A surveying and mapping intelligent analysis system based on multi-source data fusion

Abstract

The application relates to the technical field of surveying and mapping geographic information, and discloses a surveying and mapping intelligent analysis system based on multi-source data fusion, which comprises a multi-source acquisition and space-time reference alignment module, a factor graph consistency fusion and gate module and the like. The multi-source acquisition and space-time reference alignment module is configured to lock a main clock source through a hardware timing protocol, set a hard time alignment threshold, receive multi-source sensor data, judge a time synchronization deviation fed back by the sensor, and output original observation data flow when the time synchronization deviation is lower than the hard time alignment threshold. The factor graph consistency fusion and gate module is configured to receive the original observation data flow and build a global factor graph containing a full-source data constraint chain. The hardware timing protocol locking and hard time alignment threshold mechanism are adopted to establish a unified space-time reference at a physical level from the source, eliminate spatial misplacement caused by sensor sampling differences, realize high-precision consistency fusion of multi-source data, and solve the problems of difficulty in determining deviation sources and low fusion efficiency in traditional technologies.

Description

Technical Field

[0001] This invention relates to the field of surveying and mapping geographic information technology, specifically to a surveying and mapping intelligent analysis system that integrates multi-source data. Background Technology

[0002] With the deepening of the "Real-world 3D China" construction and the "Digital Twin City" strategy, the scenarios for collecting surveying and mapping geographic information data are rapidly expanding from open surface areas to underground spaces, complex indoor environments, and concealed engineering facilities. In scenarios such as urban underground integrated pipe corridors, subway tunnels, deep mines, and large underground transportation hubs, high-precision 3D geographic information data plays a crucial role in facility operation and maintenance, safety monitoring, and asset management.

[0003] To acquire high-precision 3D point cloud and panoramic images in the aforementioned scenarios, the industry widely adopts technologies based on... Mobile mapping systems utilize advanced technology. These systems typically integrate multiple sensors, including LiDAR, panoramic cameras, high-precision inertial measurement units, and odometers. Their working principle is as follows: LiDAR acquires the geometric structure information of the environment, the panoramic camera acquires the texture and color information of the environment, and so on. The system uses odometers to calculate the vehicle's attitude and employs multi-source data fusion algorithms to achieve autonomous positioning and 3D environmental reconstruction even in environments with weak or completely absent global navigation satellite system signals. While existing mobile mapping equipment has made significant progress in data acquisition efficiency due to improvements in sensor hardware performance, significant technical bottlenecks remain in areas such as spatiotemporal consistency processing of multi-source data, quantitative evaluation of results quality, and intelligent closed-loop management of operations.

[0004] When operating in underground utility tunnels or other similar environments, the sampling frequencies and timing methods of different sensors vary, causing spatial misalignment of multi-source data within a unified coordinate framework. This makes it difficult to accurately determine the source of the deviation and the boundary of responsibility, thus affecting the accuracy and efficiency of multi-source data fusion processing.

[0005] Existing mobile mapping systems present the final model or trajectory results, lacking structured quantitative analysis of fusion residuals and data quality factors. Results quality assessment mainly relies on random checks based on human experience, lacking quantifiable confidence indicators and error budgeting systems. This subjective assessment method cannot objectively reflect true accuracy and is highly prone to acceptance disputes.

[0006] When surveying quality defects are identified, existing technologies cannot generate specific resurvey paths and data acquisition parameter configurations based on uncertainty distributions. This makes it difficult for field personnel to obtain clear resurvey task instructions, forcing them to blindly expand the scope of rework for comprehensive resurveys, resulting in a significant increase in the time and manpower costs of field surveying operations.

[0007] To address the aforementioned issues, this invention proposes a multi-source data fusion intelligent surveying and mapping analysis system. Through hardware time synchronization, factor graph gating verification, and uncertainty task inversion, it achieves accurate fusion of multi-source data and structured quality measurement, realizing a shift from blind re-measurement to intelligent targeted operations. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides a multi-source data fusion intelligent mapping analysis system to solve the problems mentioned in the background section.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a multi-source data fusion intelligent mapping analysis system, comprising:

[0010] The multi-source acquisition and spatiotemporal reference alignment module is configured to lock the master clock source through a hardware timing protocol, set a hard time alignment threshold, receive multi-source sensor data, determine the time synchronization deviation fed back by the sensors, and output the raw observation data stream when the time synchronization deviation is lower than the hard time alignment threshold.

[0011] factor Figure 1 The consistency fusion and gating module is configured to receive the original observation data stream, construct a global factor graph containing the constraint chain of the whole source data, perform statistical tests on each closure constraint or observation constraint to be added during the factor graph optimization process, perform an abnormal freezing operation when the test result exceeds the set threshold, and solve and output the optimal estimated trajectory and posterior covariance matrix.

[0012] The structured quality measurement and encapsulation module is configured to receive the optimal estimated trajectory and the posterior covariance matrix, slice the trajectory, calculate the position uncertainty index and residual distribution of each slice, generate a main package containing three-dimensional results and a quality sub-package with embedded error budget fields, and hash-bind the quality sub-package with the main package.

