An engineering survey dynamic construction scene real-time positioning calibration method and system

Abstract

The application discloses an engineering survey dynamic construction scene real-time positioning calibration method and system, and relates to the technical field of positioning calibration.The steps of the method comprise the following steps: deploying a composite base station and measuring the base station coordinates, and establishing a visual feature point coordinate library; performing sensor fusion terminal installation and space-time calibration to obtain a calibrated sensor fusion terminal; reading vehicle-mounted inertial measurement unit measurement data in the calibrated sensor fusion terminal, and calculating carrier predicted positioning; collecting terminal observation values based on the calibrated sensor fusion terminal, and correcting the carrier predicted positioning by combining the terminal observation values and the visual feature point coordinate library to obtain carrier corrected positioning and positioning result confidence; and establishing a carrier calibration mechanism based on the positioning result confidence, and correcting the carrier corrected positioning through the mechanism to obtain carrier calibrated positioning.Based on the carrier calibrated positioning, engineering survey dynamic construction scene real-time positioning calibration is realized.

Description

Technical Field

[0001] This invention relates to the field of positioning calibration technology, specifically to a real-time positioning calibration method and system for dynamic construction scenarios in engineering surveying. Background Technology

[0002] As large-scale engineering construction projects continue to increase in scale and complexity, construction processes are increasingly evolving towards dynamism and automation. In such scenarios, such as bridge erection, tunnel excavation, or factory component installation, the real-time, high-precision position and attitude information of mobile carriers such as construction machinery and surveying robots has become crucial for ensuring construction quality, efficiency, and safety. Continuous, reliable, and accurate positioning of these carriers is the foundation for achieving automated construction, digital management, and intelligent decision-making.

[0003] Traditional real-time positioning calibration methods for dynamic construction scenarios in engineering surveying typically employ a loosely coupled approach to fuse positioning data from multiple sensors. This method first allows each sensor to operate independently, calculating its own positioning or feature information, such as GNSS position, inertial navigation system track, and odometry velocity. Then, a central filter, such as a particle filter, is used to fuse this positioning or feature information to obtain a final joint positioning result.

[0004] However, the loose coupling in this traditional method has inherent limitations. Since the fusion process is based on the positioning or feature information that each sensor has independently calculated, this method cannot directly optimize the original observations of all sensors under a unified mathematical model, which means that its estimation is not globally optimal and therefore cannot guarantee the robustness and accuracy of the positioning results in complex construction environments. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a real-time positioning and calibration method and system for dynamic construction scenarios in engineering surveying, thus solving the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a real-time positioning and calibration method and system for dynamic construction scenarios in engineering surveying, comprising the following steps: Step S1: Deploy the composite base station and measure its coordinates, and establish a visual feature point coordinate library; install and calibrate the sensor fusion terminal to obtain the calibrated sensor fusion terminal; Step S2: Read the measurement data of the vehicle-mounted inertial measurement unit in the calibrated sensor fusion terminal, calculate the measurement data, and obtain the carrier predictive positioning; Step S3: Based on the calibrated sensor fusion terminal, collect terminal observation values, and combine the terminal observation values ​​with the visual feature point coordinate library to correct the carrier's predicted positioning, thereby obtaining the carrier's corrected positioning and the confidence level of the positioning result; Step S4: Based on the confidence level of the positioning result, establish a carrier calibration mechanism, and calibrate the carrier correction positioning through the carrier calibration mechanism to obtain the carrier calibration positioning.

[0007] Preferably, deploying the composite base station and measuring its coordinates includes: The composite base station consists of an ultra-wideband (UWB) base station module, a power supply unit, and a wireless network communication module. It is a fixed beacon that can transmit wireless signals for terminal ranging. The specific deployment work needs to be planned according to the scope of the construction area, the terrain, and the main work routes. After the composite base station is installed, external high-precision measuring equipment is used to measure the absolute coordinates of each base station. Through resection mapping, the precise three-dimensional coordinates of each base station in a unified construction coordinate system are directly determined.

[0008] Preferably, establishing a visual feature point coordinate library includes: The process involves collecting and establishing a visual feature point coordinate library. Pre-deployed, precisely located, and stable reflective points are used as feature points. A tilting camera is used to collect data from the construction area. Multi-view photogrammetry and feature extraction techniques are employed to obtain the 3D coordinates of each feature point and its corresponding high-resolution multi-view image. Scale- and rotation-invariant visual feature descriptors are extracted from these multi-view images to encode the appearance, texture, and local geometric features of the feature points. The 3D coordinates of each feature point are associated with its visual feature descriptor and identifier, and stored in a structured manner to establish the visual feature point coordinate library.

