Three-dimensional measurement data fusion device and method based on beidou positioning and laser scanning

Abstract

The application relates to the technical field of three-dimensional measurement and data processing, and discloses a three-dimensional measurement data fusion device and method based on Beidou positioning and laser scanning. The device comprises a Beidou positioning module, an inertial measurement unit, a laser scanning module comprising a pan-tilt scanner, a data fusion processing module, a data transmission module, and a data storage and visualization module. The data fusion processing module estimates the state vector of the device and the state covariance matrix representing the uncertainty thereof based on a kinematic model. When the uncertainty of the state vector exceeds a preset threshold, the module automatically generates a control instruction to drive the pan-tilt scanner to perform target scanning and obtain high-value observation data. The obtained high-value observation data is used to update the state vector through an observation noise covariance matrix, thereby forming a closed-loop control capable of quickly converging the uncertainty. The application improves the fusion precision, long-term stability, and reliability of the measurement system through adaptive weighting and online external parameter calibration technology.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-dimensional measurement and data processing, in particular to a three-dimensional measurement data fusion device and method based on Beidou positioning and laser scanning. BACKGROUND

[0002] In the field of large-scale ports, mines and automation engineering, high-precision real-time three-dimensional positioning and attitude measurement of large mobile equipment, and synchronous acquisition of three-dimensional geometric information of its working environment, are the key technical prerequisites for realizing automated and intelligent operation processes. At present, combining Global Navigation Satellite System (GNSS) (such as China's Beidou satellite navigation system) and laser scanning technology has become the main technical approach to realize such three-dimensional measurement. Generally, GNSS is used to provide absolute position information of large mobile equipment in the global coordinate system, and laser scanners are used to obtain high-density three-dimensional point cloud data of the environment around the large mobile equipment. By fusing the two heterogeneous data sources, theoretically, a comprehensive measurement result with global accuracy and local details can be obtained.

[0003] However, in practical applications, the existing technical solutions still have deficiencies in precision, reliability and environmental adaptability; on the one hand, the existing data fusion methods are mostly passive open-loop processing modes; such methods usually fuse the input data according to the preset fixed model and algorithm, lacking dynamic evaluation and feedback mechanism for real-time quality of sensor data and uncertainty of the fusion system itself state; when the Beidou positioning module has precision decline due to satellite signal obstruction, or the laser scanning module produces noise due to interference of environmental factors such as dust, rain and fog on site, the passive fusion system cannot adaptively adjust the weight of different data sources in the fusion process, resulting in significant impact of low-quality data on the overall measurement precision and reduced reliability; on the other hand, the long-term stability of multi-sensor combined measurement system also faces challenges. The relative pose relationship between sensors installed on large equipment, i.e. the external parameter, will change slowly due to factors such as vibration, structure thermal expansion and contraction caused by long-time operation of large mobile equipment. This drift of external parameters will introduce systematic errors that are difficult to detect, and traditional offline calibration methods not only have complicated operation and affect the normal operation of large mobile equipment, but also cannot compensate for dynamic deviations generated during operation in real time.

[0004] More importantly, when the positioning or orientation accuracy of the system decreases due to the above reasons, the prior art lacks the ability to actively restore the accuracy; the laser scanning strategy is usually fixed and cannot be intelligently adjusted according to the current needs of the system; the system cannot actively focus the scanning resources on stable environmental features that can most effectively constrain the current uncertainty growth direction, so it can only passively accept the reduction of measurement accuracy, and cannot form a closed-loop control of problem discovery, active perception and problem solving, which limits its application in automated operation scenarios that require continuous high reliability.

[0005] Therefore, the application proposes a three-dimensional measurement data fusion device and method based on Beidou positioning and laser scanning to solve the problems of the prior art. SUMMARY

[0006] In view of the deficiencies of the prior art, the application provides a three-dimensional measurement data fusion device and method based on Beidou positioning and laser scanning, which solves the problem of reduced positioning and orientation accuracy caused by fluctuations in sensor data quality or missing environmental features in three-dimensional measurement technology and the inability to actively compensate.

[0007] To achieve the above purpose, the application is implemented by the following technical solutions:

[0008] The three-dimensional measurement data fusion device based on Beidou positioning and laser scanning comprises:

[0009] a Beidou positioning module for obtaining global positioning information of a mobile equipment in a global coordinate system;

[0010] an inertial measurement unit for obtaining three-dimensional angular velocity and three-dimensional acceleration information of the mobile equipment;

[0011] a laser scanning module for obtaining three-dimensional point cloud information of the environment around the mobile equipment, wherein the laser scanning module comprises multiple laser scanners, and the laser scanners comprise long-range scanners and multi-layer scanners;

[0012] a data fusion processing module for receiving the global positioning information, three-dimensional angular velocity and three-dimensional acceleration information, and three-dimensional point cloud information, based on a kinematic model, and fusing the global positioning information, three-dimensional angular velocity and three-dimensional acceleration information, and three-dimensional point cloud information to estimate a state vector of the mobile equipment and a state covariance matrix representing the uncertainty of the state vector, and real-time monitoring the state covariance matrix, and generating a control instruction for driving the long-range scanner to perform specific area scanning when the uncertainty of the state vector exceeds a preset uncertainty threshold;

[0013] a data transmission module, configured to send the control instruction from the data fusion processing module to the laser scanning module, so as to drive the laser scanner to perform targeted scanning on the stable environmental feature, thereby obtaining observation data;

[0014] a data storage and visualization module, configured to store the state vector and update the state vector by using the observation data, wherein, in the updating process, an observation noise covariance matrix is set for the observation data, so as to accelerate the convergence of the uncertainty.

[0015] In a further technical solution, the data fusion processing module comprises a scanning strategy generation unit, which is configured to generate the control instruction, and specifically comprises:

[0016] monitoring the diagonal elements of the state covariance matrix in real time to obtain the variance of each state component;

[0017] when the variance of any state component is detected to exceed the uncertainty threshold, querying a preset prior environmental map to determine the stable environmental feature capable of constraining the maximum dimension of the current uncertainty;

[0018] generating the control instruction for driving the long-range scanner to point to the stable environmental feature according to the global coordinates of the stable environmental feature.

[0019] In a further technical solution, the data fusion processing module further comprises an adaptive fusion algorithm unit, which adopts an extended Kalman filtering framework, and fuses the global positioning information, the three-dimensional angular velocity and three-dimensional acceleration information, and the three-dimensional point cloud information to estimate the state vector of the mobile equipment and the state covariance matrix representing the uncertainty of the state vector, and realizes adaptive weighting of the fusion process by dynamically adjusting the observation noise covariance matrix, thereby obtaining the optimal state vector of the mobile equipment;

[0020] wherein, when the observation data is from the Beidou positioning module, the element value of the observation noise covariance matrix is inversely proportional to the satellite signal-to-noise ratio contained in the observation data;

[0021] when the observation data is from the laser scanning module, the element value of the observation noise covariance matrix is inversely proportional to the fitting goodness of the three-dimensional point cloud information registration.