[0013] The uncertainty-driven intelligent supplementary test generation module is configured to parse the error budget field in the quality sub-package, identify defect partitions, simulate the change in accuracy after retesting based on uncertainty pre-calculation logic, and generate an electronic supplementary test task sheet containing specific acquisition parameter instructions and path planning based on the simulation results.

[0014] Preferably, the multi-source acquisition and spatiotemporal reference alignment module is configured with a basis... A standard precision time protocol master clock source is used, with LiDAR, panoramic camera, inertial measurement unit, and odometer acting as nodes that lock onto the precision time protocol master clock source via Ethernet or hardware trigger signals. When the timestamp deviation of multiple consecutive frames of data is less than... milliseconds to When the value is in the range of milliseconds, it is marked as valid data and allowed to proceed to the next level of processing.

[0015] Preferably, the multi-source acquisition and spatiotemporal reference alignment module maintains a coordinate reference constraint table, which stores the absolute coordinates of known control points. The multi-source acquisition and spatiotemporal reference alignment module maps the point cloud coordinates of the lidar, the visual optical center of the panoramic camera, and the center of the inertial measurement unit to the carrier's central coordinate system.

[0016] Preferably, the uncertainty-driven intelligent supplementary measurement generation module reads the error budget field, marks the trajectory slices with position uncertainty index higher than the preset tolerance as defect partitions, and combines adjacent defect partitions into road segments to be supplemented.

[0017] Preferably, the uncertainty-driven intelligent supplementary test generation module executes secondary generation logic when generating an electronic supplementary test task sheet. This logic includes: performing pre-calculation simulation for the defect partition; if the simulation results show that the uncertainty reduction after regular retesting meets the accuracy requirements, generating a regular supplementary test task including start and end mileage; if the simulation results show that the uncertainty reduction after regular retesting does not meet the accuracy requirements, automatically adding a forced control point suggestion instruction and a parameter downgrade instruction to the electronic supplementary test task sheet. The parameter downgrade instruction includes lowering the maximum acquisition speed limit to a preset low-speed threshold or locking the sensor to a high frame rate mode.

[0018] Preferably, the electronic supplementary measurement task sheet includes the supplementary measurement path planning trajectory, the suggested walking speed range, the suggested sensor frame rate configuration, and the control point layout location suggestions. The electronic supplementary measurement task sheet is configured to be readable and visualized by the field data acquisition terminal.

[0019] Preferably, the multi-source acquisition and spatiotemporal reference alignment module further includes:

[0020] The master-slave clock synchronization deviation calculation unit is configured to initiate message interaction between the master clock and the slave node via a precise time protocol, and record the time when the master clock sends the synchronization message. Timing of receiving synchronization messages from nodes When to send a delayed request from a node and the master clock receive delay request time Calculate the timestamp offset value of the current frame. The timestamp deviation value The calculation formula is:

[0021] ,

[0022] in, The absolute time at which the synchronization message leaves the master clock source. To synchronize the arrival time of the message at the slave node's local time, To delay the time it takes for the request message to leave the slave node's local time, To delay the absolute time of the request message arriving at the master clock source;

[0023] The master-slave clock synchronization deviation calculation unit will calculate the... With the hard time alignment threshold Perform a comparison; if the judgment criteria are met... ≤ Determine if the current clock synchronization state is valid and output a synchronized timestamp after synchronization correction. ;

[0024] A multi-sensor spatial extrinsic rigid transformation unit is configured to receive the unified timestamp. And the corresponding raw sensor observation data, calling the pre-stored rotation matrix With translation vector The observation points in the lidar coordinate system or the visual camera coordinate system Mapping to the carrier's central coordinate system yields the carrier's coordinate points. The carrier coordinate points The calculation formula is:

[0025] ,

[0026] in, This is the three-dimensional coordinate vector in the transformed carrier center coordinate system. From the sensor coordinate system to the carrier center coordinate system Rotation matrix, This is the original three-dimensional coordinate vector in the sensor coordinate system. From the sensor center to the carrier center Translation vector;

[0027] The multi-sensor spatial extrinsic rigid transformation unit further incorporates the unified timestamp. Calculate motion distortion compensation amount Output corrected observation data after spatial alignment and distortion compensation. ;

[0028] The spatiotemporal consistency validity gated output unit is configured to receive the corrected observation data. Calculate the data validity score of the current data frame. The data validity score The calculation formula is:

[0029] ,

[0030] in, To score the validity of the data, For time synchronization weighting coefficients, This is the weighting coefficient for spatial feature richness. The absolute value of the timestamp deviation. This is a hard time alignment threshold. This represents the number of feature points extracted in the current frame. The minimum number of features threshold;

[0031] The spatiotemporal consistency validity gating output unit sets validity through a threshold. When the judgment condition is met ≥ At that time, the corrected observation data will be... It is encapsulated as a raw observation data stream and transmitted to the next level module.