[0009] Preferably, the sensor fusion terminal is installed and spatiotemporally calibrated to obtain a calibrated sensor fusion terminal, including: The sensor fusion terminal is a hardware suite that integrates multiple high-precision sensors. Its core components include providing motion sensing data of the carrier and outputting real-time triaxial angular velocity measurements in the carrier coordinate system. Compared with triaxial force measurement value An inertial measurement unit (IMU) is used to receive satellite signals to provide global positioning observations and output pseudorange. phase with carrier A Global Navigation Satellite System (GNSS) receiver providing raw observations; used for wireless ranging with known coordinate base stations to provide local relative distance observations and output real-time ranging. Ultra-wideband (UWB) tags; used to capture environmental visual images, provide relative positioning observations by matching with a visual feature point coordinate library, and output feature point pixel coordinates. Industrial cameras; Spatiotemporal calibration is specifically divided into three levels: spatial, temporal, and intrinsic parameter calibration, and outputs a calibrated sensor fusion terminal.

[0010] Preferably, the measurement data from the vehicle-mounted inertial measurement unit in the calibrated sensor fusion terminal is read, and the measurement data is calculated to obtain the carrier predictive positioning. Continuously read the measurement data output in real time from the inertial measurement unit (IMU) within the calibrated sensor fusion terminal, including triaxial angular velocity measurements in the carrier coordinate system. Compared with triaxial force measurement value ; The body attitude is adopted by rotation quaternions from the carrier coordinate system to the unified construction coordinate system. Describe; the carrier at any time Predicted posture The posture of the previous moment The velocity prediction value is obtained by combining angular velocity recursion. The predicted velocity is obtained by performing discrete-time numerical integration on the continuous-time velocity differential equation; Next, the predicted location of the carrier. It is obtained by integrating the velocity; The predicted motion state of the output vehicle at the current moment, calculated by inertial navigation, is collectively referred to as vehicle predictive positioning. , .

[0011] Preferably, based on the calibrated sensor fusion terminal, the terminal observations collected include: Real-time observations from calibrated sensor fusion terminals are collected, including pseudorange data from GNSS receivers. phase with carrier , and known coordinate base station Ultra-wideband (UWB) ranging ; and real-time images captured by industrial cameras.

[0012] Preferably, the carrier's predicted positioning is corrected by combining the terminal observations and the visual feature point coordinate library, resulting in corrected carrier positioning and a confidence level for the positioning result, including: Predicting attitude using the carrier Prediction speed With predicted location This serves as the nominal state at the current moment, and the error state vector is recursively propagated based on the noise statistics of the inertial measurement unit (IMU). Prediction error covariance matrix A tightly coupled strategy is adopted to directly calculate the observation residuals based on terminal observations. To suppress abnormal observations caused by multipath effects, non-line-of-sight propagation, and visual mismatches, the Huber kernel function is introduced in the update process to robustly weight the aforementioned observation residuals. Subsequently, the filter uses this adjusted robust observation noise variance matrix. Calculate Kalman gain Then perform a state update to obtain the optimal estimate of the error state. Optimal estimate of error state This is then inverted to the nominal state to complete the predictive positioning of the carrier. Corrections; The output contains two distinct results: one is the carrier-corrected positioning after fusion and correction. The first result is the final positioning output at the current moment; the second is the confidence level of the positioning result, which is directly derived from the updated error state estimation covariance matrix. .

[0013] Preferably, the carrier calibration mechanism, based on the confidence level of the positioning results, includes: The carrier calibration mechanism is based on a logical OR judgment of the following two events, and the judgment logic deeply integrates the assessment of the confidence level of the positioning result: The first step is carrier static event detection, which measures the angular velocity modulus of the carrier over several consecutive time periods. Continuously below the preset angular velocity static threshold And velocity modulus The speed remains below the preset static threshold. At this point, it is initially determined that the carrier is stationary; based on this initial determination, the positional uncertainty of the current positioning result is further examined; and the error state estimation covariance matrix is ​​extracted. Mid-position error The corresponding 3x3 submatrix and calculate its trace ;like Exceeding the preset uncertainty threshold for static trigger position If so, the calibration process is officially triggered; Secondly, there are known visual feature recognition events, where images captured in real-time by an industrial camera mounted on a carrier are successfully matched with a visual feature database, and one or more known precise three-dimensional coordinates are identified. This event is triggered when a feature point is found; after a successful match, the covariance of the reprojection error of all matched feature points is calculated, or a local observation confidence index is calculated based on the number and spatial distribution of matched points. ; Obtain the overall confidence index of the current fusion positioning; If the local confidence index of visual observation If the overall confidence level is significantly higher than the current level, the calibration process will be strongly triggered; otherwise, if the visual observation confidence level is still low, only a routine state update may be performed, without rushing to calibrate the sensor parameters.