[0022] In a further technical solution, the data fusion processing module further comprises an online calibration and monitoring unit and a space-time registration unit;

[0023] the online calibration and monitoring unit is configured to:

[0024] Real-time identify overlapping areas in three-dimensional point cloud information collected by different laser scanners in the laser scanning module;

[0025] Extract and match corresponding feature points in the overlapping areas;

[0026] Solve a nonlinear least squares optimization problem with the objective of minimizing re-projection error, and input the corresponding feature points to real-time solve the extrinsic parameters between the laser scanners;

[0027] The space-time registration unit is configured to convert the three-dimensional point cloud information from respective sensor coordinate systems to a unified global coordinate system using the extrinsic parameters.

[0028] In further technical solutions, the online calibration and monitoring unit is further configured to:

[0029] The root mean square error of the residual after solving the nonlinear least squares optimization problem is used as a health index of the extrinsic parameters of the laser scanners;

[0030] In the adaptive fusion algorithm unit, when the observation data comes from the laser scanning module, the element values of the observation noise covariance matrix are inversely proportional to the health index.

[0031] In further technical solutions, the Beidou positioning module includes a positioning data preprocessing unit configured to perform quality control before the data enters the data fusion processing module, and the quality control includes:

[0032] According to the type of positioning solution, only data frames with fixed solution are retained;

[0033] According to the average signal-to-noise ratio of each satellite participating in the solution, data frames with an average signal-to-noise ratio lower than a preset quality threshold are removed.

[0034] In further technical solutions, the laser scanning module further includes a point cloud data preprocessing industrial computer;

[0035] The point cloud data preprocessing industrial computer is configured to perform statistical outlier removal filtering and voxel grid downsampling to obtain structured three-dimensional point cloud information.

[0036] In further technical solutions, the long-range scanner is integrated on a two-dimensional gimbal;

[0037] The multi-layer scanner is fixedly installed on both sides of the cantilever structure of the mobile equipment of the three-dimensional measurement data fusion device.

[0038] In a further technical solution, the data transmission module adopts a redundant communication architecture combining optical fiber communication and industrial wireless communication, and through implementation of a parallel redundancy protocol, a copy of a data frame is transmitted through a primary link and a backup link concurrently, and a receiving end only processes the copy of the data frame that first arrives.

[0039] The application further provides a three-dimensional measurement data fusion method based on Beidou positioning and laser scanning, applied to the three-dimensional measurement data fusion device based on Beidou positioning and laser scanning.

[0040] S1, collecting global positioning information of a mobile equipment through a Beidou positioning module, and collecting three-dimensional angular velocity and three-dimensional acceleration information of the mobile equipment through an inertial measurement unit;

[0041] S2, collecting three-dimensional point cloud information of an environment around the mobile equipment through laser scanning modules of multiple laser scanners, wherein the laser scanners include long-range scanners and multi-layer scanners;

[0042] S3, estimating a state vector of the mobile equipment and a state covariance matrix representing uncertainty of the state vector based on a kinematic model and fusing the global positioning information, the three-dimensional angular velocity and the three-dimensional acceleration information and the three-dimensional point cloud information;

[0043] S4, monitoring the state covariance matrix in real time, and when the uncertainty of the state vector exceeds a preset uncertainty threshold, generating a control instruction for driving the laser scanners to perform specific area scanning;

[0044] S5, sending the control instruction to the laser scanning modules to drive the laser scanners to perform targeted scanning on stable environmental features, so as to obtain observation data;

[0045] S6, updating the state vector by using the observation data, wherein in the updating process, an observation noise covariance matrix is set for the observation data to accelerate convergence of the uncertainty.

[0046] The application provides a three-dimensional measurement data fusion device and method based on Beidou positioning and laser scanning, and has the following beneficial effects:

[0047] 1、The device of the application can generate control instructions automatically in real time by monitoring the state covariance matrix representing the state uncertainty through the data fusion processing module, and driving the gimbal scanner to perform targeted scanning on stable environmental features that can most effectively constrain the uncertainty when the uncertainty exceeds the preset uncertainty threshold; The closed-loop control mechanism ensures that the device can quickly converge the uncertainty by actively acquiring high-value observation data in challenging scenarios where the quality of some sensor data is declining or the effective environmental features are sparse, thereby ensuring the continuous high accuracy of the output state vector.

[0048] 2、The application uses an adaptive fusion algorithm unit in the update phase of the extended Kalman filter, rather than fixed noise parameters, but dynamically adjusts the observation noise covariance matrix, which is based on the satellite signal-to-noise ratio of the Beidou positioning information, the fitting goodness of the laser point cloud registration, and the sensor health index output by the online calibration and monitoring unit. Through this adaptive weighting method, the device can quantify and utilize the data quality of each observation, automatically suppress the pollution of low-quality data on the fusion result, and output a more accurate and reliable state vector.

[0049] 3、The online calibration and monitoring unit of the application can calculate and update the external parameters between each scanner in the laser scanning module in real time, automatically compensate for the installation pose drift caused by mobile equipment vibration, structural deformation or temperature change; This online self-calibration capability without interrupting the operation and without manual intervention avoids the accumulation of systematic measurement errors caused by sensor external parameter misalignment, ensuring the data consistency and precision reliability of the three-dimensional measurement data fusion device throughout its life cycle. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 The overall architecture diagram of the three-dimensional measurement data fusion device based on Beidou positioning and laser scanning of the embodiment of the application;

[0051] Figure 2 The method flowchart of the embodiment of the application.

[0052] REFERENCE NUMERALS

[0053] 10、Beidou positioning module; 20、laser scanning module; 30、data fusion processing module; 40、data transmission module; 50、data storage and visualization module. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the specification of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0055] Figure 1 is a schematic diagram of the overall architecture according to an embodiment of the present application. As shown in Figure 1 the embodiment of the present application provides a three-dimensional measurement data fusion device based on Beidou positioning and laser scanning. The device can include a Beidou positioning module 10, a laser scanning module 20, a data fusion processing module 30, a data transmission module 40, and a data storage and visualization module 50.

[0056] The Beidou positioning module 10 and the laser scanning module 20 are data acquisition sources, respectively used to obtain global positioning information of mobile equipment and three-dimensional point cloud information of the surrounding environment. The two modules send the collected raw data to the data fusion processing module 30 through the data transmission module 40. The data fusion processing module 30 processes and fuses the received data, outputs high-precision mobile equipment state and global point cloud data, and sends them to the data storage and visualization module 50 through the data transmission module 40 for storage and subsequent upper function application, such as realizing automatic calculation of material pile volume, mobile equipment anti-collision warning and other functions. At the same time, the data fusion processing module 30 can generate control instructions according to the uncertainty of the fusion state, and return them to the laser scanning module 20 through the data transmission module 40, so as to actively control the long-range scanner in it to perform target scanning on a specific area, thereby realizing closed-loop correction of state estimation.