[0032] Preferably, the factor Figure 1 The coherent fusion and gating module further includes:

[0033] The total-source state variable and factor topology construction unit is configured to receive the original observation data stream and transmit the data at discrete times. state vector Defined as including three-dimensional position 3D velocity Posture Quaternions and sensor zero bias The combined vectors are used to construct pre-integration constraint edges from the inertial measurement data in the original observation data stream, relative pose constraint edges from the odometry data, and observation matching constraint edges from the visual features and laser point cloud data, forming the initial global optimization objective function. The initial global optimization objective function The expression is:

[0034] ,

[0035] in, It is a global set of state variables that contains the states at all times. The residuals are generated from prior information. For the observations of each sensor The constraint residuals between the state variables, For the prior information matrix, To observe the noise covariance matrix, It is the set of all constraints.

[0036] The output of the all-source state variable and factor topology construction unit contains a candidate factor graph topology structure containing closed loops to be verified;

[0037] The closed-loop constraint statistical gating unit is configured to receive the candidate factor graph topology, extract the closed-loop constraints to be added, and calculate the test statistic of the closed-loop constraint. The test statistic The calculation formula is:

[0038] ,

[0039] in, The residual vector generated by the closed loop. Let Jacobian matrix be the observation function relative to the state variables. The covariance matrix estimated from the current state. For closed-loop observation measurement noise matrix, To perform the matrix inversion operation, This is a matrix transpose operation;

[0040] The closed-loop constraint statistical gating unit sets a chi-square distribution threshold. If the judgment conditions are met ≤ The closed loop constraint is marked as a valid constraint and added to the set of valid factors. If the decision condition is met... The abnormal freeze operation is executed and the closed loop constraint is removed, and the set of effective factors is finally output.

[0041] The nonlinear incremental smoothing solution unit is configured to receive the set of effective factors, construct the incremental normal equation, and iteratively calculate the global state increment using the Levenberg-Marquardt method. The global state increment The calculation equation is as follows:

[0042] ,

[0043] in, Let be the Jacobian matrix of the global residuals relative to the global state variables. The global noise covariance matrix is... The damping coefficient is... It is the identity matrix. This is the global residual vector for the current iteration step;

[0044] The nonlinear incremental smoothing solution unit utilizes the calculated global state increment. Update the global state vector Calculate the Hessian matrix The inverse matrix is ​​used as the posterior covariance matrix, and finally the optimal estimated trajectory and the posterior covariance matrix are output to the next level module.

[0045] Preferably, the The structured quality metric and encapsulation module further includes:

[0046] The trajectory slicing and uncertainty quantification unit is configured to receive the optimal estimated trajectory and the posterior covariance matrix, and then perform quantization according to a preset mileage step size. Discretize the continuous trajectory into several trajectory slices. Extract each trajectory slice The corresponding local posterior covariance submatrix Calculate the scalar of the three-dimensional positional uncertainty of this trajectory slice. The three-dimensional position uncertainty scalar The calculation formula is:

[0047] ,

[0048] in, For the trace operation of a matrix, For the corresponding three-dimensional position state Covariance submatrix , , For along axis, axis, The variance component along the axis; the trajectory slicing and uncertainty quantization unit further sets the accuracy tolerance threshold. If the judgment conditions are met > Slice the current trajectory The defect candidate regions that require special attention are marked, and the calculated three-dimensional positional uncertainty scalar is used. This data is transmitted as the basic data to the next level unit;

[0049] The quality element mapping and confidence rating unit is configured to receive the trajectory slice. and the corresponding scalar of three-dimensional position uncertainty Based on the residual statistics in the posterior covariance matrix, according to Geographic Information Quality Standards Calculation of Comprehensive Quality Score The comprehensive quality score The calculation formula is:

[0050] ,

[0051] in, , , The weights are location accuracy weight, logical consistency weight, and integrity weight. The normalization factor for the maximum uncertainty. The average residual modulus within the current slice. Let V be the variance of the residual distribution. This represents the number of constraints that pass the chi-square test. The total number of constraints; The quality element mapping and confidence rating unit is based on the comprehensive quality score. Generate structured metadata containing location accuracy level, logical consistency records, and integrity ratio, and include the three-dimensional location uncertainty scalar. Together with the structured metadata, it is packaged into an error budget field;

[0052] The result structured encapsulation and hash anchoring unit is configured to receive the error budget field and the 3D result, generate a quality sub-package in binary format and a main package in standard format, and calculate the data integrity hash value of the main package. The data integrity hash value The calculation algorithm is as follows:

[0053] ,

[0054] in, For secure hashing algorithms, The main packet's binary data stream, For XOR operation, The timestamp value for the generated time; the structured encapsulation and hash anchoring unit will calculate the data integrity hash value. Write the fixed byte area of ​​the file header of the quality sub-package, establish a unique verification index of the quality sub-package for the main package, and output the multi-source data fusion mapping results bound by hash.