[0014] Preferably, carrier calibration and positioning includes: The final output is the carrier calibration and positioning after the calibration process is triggered by the carrier calibration mechanism. If no carrier calibration mechanism is triggered, this output will be aligned with the carrier correction positioning of the input. same.

[0015] A real-time positioning and calibration system for dynamic construction scenarios in engineering surveying includes a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps of the method.

[0016] This invention provides a real-time positioning and calibration method for dynamic construction scenarios in engineering surveying, which involves deep learning technology and has the following beneficial effects: (1) By adopting an error-state Kalman filter framework and a tightly coupled fusion strategy, the terminal observations from the sensor fusion terminal can be optimally fused under a unified mathematical model. This method does not rely on the independent positioning results of any sensor, but directly utilizes the original observation geometry, giving full play to the advantages of each sensor. It can also maintain positioning by relying on other sensors when some sensors fail or are interfered with, thus achieving higher and more robust positioning accuracy than loosely coupled or single-sensor schemes in complex construction electromagnetic and obstruction environments.

[0017] (2) By introducing the Huber kernel function to adaptively weight various observation residuals, the weight of large residual abnormal observations caused by multipath effects, non-line-of-sight propagation, and visual mismatch can be dynamically reduced when calculating the Kalman gain. This mechanism makes the fusion algorithm insensitive to occasional and unpredictable abnormal observations, avoids outlier contamination of state estimation, and significantly improves the anti-interference ability and output reliability of the entire positioning method in real, complex, and non-ideal observation environments.

[0018] (3) By designing an event-triggered online self-calibration mechanism based on the stationary position of the carrier and the recognition of known visual feature points, the process can intelligently estimate and correct the core error parameter zero bias of the inertial sensor at specific times using strong constraints. This mechanism does not rely on external equipment or manual intervention, achieving closed-loop autonomous long-term maintenance. It fundamentally suppresses the inherent drift of inertial navigation, ensures the long-term stability of fused positioning accuracy, and enables the process to maintain high-precision autonomous calculation for a longer period of time by relying on the online-calibrated inertial measurement unit when absolute observations such as GNSS are temporarily unavailable. Attached Figure Description

[0019] Figure 1 This is a flowchart of a real-time positioning and calibration method for dynamic construction scenarios in engineering surveying, as proposed in this invention.

[0020] Figure 2 This invention provides a method for real-time positioning calibration in dynamic construction scenarios for engineering surveying, which yields a hierarchical map of carrier predicted positioning.

[0021] Figure 3 This is a hierarchical diagram of carrier calibration and positioning obtained in a real-time positioning and calibration method for dynamic construction scenarios in engineering measurement proposed in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figures 1-3 This invention provides a technical solution: a real-time positioning and calibration method for dynamic construction scenarios in engineering surveying. Specifically, the following real-time positioning and calibration method for dynamic construction scenarios in engineering surveying is provided; please refer to [link / reference]. Figure 1 The method includes the following steps: Step S1: Deploy the composite base station and measure its coordinates, and establish a visual feature point coordinate library; install and calibrate the sensor fusion terminal to obtain the calibrated sensor fusion terminal.

[0024] First, the composite base station is deployed and its coordinates are measured. The composite base station consists of an ultra-wideband (UWB) base station module, a power supply unit, and a wireless network communication module. It is a fixed beacon capable of transmitting specific wireless signals for terminal ranging.

[0025] The specific deployment work needs to be planned based on the scope, terrain, and main work routes of the construction area. The deployment process includes the following key aspects: Site selection must ensure that the base station location is stable and has a wide field of vision, without continuous obstruction from large metal objects, to guarantee line-of-sight communication with mobile terminals; network planning needs to consider the geometric distribution between base stations, striving to form a good intersection pattern to improve the geometric strength of subsequent positioning calculations; at the same time, the site should be convenient for power supply and communication to ensure the base station can operate continuously and stably. After the planning is completed, the composite base station hardware will be installed at the selected site.

[0026] After the composite base stations are installed and in place, external high-precision measuring equipment, such as a total station, is used to measure the absolute coordinates of each base station. Through resection mapping, the precise three-dimensional coordinates of each base station in a unified construction coordinate system are directly determined. The coordinates of each base station can be formally represented as a position vector. .in, Indicates the first Location coordinate vectors of each base station; This is the index number of the base station, used to uniquely identify each base station in the network; These represent the coordinate components of the base station in a three-dimensional Cartesian coordinate system, specifically the X, Y, and Z axes. The purpose of this mathematical expression is to quantitatively characterize the absolute position of each base station in a unified construction coordinate system, providing a set of known and fixed control points for all subsequent positioning calculations based on geometric relationships. The output of this process is a composite base station with all known coordinates.