[0057] In a preferred embodiment, the Beidou positioning module 10 is used to obtain high-precision real-time position, speed and time information of the mobile equipment in the global coordinate system. The Beidou positioning module 10 specifically includes at least one Beidou high-precision positioning terminal supporting real-time kinematic (RTK) technology, a high-gain antenna, and a positioning data preprocessing unit. The high-gain antenna is installed at a position on the mobile equipment body that is not easily blocked, such as the top of the door leg or the root of the cantilever structure of large port machinery, to ensure good satellite signal reception effect. The positioning data preprocessing unit is used to preliminarily process the raw data output by the positioning terminal, filter out unreliable positioning data (i.e. gross errors) caused by signal interference or blocking according to the signal-to-noise ratio and other quality indicators of the satellite signal, so as to obtain a positioning data sequence that does not contain significant outliers and is purer, and provide stable and reliable global positioning observation values to the data fusion processing module 30.

[0058] The laser scanning module 20 is used to obtain high-density and high-precision three-dimensional point cloud data of the environment around the mobile equipment, and includes at least two laser scanners and a point cloud data preprocessing industrial computer. In a specific embodiment, the laser scanners can be fixedly installed on both sides of a cantilever structure of the mobile equipment, or integrated on a high-precision two-dimensional gimbal and installed on the top of the mobile equipment, to collect fine local point clouds of the working area, and can receive control instructions from the data fusion processing module 30 to realize target scanning of a specific area. The point cloud data preprocessing industrial computer is used to perform denoising and downsampling processing on the original point cloud data collected by each scanner, for example, using a statistical filtering algorithm to remove noise points caused by suspended particulate matter in the environment, and using a voxel grid downsampling algorithm to reduce the data amount while ensuring the structural features of the point cloud.

[0059] The data fusion processing module 30 internally integrates a plurality of functional units, including an online calibration and monitoring unit, a space-time registration unit, an adaptive fusion algorithm unit, and a scanning strategy generation unit. The online calibration and monitoring unit is used to calculate the relative installation position and attitude relationship (i.e., external parameters) between each scanner in the laser scanning module 20 in real time, and monitor the stability of these relative positions and attitudes (external parameters). The space-time registration unit is used to accurately register data from different sensors to the same spatial coordinate system based on a unified time reference, using real-time external parameters output by the online calibration and monitoring unit. The adaptive fusion algorithm unit is used to perform an extended Kalman filtering algorithm to fuse Beidou positioning data and laser radar data, and estimate the optimal state vector of the mobile equipment, including three-dimensional position, velocity and attitude. The scanning strategy generation unit is used to monitor the state uncertainty output by the adaptive fusion algorithm unit in real time, and when the uncertainty exceeds a preset uncertainty threshold, the scanning strategy generation unit generates a control instruction to instruct a long-range scanner in the laser scanning module 20 to perform a specific scanning task.

[0060] The data transmission module 40 is used to realize reliable exchange of data between the internal hardware modules of the device. In a preferred embodiment, the data transmission module 40 adopts a redundant communication architecture combining optical fibers and industrial wireless APs. By deploying an industrial switch supporting a high-deterministic network protocol, it is ensured that when the main link (optical fiber) fails, it can be seamlessly switched to the standby link (industrial wireless), thereby ensuring low-latency and high-reliability transmission of large-capacity point cloud data streams and real-time control instructions.

[0061] The data storage and visualization module 50 is a human-computer interaction and data archiving terminal of the device; the module specifically includes a data storage server and a three-dimensional visualization platform; the data storage server is used for long-term archiving of raw data collected by the Beidou positioning module 10 and the laser scanning module 20, and the final fusion result output by the data fusion processing module 30; the three-dimensional visualization platform loads a geographic information system map, can render the accurate motion trajectory of the mobile equipment in real time in three-dimensional form, dynamically construct a three-dimensional model of the work scene generated by the fused point cloud, and realize upper application functions such as automatic calculation of the volume of the material pile and mobile equipment anti-collision warning on this basis.

[0062] In a specific embodiment, the Beidou positioning module 10 is used for continuously acquiring the three-dimensional position, three-dimensional speed, attitude heading and accurate Coordinated-Universal-Time (UTC) timestamp of the mobile equipment in a preset global coordinate system (for example, WGS-84 or Beijing 54 coordinate system).

[0063] The Beidou positioning module 10 specifically includes a set of global navigation satellite system (GNSS) receiving units supporting real-time dynamic positioning (RTK) of multiple systems and multiple frequency points; the GNSS receiving unit contains one or more GNSS receiver terminals, a GNSS ground reference station (reference station) and a high-gain GNSS antenna connected with each receiver terminal; the GNSS receiver terminal is preferably an industrial-grade product capable of simultaneously receiving and processing satellite signals of multiple constellations such as Beidou satellite navigation system (BDS), global positioning system (GPS), GLONASS and Galileo; the high-gain GNSS antenna, for example, a choke ring antenna, has the ability to suppress multipath effects, is installed on the mobile equipment body in a position where the signal is not easily blocked by the structure itself, such as the top center or the top of the operating room of a large engineering machinery.

[0064] The working principle of the Beidou positioning module 10 is based on carrier phase differential technology; a GNSS ground reference station is erected on a known accurate coordinate point, continuously receives satellite signals, and compares them with its own known coordinates, thereby real-time solving the comprehensive errors caused by ionospheric / tropospheric delay, satellite orbit error, and satellite clock error; the ground reference station generates differential correction data from these error calculations and broadcasts it to the GNSS receiver terminal installed on the mobile equipment through a wireless data link (such as a UHF radio or a 4G / 5G mobile network); the receiver terminal receives and applies the differential correction data while receiving its own satellite signals, eliminates common mode errors by solving the integer ambiguity of the carrier phase, thereby obtaining real-time positioning accuracy of centimeters; multi-constellation fusion solving is adopted, which increases the number of available satellites and optimizes the satellite geometric distribution, thereby enhancing the availability, continuity, and reliability of the positioning solution, especially in complex environments such as ports where there is partial obstruction.