[0055] Preferably, the uncertainty-driven intelligent supplementary test generation module further includes:

[0056] The defect partitioning identification and topology combination unit is configured to parse the error budget field and extract the positional uncertainty index of each trajectory slice. This indicator is compared with a preset tolerance threshold. Perform comparison and generate defect indication function. The defect indication function The definition of is:

[0057] ,

[0058] in, This is the slice index number;

[0059] The defect partitioning and topology combination unit further performs spatial connectivity analysis to calculate the Euclidean distance between adjacent defect slices. If the judgment conditions are met ≤ Discrete defect slices are combined into a continuous set of road segments to be retested. ,in, The maximum allowable merging spacing; the defect partition identification and topology combination unit outputs the set of road segments to be supplemented. and the corresponding current information matrix To the next level unit;

[0060] The accuracy gain simulation and decision unit for supplementary measurement is configured to receive the set of road segments to be supplemented. and the current information matrix A virtual observation model is constructed based on the information filtering principle to simulate the posterior information matrix under the conventional supplementary measurement mode. The posterior information matrix The calculation formula is:

[0061] ,

[0062] in, The number of simulated virtual observations, For virtual observation Jacobian matrix, The noise covariance matrix of the conventional acquisition equipment is preset; the supplementary measurement accuracy gain simulation and decision unit is based on... Calculate the expected decrease in uncertainty If the judgment conditions are met ≥ If the judgment conditions are met, the normal supplementary test mode will be executed. The system determines whether to implement downgraded or enhanced control modes based on the execution parameters. To improve accuracy, a threshold is required; the supplementary measurement accuracy gain simulation and decision-making unit outputs the final supplementary measurement mode decision command. ;

[0063] The multi-parameter task inversion and encoding unit is configured to receive the supplementary test mode decision instruction. and the set of road sections to be supplemented for testing When the supplementary testing mode decision command determines that the mode is parameter degradation and enhanced control mode, the inversion calculation suggests an upper limit for the acquisition speed. The suggested upper limit for acquisition speed The calculation formula is:

[0064] ,

[0065] in, This refers to the scanning frequency of the lidar. This is the overlap rate coefficient between adjacent frames. To meet the target point cloud line density requirements; the multi-parameter task inversion and encoding unit simultaneously identifies the road segments to be supplemented in the set. Take the coordinates of the maximum value , the coordinates The locations marked as mandatory control point deployment points will ultimately be used to calculate the upper limit of the recommended acquisition speed. The locations of the mandatory control points and the start and end mileage codes of the set of road sections to be retested are used to generate an electronic retesting task sheet.

[0066] The multi-source acquisition and spatiotemporal reference alignment module forcibly locks the master and slave clock sources through the underlying hardware protocol and performs rigorous time synchronization verification. It eliminates time asynchrony errors caused by the difference in sampling frequency between different sensors at the physical level, ensuring that the LiDAR, panoramic image and inertial navigation data are under the same high-precision time axis and spatial coordinate reference before entering the algorithm layer. This effectively avoids subsequent fusion calculation errors or trajectory drift caused by front-end input deviations.

[0067] factor Figure 1 The consistency fusion and gating module utilizes global optimization theory to construct a mathematical model containing a full-source constraint chain and introduces a statistical verification mechanism. It can perform real-time and rigorous correctness verification of each closed loop or observation constraint to be added, automatically identify and eliminate false closures or abnormal data with severe drift, and ensure that the system can output the optimal estimated trajectory and posterior covariance matrix with extremely high robustness and geometric consistency in complex closed environments where satellite signals are denied, thus solving the problem of easy divergence in localization in weak texture environments.

[0068] The structured quality measurement and encapsulation module transforms abstract algorithm residuals into visualized confidence ratings and error budget records based on international geographic information quality standards. By performing refined slicing and hash encryption binding on the trajectory, it realizes the transformation of surveying and mapping results from black-box solution to white-box quantitative delivery. This provides objective and tamper-proof digital quality certificates for subsequent data acceptance, and solves the quality risks and delivery disputes caused by relying on human subjective experience to judge the qualification of results in traditional operations.

[0069] The uncertainty-driven intelligent supplementary survey generation module uses inverse inversion logic to analyze the distribution of quality defects and simulate the accuracy gain effect under different acquisition strategies. It automatically generates electronic task instructions that include precise path planning, acquisition speed limits, and control point layout suggestions. This guides field personnel to carry out targeted operations, avoids the waste of resources caused by blindly conducting large-scale re-surveys, and ensures that supplementary surveys can meet accuracy requirements in one go. This significantly reduces the rework rate and manpower and time costs of underground complex space surveying operations.

[0070] This invention provides a surveying and mapping intelligent analysis system based on multi-source data fusion. It has the following beneficial effects:

[0071] 1. This invention adopts a hardware timing protocol locking and hard time alignment threshold mechanism to establish a unified spatiotemporal reference at the physical level from the source, eliminate spatial misalignment caused by differences in sensor sampling, achieve high-precision consistency fusion of multi-source data, and solve the shortcomings of traditional technology in that it is not easy to determine the source of deviation and the fusion efficiency is low.

[0072] 2. This invention employs statistical gating fusion and Structured quality encapsulation technology transforms abstract algorithm residuals into quantifiable, tamper-proof confidence indicators and error budget fields, enabling strong hash binding between surveying results and quality data. This addresses the shortcomings of existing technologies that rely on subjective manual sampling and lack an objective evaluation system.