[0027] The process involved collecting and establishing a visual feature point coordinate database. This process aimed to transform pre-positioned, precisely located, and stable reflective markers into digital assets that could be recognized and utilized by computers. Using an oblique photography camera, data was collected from the construction area, focusing specifically on the aforementioned feature points. Multi-view photogrammetry and feature extraction techniques were employed to acquire the three-dimensional coordinates of each feature point and its corresponding high-resolution multi-view images. From these multi-view images, scale- and rotation-invariant visual feature descriptors—high-dimensional mathematical vectors—were extracted to encode the appearance, texture, and local geometric features of the feature points, ensuring reliable identification at different viewpoints and distances. The three-dimensional coordinates of each feature point were associated with its visual feature descriptor and unique identifier, and stored in a structured manner to establish a visual feature point coordinate database that allows for rapid querying and matching. In the database, the first... The coordinates of the feature points are as follows: .in, Indicates the first The position coordinate vector of each visual feature point; Indicates features, This refers to the index of the feature points in the database. This process generates three-dimensional coordinate components. The purpose of this process is to establish a digital representation of the construction environment, containing ground truth location values ​​and rich texture features. This representation will serve as a reference map for subsequent visual localization algorithms to perform feature matching and location calculation. The output of this process is a visual feature point coordinate library, which stores the coordinates of the feature points. The database, along with its associated visual feature descriptors that serve as key criteria for matching, will provide known coordinate references for subsequent steps in visual matching.

[0028] Next, the sensor fusion terminal was installed and calibrated spatiotemporally. The installed sensor fusion terminal is a hardware suite integrating multiple high-precision sensors. Its core components include providing motion sensing data of the carrier and outputting real-time triaxial angular velocity measurements in the carrier coordinate system. Compared with triaxial force measurement value An inertial measurement unit (IMU) is used to receive satellite signals to provide global positioning observations and output pseudorange. phase with carrier A GNSS receiver that provides raw observation data; used for wireless ranging with known coordinate base stations to provide local relative distance observations and output real-time ranging data. Ultra-wideband (UWB) tags; used to capture environmental visual images, provide relative positioning observations by matching with a visual feature point coordinate library, and output feature point pixel coordinates. The industrial camera uses a carrier coordinate system fixed to the carrier, with its origin typically located at the measurement center of the inertial measurement unit (IMU), used to represent the relative positional relationships between the sensors on the carrier. All of these sensors are mounted on the carrier, collectively forming a sensor fusion terminal.

[0029] Spatiotemporal calibration is a core preprocessing step to ensure that the data from the various sensors within the terminal can be tightly coupled and fused. It aims to eliminate errors between sensors and achieve a unified spatiotemporal reference. Specifically, it is divided into three levels: spatial, temporal, and intrinsic parameter calibration.

[0030] The core of spatial calibration is determining the boom vector so that, in the subsequent fusion process, all sensor observations can be uniformly converted and processed at the IMU center. To this end, it is necessary to determine the spatial offset vector from the IMU center (the origin) to the physical reference point of each sensor. Specifically, this includes the vector from the IMU center to the center of the GNSS antenna. Vector to industrial camera And the vector to the phase center of the ultra-wideband UWB tag antenna. .in, , , It is a three-dimensional vector. Identify sensor type, This refers to the center of the inertial measurement unit (IMU). The purpose of these lever vectors is to provide accurate geometric parameters for subsequent fusion algorithms, ensuring that observations collected by all sensors at different locations can be uniformly converted to the common point of the IMU center for calculation, thereby eliminating fixed position errors introduced by mechanical installation deviations.

[0031] Time calibration aims to establish a unified time reference at the microsecond level. Utilizing the high-precision second pulse signal output from a GNSS receiver as the master clock source, hardware triggering ensures that the data acquisition start time of the inertial measurement unit (IMU), ultra-wideband (UWB) tag, and industrial camera is strictly synchronized with the rising edge of the second pulse. This process guarantees consistent timestamps across multiple data streams, meeting the strict time alignment requirements of tightly coupled fusion algorithms, avoiding significant integration errors due to time asynchrony in dynamic scenarios, and thus ensuring the instantaneous correspondence between state prediction and observation updates.