[0065] The Beidou positioning module 10 further comprises a positioning data preprocessing unit, which is implemented at the software or firmware level, for quality control and optimization of the data before it enters the data fusion processing module 30; the detailed processing steps of the positioning data preprocessing unit include:

[0066] S101: Receive the raw positioning data frames output in real time by the GNSS receiver terminal; each data frame contains a three-dimensional position coordinate vector , a three-dimensional velocity vector , a high-precision timestamp , and a state parameter representing the solution quality, which includes the positioning solution type (such as fixed solution, floating solution, single-point solution) and the signal strength indicators of each satellite involved in the solution, such as signal-to-noise ratio ;

[0067] S102: Preliminary screening according to the positioning solution type; the positioning data preprocessing unit checks the solution type of each data frame and only retains the data frame with the highest precision level of fixed solution; any floating solution, single-point solution, or other low-precision type of positioning data is considered as invalid observation and directly discarded to prevent low-quality data from polluting the subsequent fusion filtering process;

[0068] S103: Secondary discrimination according to signal quality indicators; for the fixed solution data screened in step S102, the positioning data preprocessing unit further evaluates its reliability; specifically, the average signal-to-noise ratio of all satellites involved in this solution is calculated; if the average signal-to-noise ratio is lower than a preset quality threshold (such as 35 dB), it is considered that the positioning result is a fixed solution but obtained in a weak signal environment, its reliability is questionable, and it is also excluded;

[0069] S104: performing smoothing filtering on the valid positioning data sequence passing through all filtering steps; the positioning data preprocessing unit adopts a moving average filtering algorithm based on a sliding time window to suppress random high-frequency noise existing in the valid data; the specific implementation is as follows:

[0070] ;

[0071] wherein, is the current discrete time step; is the time step is the valid position coordinate vector passing through the above filtering steps; is the size of the sliding window, which is a positive integer, and the value can be configured according to the motion characteristics of the mobile equipment; is the time step is the output smoothed position coordinate vector.

[0072] The same filtering process is also applied to the velocity vector After the above steps, the positioning data preprocessing unit outputs a more pure, smooth and reliable positioning observation data stream for the data fusion processing module 30.

[0073] The laser scanning module 20 is used to acquire three-dimensional space point cloud data of the environment around the mobile equipment in real time, and to pre-process the original point cloud data; the laser scanning module 20 provides high-quality, structured three-dimensional geometric information for subsequent data fusion and scene three-dimensional reconstruction.

[0074] In a preferred embodiment, the laser scanning module 20 includes at least two laser scanners and a dedicated point cloud data preprocessing industrial computer; the laser scanners can be divided into two types according to their installation position and function positioning on the mobile equipment.

[0075] The laser scanners have a 360-degree horizontal field of view and a multiple echo recognition function, a part of the laser scanners are fixedly installed on the key motion components of the mobile equipment, such as the two sides of the cantilever structure of the bucket-wheel stacker-reclaimer, for collecting fine local point clouds of the working area; another part of the laser scanners are installed on a high-precision two-dimensional servo cloud platform and fixedly installed at the top of the mobile equipment, for receiving and executing specific control instructions from the data fusion processing module 30 to realize targeted scanning of specific targets or areas.

[0076] The laser scanning module 20 further comprises a point cloud data preprocessing industrial computer, which is a high-performance industrial computer running a point cloud processing software environment internally, such as a software program based on a Linux operating system and a Point-Cloud-Library (PCL); the industrial computer is used to receive raw point cloud data output by all laser scanners and execute a series of preprocessing algorithms; the preprocessing process specifically comprises the following steps:

[0077] S201: receiving raw three-dimensional point cloud data frames output by each laser scanner with accurate time stamps . .

[0078] S202: performing a Statistical-Outlier-Removal (SOR) filtering on the raw three-dimensional point cloud data frames, which removes isolated noise points generated by dust, water vapor suspended in the air or the sensor itself, the principle of which is that, for each point in the raw three-dimensional point cloud data frames, the average distance of the point to its nearest neighbor points is calculated; then, the mean value and the standard deviation of the distribution of the average distances of all points are calculated; a point is determined to be an outlier if its average distance exceeds a threshold value determined by global statistical characteristics: ; wherein: is the current point to be judged; is the average distance of the point to its nearest neighbor points; is the mean value of the average distances of all points in the point cloud; is the standard deviation of the average distances of all points in the point cloud; is a threshold value of the standard deviation multiple, which is a configurable parameter.

[0079] All points satisfying the above condition are removed from the raw three-dimensional point cloud data frames to obtain a cleaner point cloud ;

[0080] S203: performing a voxel grid downsampling on the filtered point cloud ; the step reduces the data volume of the point cloud under the premise that the macroscopic geometric structure of the point cloud is unchanged to meet the real-time requirements of subsequent algorithms; the principle of which is that the entire three-dimensional space in which the point cloud is located is divided into a grid composed of a large number of tiny cubes (i.e., voxels); for all points in each non-empty voxel , the geometric center (centroid) of the points is calculated , and the centroid point is used to represent all points in the voxel. The formula for calculating the centroid is: ; wherein: is the i-th non-empty voxel; is the number of points contained in the voxel is the i-th point in the voxel is the centroid point of the voxel calculated; The centroid points of all voxels together constitute the final down-sampled point cloud ; After the above preprocessing steps, the point cloud data preprocessing industrial computer sends the clean and structured point cloud data stream with the original timestamp to the data fusion processing module 30.

[0081] As shown in , the data fusion processing module 30 can be physically deployed on one or more high-performance and high-reliability industrial computers. To ensure uninterrupted operation for 7x24 hours in a harsh industrial environment, the hardware platform is preferably a server with redundant design, such as a central processor with fault-tolerant support, large-capacity error correction code (ECC) memory, and redundant power supply. The data fusion processing module 30 receives data from the Beidou positioning module 10 and the laser scanning module 20, and performs a series of complex algorithms to output high-precision and high-reliability mobile equipment status and globally time-space registered three-dimensional point cloud data.

[0082] In one specific embodiment, the functions of the data fusion processing module 30 are realized by a plurality of software units integrated therein; these software units include: an online calibration and monitoring unit, a time-space registration unit, an adaptive fusion algorithm unit, and a scanning strategy generation unit. Figure 1 The online calibration and monitoring unit exists as a continuously running background process, and its function is to real-time calibrate and monitor the relative pose relationship (external parameters) between the scanners in the laser scanning module 20 to compensate for installation errors caused by mobile equipment vibration, structural deformation, or temperature changes. The execution steps of the online calibration and monitoring unit include:

[0083] S301: According to the nominal installation pose and field of view parameters of each scanner, the overlapping area of the point cloud data collected by different scanners (such as long-range scanners and multi-layer scanners) in three-dimensional space is calculated and identified in real time.

[0084] S302: According to the overlapping area of the point cloud data calculated in S301, the relative pose relationship (external parameters) between the scanners is calculated and updated in real time.

[0085] S303: According to the updated relative pose relationship (external parameters) between the scanners, the point cloud data collected by different scanners is time-space registered.

[0086] S304: According to the time-space registered point cloud data, the online calibration and monitoring unit generates a series of real-time monitoring reports, and sends them to the data fusion processing module 30.​S302: Within the identified overlapping region, stable geometric feature points are independently extracted from the two point cloud datasets and described using descriptors such as Fast-Point-Feature-Histograms (FPFH). Subsequently, a set of corresponding feature point pairs is established using robust matching algorithms such as random sampling consistency. .