[0073] 3. This invention adopts uncertainty-driven precision simulation and task inversion logic, and automatically generates electronic task sheets containing specific parameters and control point suggestions based on defect partitioning, realizing the transformation from blind coverage and retesting to intelligent targeted operation, solving the shortcomings of field work, such as large rework scope and high cost due to lack of clear instructions. Attached Figure Description

[0074] Figure 1 This is a block diagram illustrating the overall architecture principle of a multi-source data fusion intelligent analysis system for surveying and mapping according to the present invention.

[0075] Figure 2 This is a schematic diagram of the decision logic of the uncertainty-driven intelligent supplementary test generation module in this invention. Detailed Implementation

[0076] To enable those skilled in the art to understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0077] The present invention will now be described in detail with reference to the accompanying drawings:

[0078] Example 1: Routine Inspection of Urban Underground Utility Tunnels

[0079] Please see the appendix Figure 1 and attached Figure 2 This embodiment verifies the system's spatiotemporal synchronization capability, anomaly removal capability, and routine supplementary measurement task generation process in a conventional underground enclosed environment.

[0080] Multi-source acquisition and spatiotemporal reference alignment stage

[0081] Workers carry integrated A mobile mapping backpack equipped with a line lidar, panoramic camera, high-precision fiber optic inertial navigation system, and odometer is deployed into the underground utility tunnel. After activation, based on... The standard precision time protocol master clock source starts working, distributing synchronization messages to each sensor via Ethernet; real-time monitoring of the hardware-level synchronization status; during the acquisition process, the time synchronization deviation between the inertial navigation system and the lidar is controlled within a specified range. Around milliseconds, far below the set value Millisecond hard alignment threshold. At this point, the clock synchronization status is determined to be valid, and the coordinates of each sensor are automatically mapped to the center of the carrier. The original observation data stream with a uniform high-precision timestamp is output to ensure that the initial spatial position of the point cloud and the image is strictly aligned in the narrow space of the tube gallery.

[0082] factor Figure 1 Sexual fusion and gating phase

[0083] The raw data stream enters the background processing system. A global factor graph is constructed, using the pre-integration results of the inertial navigation system as constraint edges and the feature point cloud of the lidar as observation edges. When a worker walks to the intersection of the utility tunnel and turns back, a loop closure opportunity is identified. At this point, the closure constraint statistical gating unit intervenes and performs a chi-square test on the loop constraint. The calculation shows that the test statistic of the loop closure is lower than the set chi-square distribution threshold, indicating that the loop closure is accurately and reliably matched, and it is immediately added to the effective factor set for optimization. A smooth, ghost-free optimal estimated trajectory is calculated using the Levenburg-Marquardt method, and the posterior covariance matrix of the trajectory is output simultaneously.

[0084] Structured quality metrics and packaging stage

[0085] Received long-term contract Kilometer trajectory data, according to Slicing is performed in increments of meters; in the first... Rice to Within a meter interval, due to the uniform characteristics of the inner wall of the utility tunnel, the calculated scalar uncertainty of the three-dimensional position of this slice is slightly higher than the standard value; based on Geographic information quality standards, combining location accuracy, residual distribution, and integrity ratio, generate data including " The quality data is then packaged into a "quality sub-package" using structured metadata identified by "level precision". This quality data is then used to generate a timestamp salt value. The algorithm calculates a hash value and binds it to the main packet storing the 3D point cloud results to generate a tamper-proof final data packet.

[0086] Uncertainty-driven intelligent supplementary test generation stage

[0087] Parse the quality sub-packet and identify the first Rice to The system delineates the defect zone by meter; its internal pre-calculation logic simulates the accuracy change after a routine retest in this area. The simulation results show that further data acquisition is necessary, and by increasing observation redundancy, the uncertainty can be reduced to an acceptable range. Therefore, the system generates a standard electronic retest task sheet. The task sheet highlights the start and end mileage on the map and recommends that field personnel collect data at a normal walking pace, eliminating the need for additional control points. Field personnel can then quickly complete targeted retests, avoiding a complete rework.

[0088] Example 2: As-built Survey of a New Subway Shield Tunnel

[0089] Please see the appendix Figure 1 and attached Figure 2 This embodiment verifies the system's anti-interference capability, severe quality defect identification, and the execution flow of parameter degradation / enhancement control supplementary testing mode in extreme low-texture environments.

[0090] Multi-source acquisition and spatiotemporal reference alignment stage

[0091] The surveying system, mounted on a track inspection trolley, entered the newly constructed shield tunnel. Due to the uneven track, the equipment experienced high-frequency and severe vibrations. The spatiotemporal reference alignment module detected in real-time that several frames of panoramic camera data experienced hardware trigger signal delays due to vibration, resulting in a timestamp deviation of 6 milliseconds, exceeding the set threshold of 5 milliseconds. The system automatically marked these frames as invalid data and blocked them from entering the backend fusion process.