[0032] Intrinsic parameter calibration is specifically designed for industrial cameras to determine their internal imaging geometry. By photographing a calibration board with a known pattern from multiple angles, the camera's intrinsic parameter matrix can be calculated. With lens distortion coefficient vector Intrinsic parameter matrix Includes focal length , and principal point coordinates , Key parameters such as lens distortion coefficient vector. Used to correct radial and tangential distortion of the lens. The purpose of intrinsic parameter calibration is to establish an accurate projection and backprojection model from three-dimensional spatial points to two-dimensional image pixel coordinates. This is the geometric basis for calculating reprojection errors in subsequent visual positioning, and its accuracy directly determines the validity of visual observations.

[0033] By completing the calibration of spatial, temporal, and intrinsic parameters, the sensor fusion terminal is transformed into a data acquisition unit with a unified spatiotemporal reference and an accurate observation model. The output of this process is a calibrated sensor fusion terminal, a sensor suite capable of outputting spatiotemporally unified raw data, which will serve as the sole source for all subsequent real-time data acquisition.

[0034] Step S2: Read the measurement data of the vehicle-mounted inertial measurement unit in the calibrated sensor fusion terminal, calculate the measurement data, and obtain the carrier predictive positioning.

[0035] Continuously read the measurement data output in real time from the inertial measurement unit (IMU) within the calibrated sensor fusion terminal, including triaxial angular velocity measurements in the carrier coordinate system. Compared with triaxial force measurement value These data form the initial observational basis for inertial navigation calculations. The core of the processing lies in the mechanical orchestration calculation of inertial navigation, which uses the optimal state estimate of the vehicle at the previous moment after fusion optimization as the initial condition for this recursion. All position, velocity, and attitude parameters relied upon and output in this process are described and calculated under a unified construction coordinate system. The purpose of using a unified construction coordinate system is to ensure the predicted positioning of the vehicle obtained from the inertial navigation calculations. It can be directly compared and fused with observation constraints from known coordinate base stations and known coordinate feature points in the same geometric space, thus providing a consistent mathematical basis for subsequent tight-coupled data correction.

[0036] First, calculate the spatial orientation of the carrier under the unified construction coordinate system. The carrier attitude is determined using a rotation quaternion from the carrier coordinate system to the unified construction coordinate system. Describe the rate of change of the attitude quaternion over time. It follows the following quaternion differential equation: ; in, It is a quaternion describing the attitude of the carrier; This represents quaternion multiplication. It is the projection of the rotational angular velocity of the carrier relative to the unified construction coordinate system onto the carrier coordinate system, derived from the original triaxial angular velocity measurements. This was obtained after compensating for effects such as Earth's rotation. This represents the derivative with respect to time. The purpose of this differential equation is to establish an accurate mathematical model between the rate of change of the carrier's attitude and its angular velocity.

[0037] In practical discrete-time processing, attitude recursion is achieved by numerical integration of this equation. The carrier at time... Predicted posture The posture of the previous moment Based on the recursive derivation of angular velocity, its recursive relationship can be expressed as: ; in, This represents the operation of mapping a rotation vector to an incremental quaternion; This represents the step size for the discrete-time recursion. The purpose of this recursive formula is to transform the continuous attitude dynamics equations into a discrete-time iterative algorithm that can be executed in a processor, thereby achieving real-time recursive attitude calculation.

[0038] To obtain the predicted attitude at the current moment Then, the velocity and position of the carrier are calculated in a unified construction coordinate system using triaxial force measurements. This calculation follows the velocity differential equation: ; in, This represents the time derivative of the velocity of the carrier relative to the Earth in a unified construction coordinate system. It is based on the current predicted attitude quaternion The calculated rotation matrix is ​​used to transform the triaxial force measurement values ​​from the carrier coordinate system to the unified construction coordinate system; It is the projection of the Earth's rotational angular velocity onto a unified construction coordinate system; It is the angular velocity of the unified construction coordinate system relative to the Earth, which is related to the speed of the carrier; Represents the cross product of vectors; It is the local gravity vector.

[0039] Speed ​​prediction value The velocity is obtained by performing discrete-time numerical integration on the continuous-time velocity differential equation. Specifically, the velocity at the known previous moment... Using the initial values, the terms on the right-hand side of the differential equation are approximated and summed to obtain an equivalent average acceleration, which is then multiplied by the time step. The velocity increment is obtained and then added to the initial velocity. Its recursive form can be expressed as: ; The purpose of this recursive formula is to transform the complex velocity change model into a discrete, step-by-step numerical calculation, thereby recursively calculating the predicted velocity of the carrier at the current moment using the triaxial specific force measurements from the inertial measurement unit (IMU), combined with known geophysical parameters and the state at the previous moment. .