[0087] S303: Calculate the optimal relative rotation matrix by solving a nonlinear least squares optimization problem. With translation vector The objective function is to minimize the reprojection error between all pairs of points with the same name. ;in: The optimal solution to this optimization problem, together constitute the solution from the scanner. Coordinate system to scanner Optimal rigid body transformation of the coordinate system. Among them, This is the optimal 3x3 rotation matrix. The optimal 3x1 translation vector; This represents an optimization operation aimed at finding parameters that minimize the objective function. and This operation will solve for the rotation matrix. Constraints in a 3D special orthogonal group The inner set (i.e., the set of all 3x3 rotation matrices) and the translation vector to be solved. Constrained in three-dimensional real vector space Inside;

[0088] This represents the total number of identical feature point pairs established in step S302. For from the scanner The Coordinates of one feature point; For it in the scanner The coordinates of the corresponding points with the same name in the table; This represents the square of the L2 norm (i.e., the Euclidean norm) of a vector; in this formula, it calculates the source point after rotation and translation transformations. Its corresponding target point The square of the Euclidean distance between them.

[0089] S304: The optimal extrinsic parameters obtained from the solution The update is into the device for the space-time registration unit, while the optimized residual mean square error is calculated as the health index of the sensor external parameter. If the index is stable at a low level for a long time, it indicates that the sensor installation state is good. If the index shows a persistent increase or mutation, it is determined that the sensor external parameter has been misaligned. The data fusion processing module 30 will reduce the weight of its data in subsequent fusion, and can issue a maintenance alarm to the data storage and visualization module 50.

[0090] The space-time registration unit is used to unify all heterogeneous and asynchronous sensor data to the same space-time reference. The execution steps of the space-time registration unit include:

[0091] S311: Time synchronization. For any point cloud data with a timestamp of , the mobile equipment pose at the point cloud collection moment is obtained by linear interpolation between the timestamps and of the two nearest Beidou positioning modules 10 before and after it, so as to realize millisecond-level time alignment.

[0092] S312: Space registration. Each preprocessed frame of point cloud is converted from its respective sensor coordinate system to the unified global coordinate system. For any point in the local coordinate system of the laser scanner , its coordinate in the global coordinate system is calculated by the following formula: ; wherein: is the external parameter of the laser scanner relative to the mobile equipment body coordinate system , which is provided and updated in real time by the online calibration and monitoring unit; is the pose of the mobile equipment body coordinate system in the global coordinate system, which is provided by the best state estimation output by the adaptive fusion algorithm unit at the last time.

[0093] The adaptive fusion algorithm unit adopts an Extended-Kalman-Filter (EKF) framework to optimally fuse multi-source data to estimate the complete state of the mobile equipment. First, the state vector of the device is defined as , which at least includes the three-dimensional position , three-dimensional velocity and unit quaternion representing the attitude of the mobile equipment in the global coordinate system. The filtering process is divided into two stages of prediction and update.

[0094] During the prediction phase, based on the kinematic model of the mobile equipment, readings from the internal Inertial Measurement Unit (IMU) are used. This IMU measures the three-dimensional angular velocity and three-dimensional acceleration of the mobile equipment body, which is equipped with a three-dimensional measurement data fusion device, in real time at a high frequency (e.g., 100Hz or higher). The output data of the IMU is directly fed into the adaptive fusion algorithm unit as the basic input for performing the extended Kalman filter prediction step, providing crucial dynamic information for solving the kinematic model. Especially when the BeiDou positioning signal is briefly interrupted or its quality degrades, it can ensure the continuity and smoothness of the state estimation from the previous moment. State estimation Predict the current moment Prior state and its covariance : ; ;in: For IMU at time The measured value; It is a nonlinear state transition function; for The Jacobian matrix for the state; Let be the process noise covariance matrix.

[0095] During the update phase, when new external observations (such as position observations from BeiDou positioning module 10) are available... Or pose observations obtained by matching laser point clouds with prior maps. Upon arrival, the prior estimate is corrected using the following formula to obtain the optimal posterior estimate. and its covariance : ; ; ;in: This represents the observation vector at the current moment; For observation model functions; for The Jacobian matrix for the state; Kalman gain; Represent an identity matrix;

[0096] The key to this adaptive fusion algorithm unit lies in the observation noise covariance matrix. It is adaptively adjusted. Specifically, when the observation comes from Beidou positioning module 10, The element values ​​and the satellite signal-to-noise ratio calculated in this positioning solution Inversely proportional; when the observation comes from lidar, The element value of the matrix is inversely proportional to the goodness of fit of the point cloud registration and the sensor health index output by the online calibration and monitoring unit. In this way, the device automatically trusts high-quality observation data and suppresses poor-quality observation data, achieving adaptive weighting in the fusion process.

[0097] The scanning strategy generation unit is beneficial to the active sensing capability of the device, and the execution steps of the scanning strategy generation unit include:

[0098] S321: Real-time monitoring of the posterior state covariance matrix output by the adaptive fusion algorithm unit The diagonal elements of the matrix directly quantify the variance, i.e., the degree of uncertainty, of the first state component (e.g., X-direction position, pitch angle).

[0099] S322: Comparing the variance of each state component with a preset uncertainty threshold vector related to the task accuracy requirement; if the uncertainty of any state exceeds its threshold, i.e. , it is determined that the positioning or orientation accuracy of the device in that degree of freedom is decreasing and active intervention is needed.

[0100] S323: When active intervention is triggered, the scanning strategy generation unit queries the prior environmental map (e.g., GIS map or historical three-dimensional point cloud map) preloaded in the data storage and visualization module 50, searches for long-existing stable environmental features, such as the right-angle edges of buildings, fixed lighthouses, or track lines, near the current estimated position of the mobile equipment, which can most effectively constrain the current uncertainty in the largest dimension.

[0101] S324: The scanning strategy generation unit calculates the pitch angle and yaw angle that the long-range scanner in the laser scanning module 20 needs to point to according to the global coordinates of the selected target feature, generates the corresponding control instructions, and sends them to the scanner through the data transmission module 40. The scanner executes the instructions and performs a high-density, high-precision targeted scan. The obtained point cloud data is given a very high weight (i.e., a very small observation noise ) when updating the EKF, so that the divergent state uncertainty can be quickly and significantly converged, and the device can be pulled back to a high-precision working state.

[0102] Referring to Figure 1The data transmission module 40 establishes a bidirectional data communication link between the hardware modules (including the Beidou positioning module 10, the laser scanning module 20, the data fusion processing module 30, and the data storage and visualization module 50) inside the device. The function of the data transmission module 40 is to ensure that the large-capacity three-dimensional point cloud data, high-frequency positioning data, and control commands and synchronization signals with extremely high real-time requirements generated by the device can be transmitted between units with low delay and high reliability.