[0092] factor Figure 1 Sexual fusion and gating phase

[0093] During processing, due to the highly repetitive textures of the tunnel lining segments, visual feature matching is prone to errors. When processing a section of curved data, the front-end algorithm incorrectly matched the 100th ring segment to the 105th ring, generating an incorrect closed-loop constraint. Before factor graph optimization, the closure constraint statistical gating unit calculated the test statistic of this constraint and found that the value far exceeded the chi-square distribution threshold. The system immediately triggered an abnormal freeze operation, decisively removing the erroneous closed loop.

[0094] Structured quality metrics and packaging stage

[0095] After processing, the system performs a sliced ​​evaluation of the trajectory. In the curved areas where closed loops were eliminated, the system calculations revealed a sharp increase in the three-dimensional position uncertainty index, far exceeding the acceptance criteria. When generating the quality sub-package, the system marked this area as a serious defect and explicitly recorded the low logical consistency and high residual distribution in the error budget field.

[0096] Uncertainty-driven intelligent supplementary test generation stage

[0097] The intelligent retesting module detected this serious defect. A simulated retest revealed that even after repeating the process, the texture remained repetitive, resulting in limited accuracy improvement and failing to meet requirements. Therefore, the system triggered parameter degradation and secondary generation logic, calculating that reducing the acquisition speed to 0.5 m / s, locking the LiDAR to high frame rate mode to increase point cloud density, and simultaneously deploying forced control points at the midpoint of curves would achieve the required accuracy. Finally, the system generated a complex electronic retesting task sheet. The task sheet clearly stated the instructions:

[0098] Supplementary measurements will be conducted between mileage K12+200 and K12+400; driving speed is required to be no higher than 0.5 m / s; it is recommended to set up spherical target control points at K12+300.

[0099] After receiving the instructions, the field staff strictly followed the parameters and successfully completed the retest on the first attempt, thus solving the mapping problem in a low-texture environment.

[0100] Embodiments of the present invention have been presented and described. It will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to the 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 surveying and mapping intelligent analysis system that integrates multi-source data, characterized in that, include: The multi-source acquisition and spatiotemporal reference alignment module is configured to lock the master clock source through a hardware timing protocol, set a hard time alignment threshold, receive multi-source sensor data, determine the time synchronization deviation fed back by the sensors, and output the raw observation data stream when the time synchronization deviation is lower than the hard time alignment threshold. The factor graph consistency fusion and gating module is configured to receive the original observation data stream, construct a global factor graph containing a constraint chain of all source data, perform statistical tests on each closure constraint or observation constraint to be added during the factor graph optimization process, perform an abnormal freezing operation when the test results exceed a set threshold, and solve and output the optimal estimated trajectory and posterior covariance matrix. The structured quality measurement and encapsulation module is configured to receive the optimal estimated trajectory and the posterior covariance matrix, slice the trajectory, calculate the position uncertainty index and residual distribution of each slice, generate a main package containing three-dimensional results and a quality sub-package with embedded error budget fields, and hash-bind the quality sub-package with the main package. The uncertainty-driven intelligent supplementary test generation module is configured to parse the error budget field in the quality sub-package, identify defect partitions, simulate the change in accuracy after retesting based on uncertainty pre-calculation logic, and generate an electronic supplementary test task sheet containing specific acquisition parameter instructions and path planning based on the simulation results.

2. The intelligent mapping analysis system based on multi-source data fusion according to claim 1, characterized in that, The multi-source acquisition and spatiotemporal reference alignment module is configured with a basis... A standard precision time protocol master clock source is used, with LiDAR, panoramic camera, inertial measurement unit, and odometer acting as nodes that lock onto the precision time protocol master clock source via Ethernet or hardware trigger signals. When the timestamp deviation of multiple consecutive frames of data is less than... milliseconds to When the value is in the range of milliseconds, it is marked as valid data and allowed to proceed to the next level of processing.

3. The intelligent mapping analysis system based on multi-source data fusion according to claim 1, characterized in that, The multi-source acquisition and spatiotemporal reference alignment module maintains a coordinate reference constraint table, which stores the absolute coordinates of known control points. The multi-source acquisition and spatiotemporal reference alignment module maps the point cloud coordinates of the lidar, the visual center of the panoramic camera, and the center of the inertial measurement unit to the carrier's central coordinate system.

4. The intelligent mapping analysis system based on multi-source data fusion according to claim 1, characterized in that, The uncertainty-driven intelligent supplementary measurement generation module reads the error budget field, marks the trajectory slices with location uncertainty index higher than the preset tolerance as defect partitions, and combines adjacent defect partitions into road segments to be supplemented.

5. The intelligent mapping analysis system based on multi-source data fusion according to claim 1, characterized in that, The uncertainty-driven intelligent supplementary test generation module executes secondary generation logic when generating an electronic supplementary test task sheet. This logic includes: performing pre-calculation simulation for the defect partition; if the simulation results show that the uncertainty reduction after regular retesting meets the accuracy requirements, generating a regular supplementary test task including start and end mileage; if the simulation results show that the uncertainty reduction after regular retesting does not meet the accuracy requirements, automatically adding a forced control point suggestion instruction and a parameter downgrade instruction to the electronic supplementary test task sheet. The parameter downgrade instruction includes lowering the maximum acquisition speed limit to a preset low-speed threshold or locking the sensor to a high frame rate mode.