[0040] After obtaining the predicted speed Next, the predicted location of the carrier. This is obtained by integrating the velocity. Position and velocity satisfy the fundamental kinematic relationship, i.e., the time derivative of position equals velocity, expressed as... The purpose of this relationship is to establish a fundamental integral connection between position and velocity. In the discrete-time implementation, this relationship is numerically integrated to obtain the position at the previous time step. Starting from the calculated predicted velocity, a recursive approach is used. The recursive relationship can be expressed as: ; The purpose of this formula is to discretize the continuous position integral relationship and recursively calculate the predicted position of the carrier at the current moment by accumulating the estimated displacement of the carrier movement within each time step.

[0041] The process ultimately outputs the predicted motion state of the carrier at the current moment, calculated by inertial navigation, collectively referred to as carrier predictive positioning. , This carrier predicts positioning. This will serve as a key comparison benchmark and state prediction value for subsequent multi-source data fusion and correction. At the same time, the predicted state output at this moment will also serve as the initial condition for the inertial recursive calculation at the next moment, thereby realizing continuous and recursive estimation of the carrier's motion state, forming a high-frequency and continuous trajectory skeleton in the entire positioning process.

[0042] Step S3: Based on the calibrated sensor fusion terminal, collect terminal observation values, and combine the terminal observation values ​​with the visual feature point coordinate library to correct the carrier's predicted positioning, thereby obtaining the carrier's corrected positioning and the confidence level of the positioning result; First, real-time observations are collected from the calibrated sensor fusion terminal. These include pseudorange data from the GNSS receiver. phase with carrier , and known coordinate base station Ultra-wideband (UWB) ranging The system also includes real-time images captured by industrial cameras. Next, the real-time images captured by the industrial cameras are processed to extract visual feature points and calculate their visual feature descriptors. Subsequently, feature matching is achieved by comparing the similarity of the visual feature descriptors with pre-stored visual feature descriptors in a visual feature point coordinate library, thus obtaining the pixel coordinates of the feature points. and their corresponding known three-dimensional coordinates .

[0043] The core of the processing is the recursive operation of the error state Kalman filter, which starts from the optimal state of the previous time step to predict and locate the carrier. Tightly coupled fusion and optimal correction are performed. This filter does not directly estimate the absolute state of the carrier, but rather estimates its position relative to the carrier's predicted location, which is considered the nominal state. The deviation between them, its error state vector ,in Indicates three-dimensional position error. Indicates the three-dimensional velocity error. Indicates the three-dimensional attitude error. and These represent the zero bias errors of the gyroscope and acceleration, respectively. The aim is to unify all error quantities that need to be corrected online into a mathematical framework for optimal estimation.

[0044] To ensure that terminal observations can be meaningfully compared with the nominal state, which serves as a unified benchmark, the known coordinates acquired at their respective sensor reference points must be uniformly transformed to the measurement center of the inertial measurement unit (IMU) corresponding to that nominal state before calculating any observation residuals. This transformation utilizes the lever vector and the vehicle's predicted positioning at the current moment. Completed. Specifically, for a known coordinate point, if it was measured using a certain sensor reference point, its position corresponding to the center of the inertial measurement unit (IMU) in the unified construction coordinate system is calculated as follows: For known base station coordinates... In the calculation and carrier prediction position When determining the geometric distance, the actual base station coordinates used, converted to the center of the inertial measurement unit (IMU), should be: ,in This represents the vector pointing from the center of the UWB tag antenna to the center of the inertial measurement unit (IMU). Similarly, for the coordinates of visual feature points... Its coordinates are converted to the center of the inertial measurement unit (IMU). ,in This represents the vector pointing from the center of the industrial camera antenna to the center of the inertial measurement unit (IMU). For GNSS satellites, due to the extremely large distance between the satellite and the ground, the vector direction between the phase centers of the satellite and the receiver antenna is negligibly affected by the small lever arm. Therefore, in subsequent calculations, the satellite coordinates can be approximated as... No lever compensation is required for this purpose. After the above-mentioned coordinate unification preprocessing, all subsequent calculations of observation residuals are performed on the common geometric reference point of the IMU center, ensuring consistency.

[0045] First, the attitude is predicted based on the carrier. Prediction speed With predicted location This serves as the nominal state at the current moment, and the error state vector is recursively propagated based on the noise statistics of the inertial measurement unit (IMU). Prediction error covariance matrix This is used to quantify the uncertainty of the predicted value of the current nominal state.