[0103] In a preferred embodiment, the data transmission module 40 adopts a physically redundant and logically seamless switching communication architecture. At the physical level, the architecture deploys at least two independent communication links. The first link, as the primary link, adopts high-bandwidth industrial-grade optical fiber communication. For scenarios requiring data exchange between rotating parts (such as the slewing mechanism of a stacker-reclaimer) and fixed parts of mobile equipment, the optical fiber link is connected through an industrial-grade optical fiber slip ring, ensuring stable transmission of optical fiber signals during continuous rotation of mobile equipment. The second link, as a parallel backup link, adopts industrial wireless communication technology, such as deploying enterprise-level wireless access points (APs) and clients conforming to IEEE-802.11ac or higher standards to provide high-bandwidth wireless network coverage.

[0104] At the logical level, to achieve high-reliability and zero-interruption switching between the two physical links, the module adopts a parallel redundancy protocol (PRP), such as the protocol defined in IEC62439-3 standard. The implementation of this protocol relies on network adapters or switches that support the protocol at both the data sending and receiving ends. The specific working steps are as follows:

[0105] S401: Any node that needs to send data (such as the data fusion processing module 30 mounted on mobile equipment) duplicates each data frame it generates into two identical copies at the network layer;

[0106] S402: The sending node sends one copy of the data frame through the first physical link (i.e., the optical fiber link);

[0107] S403: At the same time, the sending node concurrently sends another identical copy of the data frame through the second physical link (i.e., the industrial wireless link);

[0108] S404: At the receiving end (e.g. the data storage and visualization module 50 located in the central control room), its network interface simultaneously listens to data from both physical links. When the first valid data frame from either link is received, the receiving end node immediately passes this data frame to the upper layer application and records the unique identifier of the data frame;

[0109] S405: When the receiving end node subsequently receives a duplicate data frame with the same unique identifier from the other link, it recognizes this as a redundant copy and discards it directly without passing it to the upper layer application.

[0110] Through this parallel redundancy mechanism, when any link (e.g. the optical fiber link is interrupted due to physical damage) fails, data transmission will not be interrupted because the receiving end can always receive data from the other normal link (industrial wireless link); from the perspective of the upper layer application, the data stream is completely continuous, thus achieving zero switching time fault recovery and ensuring high availability of data transmission.

[0111] To further ensure the quality of service of different types of data transmission, the data transmission module 40 also implements a Quality-of-Service (QoS) policy; specifically, by using industrial switches that support Virtual-Local-Area-Network (VLAN) technology, different priorities are assigned to different types of data streams; for example, for control instructions generated by the scanning strategy generation unit to control the long-range scanner, and synchronization signals for multi-sensor time synchronization, the device marks a higher Priority-Code-Point (PCP) for their data frames; while for three-dimensional point cloud data streams with huge data volume but not extremely sensitive to single frame delay, a lower priority is marked; the switch will prioritize high-priority data frames when processing data forwarding, ensuring that the transmission delay of critical control and synchronization information is minimized, even in the case of extremely high network load.

[0112] As shown in Figure 1 , the data storage and visualization module 50 serves as the final output of the device and the human-machine interface, used for permanent archiving of the results generated by the data fusion processing module 30, and real-time display and application analysis in an intuitive three-dimensional form.

[0113] In a preferred embodiment, the data storage and visualization module 50 is composed of one or more data storage servers and a three-dimensional visualization platform; the hardware platform of the data storage server is preferably an enterprise-level server, which is configured with a large-capacity, high-reading and writing speed disk array (such as RAID5 or RAID6) to ensure the redundancy backup and security of data; the server implements a specific data archiving strategy, the specific steps of which include:

[0114] S501: Classify and store the received data, the server creates different storage paths for storing the raw data collected by the Beidou positioning module 10 and the laser scanning module 20, and the final result data output by the data fusion processing module 30, respectively. This separate storage method facilitates problem tracing, algorithm debugging, and historical data playback in the later stage;

[0115] S502: Establish a structured data organization method, for raw data, organize folders according to the hierarchical structure of sensor type, mobile equipment number, and collection time, for result data, store the high-frequency output mobile equipment state vector (including position, speed, attitude, timestamp, etc.) into a time series database for quick retrieval and time series analysis; store the huge amount of global point cloud data in a file format with timestamp (such as PCD or LAS format);

[0116] S503: Perform the preset data life cycle management, set the data retention period according to the storage capacity and business needs, for example, retain all data on the online storage server for the past three months for immediate access, and automatically transfer data exceeding three months to lower-cost nearline or offline storage media for long-term archiving;

[0117] The three-dimensional visualization platform is a software application developed based on a real-time three-dimensional graphics engine (such as Unity3D or Unreal-Engine), which is a client that subscribes to the required data stream from the data storage server or directly from the data fusion processing module 30 through the data transmission module 40, and performs real-time rendering and function implementation in a virtual three-dimensional scene. The core function implementation steps of the three-dimensional visualization platform are as follows:

[0118] S511: Load and display basic geographic information, when the three-dimensional visualization platform starts, first load a two-dimensional or three-dimensional Geographic-Information-System (GIS) base map consistent with the global coordinate system used by the device; the base map can include factory roads, building outlines, fixed mobile equipment, and other static environmental elements, providing a geographical spatial context for subsequent dynamic data visualization;

[0119] S512: Real-time rendering of mobile equipment motion trajectory, the three-dimensional visualization platform receives and parses the mobile equipment state vector output by the adaptive fusion algorithm unit in real time, and updates the spatial pose of a preset mobile equipment three-dimensional model in the three-dimensional scene according to the position vector in each frame of received state and attitude quaternion , the three-dimensional visualization platform updates the spatial pose of a preset mobile equipment three-dimensional model in the three-dimensional scene in real time, thereby accurately and smoothly reproducing the actual motion state of the mobile equipment; at the same time, a trajectory line can be drawn behind the model to show its historical motion path;

[0120] S513: Dynamically constructing and updating a three-dimensional scene model, the three-dimensional visualization platform receives the three-dimensional point cloud data stream output by the space-time registration unit, which has been unified to the global coordinate system, the three-dimensional visualization platform internally maintains a global three-dimensional data structure, such as an Octree or a voxel grid, when a new point cloud frame arrives, the three-dimensional visualization platform inserts the points in it into the data structure, realizing incremental updating of the scene model, for the areas that are repeatedly scanned, the new point cloud can be used to update or encrypt the existing area, thereby dynamically reflecting the changes of the scene, such as the accumulation or taking process of the stockpile, in order to obtain better visualization effect, the three-dimensional visualization platform can generate a triangular mesh model in real time based on the point cloud data structure, using algorithms such as Marching-Cubes, to display the surface of irregular objects such as stockpiles in solid form;

[0121] S514: Real-time collision warning, the three-dimensional visualization platform defines virtual bounding boxes or more accurate collision bodies for the three-dimensional model of the mobile equipment and obstacles (other mobile equipment, buildings, etc.) identified from the GIS map or real-time point cloud in the virtual scene; the three-dimensional visualization platform continuously detects whether there is spatial interference or the distance between the collision bodies of the mobile equipment model and other objects in the scene is less than the preset safety threshold at a high frequency (such as 30 times per second); when a potential collision risk is detected, the three-dimensional visualization platform immediately triggers an alarm, such as highlighting the collision risk area on the screen and sounding an audible alarm.