6. The intelligent mapping analysis system based on multi-source data fusion according to claim 1, characterized in that, The electronic supplementary measurement task sheet includes the supplementary measurement path planning trajectory, the suggested walking speed range, the suggested sensor frame rate configuration, and the control point layout location suggestions. The electronic supplementary measurement task sheet is configured to be readable and visualized by the field data acquisition terminal.

7. The intelligent mapping analysis system based on multi-source data fusion according to claim 1, characterized in that, The multi-source acquisition and spatiotemporal reference alignment module further includes: The master-slave clock synchronization deviation calculation unit is configured to initiate message interaction between the master clock and the slave node via a precise time protocol, and record the time when the master clock sends the synchronization message. Timing of receiving synchronization messages from nodes When to send a delayed request from a node and the master clock receive delay request time Calculate the timestamp offset value of the current frame. The timestamp deviation value The calculation formula is: ， in, The absolute time at which the synchronization message leaves the master clock source. To synchronize the arrival time of the message at the slave node's local time, To delay the time it takes for the request message to leave the slave node's local time, To delay the absolute time of the request message arriving at the master clock source; The master-slave clock synchronization deviation calculation unit will calculate the... With the hard time alignment threshold Perform a comparison; if the judgment criteria are met... ≤ Determine if the current clock synchronization state is valid and output a synchronized timestamp after synchronization correction. ; A multi-sensor spatial extrinsic rigid transformation unit is configured to receive the unified timestamp. And the corresponding raw sensor observation data, calling the pre-stored rotation matrix With translation vector The observation points in the lidar coordinate system or the visual camera coordinate system Mapping to the carrier's central coordinate system yields the carrier's coordinate points. The carrier coordinate points The calculation formula is: ， in, This is the three-dimensional coordinate vector in the transformed carrier center coordinate system. From the sensor coordinate system to the carrier center coordinate system Rotation matrix, This is the original three-dimensional coordinate vector in the sensor coordinate system. From the sensor center to the carrier center Translation vector; The multi-sensor spatial extrinsic rigid transformation unit further incorporates the unified timestamp. Calculate motion distortion compensation amount Output corrected observation data after spatial alignment and distortion compensation. ; The spatiotemporal consistency validity gated output unit is configured to receive the corrected observation data. Calculate the data validity score of the current data frame. The data validity score The calculation formula is: ， in, To score the validity of the data, For time synchronization weighting coefficients, This is the weighting coefficient for spatial feature richness. The absolute value of the timestamp deviation. This is a hard time alignment threshold. This represents the number of feature points extracted in the current frame. The minimum number of features threshold; The spatiotemporal consistency validity gating output unit sets validity through a threshold. When the judgment condition is met ≥ At that time, the corrected observation data will be... It is encapsulated as a raw observation data stream and transmitted to the next level module.

8. The intelligent mapping analysis system based on multi-source data fusion according to claim 7, characterized in that, The factor graph consistency fusion and gating module further includes: The total-source state variable and factor topology construction unit is configured to receive the original observation data stream and transmit the data at discrete times. state vector Defined as including three-dimensional position 3D velocity Posture Quaternions and sensor zero bias The combined vectors are used to construct pre-integration constraint edges from the inertial measurement data in the original observation data stream, relative pose constraint edges from the odometry data, and observation matching constraint edges from the visual features and laser point cloud data, forming the initial global optimization objective function. The initial global optimization objective function The expression is: ， in, It is a global set of state variables that contains the states at all times. The residuals are generated from prior information. For the observations of each sensor The constraint residuals between the state variables, For the prior information matrix, To observe the noise covariance matrix, It is the set of all constraints. The output of the all-source state variable and factor topology construction unit contains a candidate factor graph topology structure containing closed loops to be verified; The closed-loop constraint statistical gating unit is configured to receive the candidate factor graph topology, extract the closed-loop constraints to be added, and calculate the test statistic of the closed-loop constraint. The test statistic The calculation formula is: ， in, The residual vector generated by the closed loop. Let Jacobian matrix be the observation function relative to the state variables. The covariance matrix estimated from the current state. For closed-loop observation measurement noise matrix, To perform the matrix inversion operation, This is a matrix transpose operation; The closed-loop constraint statistical gating unit sets a chi-square distribution threshold. If the judgment conditions are met ≤ The closed loop constraint is marked as a valid constraint and added to the set of valid factors. If the decision condition is met... The abnormal freeze operation is executed and the closed loop constraint is removed, and the set of effective factors is finally output. The nonlinear incremental smoothing solution unit is configured to receive the set of effective factors, construct the incremental normal equation, and iteratively calculate the global state increment using the Levenberg-Marquardt method. The global state increment The calculation equation is as follows: ， in, Let be the Jacobian matrix of the global residuals relative to the global state variables. The global noise covariance matrix is... The damping coefficient is... It is the identity matrix. This is the global residual vector for the current iteration step; The nonlinear incremental smoothing solution unit utilizes the calculated global state increment. Update the global state variables and calculate the Hessian matrix. The inverse matrix is ​​used as the posterior covariance matrix, and finally the optimal estimated trajectory and the posterior covariance matrix are output to the next level module.