[0046] The algorithm then employs a tightly coupled strategy, directly calculating the observation residual based on the terminal observations; that is, the difference between the actual observations and the observations predicted based on the current nominal state. For GNSS pseudorange observations, the pseudorange residual... The calculation is as follows: ; in, Nominal position ; It refers to the satellite's location; It's the speed of light; The receiver clock bias to be estimated is the residual, which reflects the error in position and clock bias prediction.

[0047] For ultra-wideband (UWB) ranging observations, the ranging residual... The calculation is as follows: ; in, Nominal position ; These are the known coordinates of the base station. This residual directly provides the geometric distance error between the carrier and the known fixed point.

[0048] For visual observation, the reprojection residual is a two-dimensional vector. : ; in, It is a nominal posture The calculated rotation matrix; It is the camera projection function; Nominal position This residual is a strong constraint for optimizing carrier positioning.

[0049] Of all the above observation residuals, the known coordinates of the base station and visual feature point coordinates The coordinates are in the unified construction coordinate system corresponding to the center of the inertial measurement unit (IMU). and Pseudorange residuals of all GNSS satellites Ranging residuals of all UWB base stations and the reprojection residuals of all matching visual feature points The two pixel components are stacked in a predetermined order to form the observation residual vector at the current moment. .

[0050] To suppress anomalous observations caused by multipath effects, non-line-of-sight propagation, and visual mismatches, the Huber kernel function is introduced in the update process to robustly weight the aforementioned observation residuals. This is achieved through a weighting function... The implementation, its expression is: ; in, It is the scalar residual of a certain observation; The threshold for distinguishing between normal and outlier values ​​is set based on the sensor's noise characteristics; for example, for a nominal noise standard deviation of [value missing]. UWB ranging of meters, setting Meters. The purpose of this function is to dynamically evaluate the reliability of each observation: when the absolute value of the residual is... Less than or equal to the threshold At that time, the observation is given full weight. ;when Greater than When it is considered potentially abnormal, the weights are adjusted accordingly. Reduce, thereby lessening its impact.

[0051] This weighting function acts directly on the observation noise covariance matrix. ,standard It is a diagonal matrix, and its elements are... The prior noise variance of each observation, after being weighted by the Huber kernel function, is the equivalent robust observation noise variance matrix. This means that for normal observation Its noise level remains unchanged; for anomalous observations Its equivalent noise variance It is amplified and characterized as unreliable.

[0052] Subsequently, the filter uses this adjusted robust observation noise variance matrix. Calculate Kalman gain Then, a state update is performed. The complete update process is described by the following set of equations: ; ; ; In this system of equations, This represents the calculated Kalman gain matrix; Represents the error state vector from the prediction step. The prediction error covariance matrix; This represents the observation matrix, i.e., the error state vector of various observation residual models. The Jacobian matrix; The observation residual vector is formed by stacking. This represents the optimal estimate of the error state to be determined. This represents the updated error state estimate covariance matrix. The purpose of this system of equations is to perform an optimal data fusion based on minimizing the mean squared error of the estimate.

[0053] Its core mechanism lies in calculating the Kalman gain using the first formula. The calculation process involves uncertainty in state prediction. With the uncertainty of observation This involves making an optimal trade-off between the two. The robustness of the Huber kernel function is centrally demonstrated here: for observations judged as anomalous, its performance is optimized. The corresponding noise variance is significantly increased, which leads to difficulties in calculating the inverse matrix. At that time, the contribution of the observation was suppressed, ultimately resulting in a decrease in the calculated gain. The component values ​​corresponding to the anomaly observations become extremely small. Then, in the second formula... In this process, this extremely small gain component will significantly reduce the impact of the anomalous residual on the final state correction. The third formula simultaneously updates the quantitative assessment of uncertainty, yielding the updated error state estimate covariance matrix. .

[0054] To obtain the optimal estimate of the error state Optimal estimate of error state This is then inverted to the nominal state to complete the predictive positioning of the carrier. Correction: The position is corrected to: ; Speed ​​correction: ; The attitude was corrected to: ; in, Represents quaternion multiplication; This represents the operation of converting a rotation vector into a gain quaternion. The purpose of these correction formulas is to directly apply the precise deviation calculated from the observation information to compensate for the prediction error of inertial navigation, thereby obtaining a higher accuracy positioning result.

