition module is used to acquire lidar point cloud, pipeline environment data, pipeline visual data and robot motion data collected by different sensors on the robot during the robot's movement in the pipeline. The pose prediction module is used to predict the initial pose of the robot based on the motion data at the current map construction time when the current map construction time is reached. The compensation factor determination module is used to determine the refractive index compensation factor of the pipeline based on the pipeline environment data at the current map construction time. The point cloud determination module is used to determine the initial map point cloud for the target time period based on the initial pose, the refractive index compensation factor, and the lidar point cloud and pipeline visual data within the target time period; wherein, the target time period is the time period between the current map construction time and the previous map construction time. A map construction module is used to divide the initial map point cloud into a first point cloud region and a second point cloud region according to a preset curvature threshold; wherein the curvature of the first point cloud region is less than the preset curvature threshold, and the curvature of the second point cloud region is not less than or greater than the preset curvature threshold; interpolation and fusion are performed on the boundary region between the first point cloud region and the second point cloud region to obtain an initial local map; and intensity correction is performed on the initial local map according to the refractive index compensation factor to obtain the target local map for the target time period. The map update module is used to update the initial global map at the previous map construction time based on the target local map, so as to obtain the target global map at the current map construction time.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images