[0122] S515: Realize the automatic calculation of the volume of the job object; the three-dimensional visualization platform provides interactive tools to allow users to frame the boundary of a specific job object (such as a certain pile) in a two-dimensional map or a three-dimensional scene; after the user confirms the range, the three-dimensional visualization platform performs the following calculations: first, extract all three-dimensional point sets located within the user-defined boundary from the dynamically updated global three-dimensional scene model; then, define a reference surface, which can be a pre-measured yard ground elevation model or a user-specified horizontal plane; finally, calculate the volume using the grid-based digital elevation model (DEM) method, that is, project the selected point set onto a two-dimensional horizontal grid, calculate the average height of the points in each grid cell, multiply the area of the grid cell by the height of the reference surface at that location, and subtract the height of the reference surface at that location to obtain the volume of each grid column. The total volume of the pile is obtained by summing the volumes of all grid columns; the calculation result can be displayed and updated periodically in real time.

[0123] The implementation process of the adaptive cooperative three-dimensional measurement method provided by the embodiment of the application will be described in detail below with reference to the accompanying drawings. As shown in Figure 2 Figure 2 is a method flowchart according to an embodiment of the application, which mainly includes the following steps:

[0124] Step S1: Device initialization and online calibration of external parameters; in a specific embodiment, when the device is started, the device initialization and online calibration of external parameters are first performed; in this step, the data fusion processing module 30 loads the initial or last time calibrated external parameters (including rotation and translation parameters) of each laser scanner from the local configuration file, and at the same time, as a separate and continuously running thread in the background, the online calibration and monitoring unit is activated; the online calibration and monitoring unit concurrently performs a series of operations: real-time data acquisition from the laser scanning module 20, automatic recognition of overlapping areas in the fields of view of different scanners; extraction and matching of robust geometric feature points in these overlapping areas; based on the matched point pair set, the current optimal sensor external parameters are calculated by optimization; finally, the online calibration and monitoring unit evaluates the health of the external parameters of each sensor according to the residual of the optimization result, and this online calibration process is uninterrupted during the entire operation of the device, ensuring the real-time accuracy of the external parameters.

[0125] ​Step S2: Concurrent data acquisition and preprocessing, in this step, the Beidou positioning module 10 and the laser scanning module 20 work independently in parallel and high frequency; the Beidou positioning module 10 obtains the original positioning data containing position, speed, satellite signal quality and other information through its GNSS receiver, and filters and smooths it by the internal positioning data preprocessing unit to output stable and reliable global positioning observation values; at the same time, all laser scanners in the laser scanning module 20 collect three-dimensional point cloud data within their respective field of view, and send the original point cloud data to the point cloud data preprocessing industrial computer in real time, which performs outlier rejection and data downsampling operations to output structured point cloud with lower noise and more refined data volume;

[0126] Step S3: Real-time high-precision space-time registration is performed; the space-time registration unit in the data fusion processing module 30 receives the preprocessed positioning data stream and point cloud data stream from the previous step S2 concurrent data acquisition and preprocessing step; for each frame of point cloud data, the space-time registration unit calculates the mobile equipment pose at the moment of collecting the frame of point cloud to millisecond level according to its timestamp in the positioning data stream by interpolation algorithm; then, the space-time registration unit calls the latest external parameters updated by the online calibration and monitoring unit in the device initialization and external parameter online calibration step of step S1, and combines the mobile equipment pose obtained by interpolation to accurately convert each point in the frame of point cloud from its sensor's local coordinate system to the unified global coordinate system, completing the space-time reference unification of all data;

[0127] Step S4: Adaptive data fusion and state estimation is performed; this is the core iterative cycle of the method, in a complete iteration cycle, the adaptive fusion algorithm unit first predicts the state (position, speed, attitude, etc.) of the mobile equipment at the current time based on the readings of the inertial measurement unit and the kinematic model of the mobile equipment, to obtain a prior state estimation and its uncertainty (covariance matrix); when a new, space-time registered observation data (such as a GNSS positioning point or a pose obtained by laser radar point cloud registration) arrives, the adaptive fusion algorithm unit does not update immediately, but first sets the noise covariance matrix of the observation according to the quality index of the observation data, such as the signal-to-noise ratio of the GNSS observation, or the point cloud matching score and sensor health of the laser radar observation; a high-quality observation will be given a small noise value (high weight), and vice versa; after completing the noise setting, the adaptive fusion algorithm unit performs the update step of extended Kalman filter, combines the prior estimation with the observation with weight to calculate the optimal posterior state estimation and its updated covariance matrix at the current time

[0128] Step S5: Uncertainty-based active scanning closed-loop control; This method executes this step in parallel with the above state estimation, triggering a condition; Under normal operating conditions, when the state covariance output by the adaptive fusion algorithm unit... Within the preset accuracy range, this active scanning closed-loop control step is in a silent monitoring state; the scanning strategy generation unit in the data fusion processing module 30 continuously checks the covariance matrix. The diagonal elements represent the variance of each state component. When the variance of any state component exceeds the preset task accuracy threshold, it indicates that the uncertainty of the device in that degree of freedom is growing uncontrollably, and the active scanning closed-loop control step is immediately activated. After activation, the scanning strategy generation unit queries prior map data to determine an environmental stability feature point that can most effectively constrain the uncontrollable state, and generates a target scanning command. This command is sent to the long-range scanner in the laser scanning module 20, driving it to accurately point to the target feature and perform a high-density scan. This actively acquired high-value observation data will be given extremely high weight and immediately sent back to the update stage of the adaptive data fusion and state estimation step S4, thus forming a complete closed loop of problem discovery, decision-making, active perception, and problem correction, quickly bringing the state uncertainty of the device back to a normal level.

[0129] Step S6: Integration of data distribution, storage, and application steps; the final fusion result generated from the adaptive data fusion and state estimation steps in Step S4 is a high-frequency, high-precision mobile equipment state vector sequence. The global 3D point cloud, which has undergone precise spatiotemporal registration, is distributed to two destinations via the data transmission module 40. On the one hand, it is sent to the data storage server in the data storage and visualization module 50, where it is classified, archived, and stored long-term according to a predetermined strategy. On the other hand, it is pushed to the 3D visualization platform in real time. The 3D visualization platform uses this real-time data to render the precise trajectory of the mobile equipment in a 3D virtual environment, dynamically construct a 3D model of the work scene, and provide operators with advanced auxiliary functions such as real-time collision warning and automatic measurement of material pile volume.