9. A surveying and mapping intelligent analysis system based on multi-source data fusion according to claim 8, characterized in that, The The structured quality metric and encapsulation module further includes: The trajectory slicing and uncertainty quantification unit is configured to receive the optimal estimated trajectory and the posterior covariance matrix, and then perform quantization according to a preset mileage step size. Discretize the continuous trajectory into several trajectory slices. Extract each trajectory slice The corresponding local posterior covariance submatrix Calculate the scalar of the three-dimensional positional uncertainty of this trajectory slice. The three-dimensional position uncertainty scalar The calculation formula is: ， in, For the trace operation of a matrix, For the corresponding three-dimensional position state Covariance submatrix , , For along axis, axis, The variance component along the axis; the trajectory slicing and uncertainty quantization unit further sets the accuracy tolerance threshold. If the judgment conditions are met > Slice the current trajectory The defect candidate regions that require special attention are marked, and the calculated three-dimensional positional uncertainty scalar is used. This data is transmitted as the basic data to the next level unit; The quality element mapping and confidence rating unit is configured to receive the trajectory slice. and the corresponding scalar of three-dimensional position uncertainty Based on the residual statistics in the posterior covariance matrix, according to Geographic Information Quality Standards Calculation of Comprehensive Quality Score The comprehensive quality score The calculation formula is: ， in, , , The weights are location accuracy weight, logical consistency weight, and integrity weight. The normalization factor for the maximum uncertainty. The average residual modulus within the current slice. Let V be the variance of the residual distribution. This represents the number of constraints that pass the chi-square test. The total number of constraints; The quality element mapping and confidence rating unit is based on the comprehensive quality score. Generate structured metadata containing location accuracy level, logical consistency records, and integrity ratio, and include the three-dimensional location uncertainty scalar. Together with the structured metadata, it is packaged into an error budget field; The result structured encapsulation and hash anchoring unit is configured to receive the error budget field and the 3D result, generate a quality sub-package in binary format and a main package in standard format, and calculate the data integrity hash value of the main package. The data integrity hash value The calculation algorithm is as follows: ， in, For secure hashing algorithms, The main packet's binary data stream, For XOR operation, The timestamp value for the generated time; the structured encapsulation and hash anchoring unit will calculate the data integrity hash value. Write the fixed byte area of ​​the file header of the quality sub-package, establish a unique verification index of the quality sub-package for the main package, and output the multi-source data fusion mapping results bound by hash.

10. A surveying and mapping intelligent analysis system based on multi-source data fusion according to claim 9, characterized in that, The uncertainty-driven intelligent supplementary test generation module further includes: The defect partitioning identification and topology combination unit is configured to parse the error budget field and extract the positional uncertainty index of each trajectory slice. This indicator is compared with a preset tolerance threshold. Perform comparison and generate defect indication function. The defect indication function The definition of is: ， in, This is the slice index number; The defect partitioning and topology combination unit further performs spatial connectivity analysis to calculate the Euclidean distance between adjacent defect slices. If the judgment conditions are met ≤ Discrete defect slices are combined into a continuous set of road segments to be retested. ,in, The maximum allowable merging spacing; the defect partition identification and topology combination unit outputs the set of road segments to be supplemented. and the corresponding current information matrix To the next level unit; The accuracy gain simulation and decision unit for supplementary measurement is configured to receive the set of road segments to be supplemented. and the current information matrix A virtual observation model is constructed based on the information filtering principle to simulate the posterior information matrix under the conventional supplementary measurement mode. The posterior information matrix The calculation formula is: ， in, The number of simulated virtual observations, For virtual observation Jacobian matrix, The noise covariance matrix of the conventional acquisition equipment is preset; the supplementary measurement accuracy gain simulation and decision unit is based on... Calculate the expected decrease in uncertainty If the judgment conditions are met ≥ If the judgment conditions are met, the normal supplementary test mode will be executed. The system determines whether to implement downgraded or enhanced control modes based on the execution parameters. To improve accuracy, a threshold is required; the supplementary measurement accuracy gain simulation and decision-making unit outputs the final supplementary measurement mode decision command. ; The multi-parameter task inversion and encoding unit is configured to receive the supplementary test mode decision instruction. and the set of road sections to be supplemented for testing When the supplementary testing mode decision command determines that the mode is parameter degradation and enhanced control mode, the inversion calculation suggests an upper limit for the acquisition speed. The suggested upper limit for acquisition speed The calculation formula is: ， in, This refers to the scanning frequency of the lidar. This is the overlap rate coefficient between adjacent frames. To meet the target point cloud line density requirements; the multi-parameter task inversion and encoding unit simultaneously identifies the road segments to be supplemented in the set. Take the coordinates of the maximum value , the coordinates The locations marked as mandatory control point deployment points will ultimately be used to calculate the upper limit of the recommended acquisition speed. The locations of the mandatory control points and the start and end mileage codes of the set of road sections to be retested are used to generate an electronic retesting task sheet.

Images