[0130] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A three-dimensional measurement data fusion device based on Beidou positioning and laser scanning, characterized in that, The three-dimensional measurement data fusion device comprises: a Beidou positioning module, configured to acquire global positioning information of the mobile equipment in a global coordinate system; an inertial measurement unit, configured to acquire three-dimensional angular velocity and three-dimensional acceleration information of the mobile equipment; a laser scanning module, configured to acquire three-dimensional point cloud information of an environment around the mobile equipment, and the laser scanning module comprises a plurality of laser scanners, and the laser scanners comprise a long-range scanner and a multi-layer scanner; a data fusion processing module, configured to receive the global positioning information, the three-dimensional angular velocity and three-dimensional acceleration information, and the three-dimensional point cloud information, estimate a state vector of the mobile equipment and a state covariance matrix representing uncertainty of the state vector based on a kinematic model and by fusing the global positioning information, the three-dimensional angular velocity and three-dimensional acceleration information, and the three-dimensional point cloud information, monitor the state covariance matrix in real time, and generate a control instruction for driving the long-range scanner to perform specific area scanning when the uncertainty of the state vector exceeds a preset uncertainty threshold; a data transmission module, configured to send the control instruction from the data fusion processing module to the laser scanning module to drive the long-range scanner to perform targeted scanning on stable environment features, so as to acquire observation data; a data storage and visualization module, configured to store the state vector and update the state vector by using the observation data, wherein an observation noise covariance matrix is set for the observation data in the updating process to accelerate convergence of the uncertainty. 2.The apparatus for fusing three-dimensional measurement data based on Beidou positioning and laser scanning according to claim 1, characterized in that, The data fusion processing module comprises a scanning strategy generation unit, and the scanning strategy generation unit is configured to generate the control instruction, and specifically comprises: monitoring diagonal elements of the state covariance matrix in real time to acquire variances of state components; when detecting that the variance of any state component exceeds the uncertainty threshold, querying a preset prior environment map to determine stable environment features capable of constraining a maximum dimension of current uncertainty; generating a control instruction for driving the long-range scanner to point to the stable environment features according to global coordinates of the stable environment features. 3.The apparatus for fusing three-dimensional measurement data based on Beidou positioning and laser scanning according to claim 1, characterized in that, The data fusion processing module further comprises an adaptive fusion algorithm unit, which adopts an extended Kalman filtering framework, estimates the state vector of the mobile equipment and the state covariance matrix representing uncertainty of the state vector by fusing the global positioning information, the three-dimensional angular velocity and three-dimensional acceleration information, and the three-dimensional point cloud information, and realizes adaptive weighting of the fusion process by dynamically adjusting an observation noise covariance matrix, to obtain an optimal state vector of the mobile equipment; when the observation data is from the Beidou positioning module, element values of the observation noise covariance matrix are inversely proportional to satellite signal-to-noise ratios contained in the observation data; when the observation data is from the laser scanning module, element values of the observation noise covariance matrix are inversely proportional to fitting degrees of three-dimensional point cloud information registration.

4. The apparatus for fusing three-dimensional measurement data based on Beidou positioning and laser scanning according to claim 1, characterized in that, The data fusion processing module further comprises an online calibration and monitoring unit and a space-time registration unit; the online calibration and monitoring unit is configured to: Real-time recognizing overlapping areas in three-dimensional point cloud information collected by different laser scanners in the laser scanning module; Extracting and matching homonymous feature point pairs in the overlapping areas; Solving a nonlinear least squares optimization problem with the objective of minimizing re-projection error and taking the homonymous feature point pairs as input to real-time solve the extrinsic parameters between the laser scanners; The space-time registration unit is configured to convert the three-dimensional point cloud information from respective sensor coordinate systems to a unified global coordinate system using the extrinsic parameters.

5. The apparatus for fusing three-dimensional measurement data based on Beidou positioning and laser scanning according to claim 4, characterized in that, The online calibration and monitoring unit is further configured to: Taking the root mean square error of the residual after solving the nonlinear least squares optimization problem as a health index of the extrinsic parameters of the laser scanners; In the adaptive fusion algorithm unit, when the observation data comes from the laser scanning module, the element values of the observation noise covariance matrix are inversely proportional to the health index. 6.The apparatus for BDS positioning and laser scanning based three-dimensional measurement data fusion according to claim 1, wherein, The Beidou positioning module includes a positioning data preprocessing unit configured to perform quality control before the data enters the data fusion processing module, and the quality control includes: Filtering according to the type of positioning solution, and only keeping data frames with fixed solution; Performing secondary screening according to the average signal-to-noise ratio of each satellite participating in the solution, and removing data frames with an average signal-to-noise ratio lower than a preset quality threshold.

7. The apparatus for fusing three-dimensional measurement data based on Beidou positioning and laser scanning according to claim 1, characterized in that, The laser scanning module further includes a point cloud data preprocessing industrial computer; The point cloud data preprocessing industrial computer is configured to perform statistical outlier removal filtering and voxel grid downsampling to obtain structured three-dimensional point cloud information. 8.The apparatus for BDS positioning and laser scanning based three-dimensional measurement data fusion according to claim 1, wherein, The long-range scanner is integrated on a two-dimensional gimbal; The multi-layer scanner is fixedly installed on both sides of a cantilever structure of a mobile equipment of the three-dimensional measurement data fusion device. 9.The apparatus for BDS positioning and laser scanning based three-dimensional measurement data fusion according to claim 1, wherein, The data transmission module adopts a redundant communication architecture combining optical fiber communication and industrial wireless communication, and by implementing a parallel redundancy protocol, a copy of a data frame is transmitted through a primary link and a backup link in parallel, and a receiving end only processes the copy of the data frame that first arrives.

10. The method for fusing three-dimensional measurement data based on Beidou positioning and laser scanning, applied to the device for fusing three-dimensional measurement data based on Beidou positioning and laser scanning according to any one of claims 1-9, characterized in that, The method includes the following steps: S1, collecting global positioning information of a mobile equipment through a Beidou positioning module, and collecting three-dimensional angular velocity and three-dimensional acceleration information of the mobile equipment through an inertial measurement unit; S2, collecting three-dimensional point cloud information of an environment around the mobile equipment through a laser scanning module of multiple laser scanners, the laser scanners including a long-range scanner and a multi-layer scanner; S3, estimating a state vector of the mobile equipment and a state covariance matrix representing uncertainty of the state vector based on a kinematic model and by fusing the global positioning information, the three-dimensional angular velocity and three-dimensional acceleration information, and the three-dimensional point cloud information; S4, real-time monitoring the state covariance matrix, and when it is determined based on the state covariance matrix that the uncertainty of the state vector exceeds a preset uncertainty threshold, generating a control instruction for driving the long-range scanner to perform specific area scanning; S5, sending the control instruction to the laser scanning module to drive the long-range scanner to perform targeted scanning on stable environmental features to obtain observation data; S6, updating the state vector with the observation data, wherein in the updating process, an observation noise covariance matrix is set for the observation data to accelerate the convergence of the uncertainty.

Images