[0055] This process ultimately outputs two distinct results: one is the carrier-corrected positioning after fusion and correction. The first result is the final positioning output at the current moment; the second is the confidence level of the positioning result, which is directly derived from the updated error state estimation covariance matrix. The elements on the diagonal of this matrix quantitatively characterize the uncertainty level of the fused positioning results; the smaller the value, the higher the confidence level of the positioning results. This carrier-corrected positioning... The location reliability will serve as the core input for the backend intelligent calibration decision. Simultaneously, the optimized state at this moment represents the corrected carrier positioning. and gyroscope zero bias estimate and zero bias estimate of acceleration It comes from the calculated optimal estimate of the error state. In this process, it will serve as the initial state and parameters for inertial deduction and filter prediction at the next moment, thereby realizing the recursion and closed loop of the entire positioning process in time.

[0056] Step S4: Based on the confidence level of the positioning result, establish a carrier calibration mechanism, and calibrate the carrier correction positioning through the carrier calibration mechanism to obtain the carrier calibration positioning.

[0057] This step uses the confidence level of the positioning results output from the previous steps and the carrier-corrected positioning. This forms the basis for monitoring and decision-making. The confidence level of the positioning result is directly derived from the updated error state estimation covariance matrix. The elements on the diagonal of this matrix quantitatively characterize the uncertainty level of each state component in the fused positioning result. The core purpose of the carrier calibration mechanism is to intelligently trigger online estimation of sensor parameters when the current positioning result is determined to have high confidence or the carrier is under special observation conditions that provide strong constraints for parameter estimation, thereby correcting the zero-bias error and suppressing long-term drift in positioning accuracy. The establishment of the calibration mechanism relies heavily on the continuous evaluation of the confidence of the positioning result and on comprehensive decision-making combined with the carrier's motion state and external observation conditions.

[0058] The carrier calibration mechanism is based on a logical OR judgment of the following two events, and the judgment logic deeply integrates the assessment of the confidence level of the positioning result: The first step is the detection of stationary events on the carrier. This detection is achieved by analyzing the triaxial angular velocity measurements from the raw output of the inertial measurement unit (IMU). Compared with triaxial force measurement value And the speed after fusion correction A comprehensive judgment is made. When the carrier's angular velocity modulus exceeds a certain value over several consecutive time periods... Continuously below the preset angular velocity static threshold And velocity modulus The speed remains below the preset static threshold. At that time, it was initially determined that the carrier was in a stationary state. Angular velocity stationary threshold. The value was set to 0.01 rad / s. This value was determined by performing Allen's variance analysis on the MEMS gyroscope of the model used, taking 3.5 times its zero-bias instability index, and combining it with statistics on angular velocity jitter caused by environmental vibrations under typical stationary conditions. (Velocity stationary threshold) The velocity was set to 0.05 m / s, a value determined based on three standard deviations of the velocity estimation error during a one-hour static test of the achieved fusion positioning process. Based on the initial determination of stillness, the positional uncertainty of the positioning result at the current moment was further examined. The error state estimation covariance matrix was extracted. Mid-position error The corresponding 3x3 submatrix and calculate its trace .like Exceeding the preset uncertainty threshold for static trigger position If this occurs, the calibration process is formally triggered. The uncertainty threshold for the stationary trigger position. The value was set at 0.04 square meters. This value was determined through Monte Carlo simulation, simulating the average trace value corresponding to the increase in positioning error covariance to the acceptable accuracy loss boundary under typical construction noise conditions. This design ensures that, when calibration resources are limited, parameter correction is prioritized when the carrier is stationary and the confidence level of the positioning results is significantly reduced, thereby improving calibration efficiency and necessity.

[0059] Secondly, there are events involving the recognition of known visual features. This occurs when an image captured in real-time by an industrial camera mounted on a carrier successfully matches a visual feature database and identifies one or more known, precise 3D coordinates. This event is triggered when a feature point is found. This event signifies that the carrier has obtained an absolutely reliable observation benchmark derived from a high-precision mapping dataset. After a successful match, the covariance of the reprojection error of all matched feature points is calculated, or a local observation confidence index is calculated based on the number and spatial distribution of matched points. Simultaneously, obtain the overall confidence index of the current fused positioning, such as the location error covariance matrix. traces If the local confidence level of visual observation Significantly higher than the current overall confidence index, i.e., meets the requirements This strongly triggers the calibration process. Among these factors, the proportionality coefficient... The value is set to 5. This value was determined through analysis of a large amount of experimental data, based on the empirical relationship that visual observation accuracy is generally more than 5 times better than fused positioning accuracy. Conversely, if the confidence level of visual observation is still low, only routine state updates may be performed, without hastily calibrating sensor parameters. This mechanism allows calibration to intelligently utilize external, highly reliable absolute benchmarks to correct for biases, especially when the confidence level of fused positioning is insufficient.

[0060] The calibration calculation process begins after the carrier calibration mechanism is triggered. Its core is to solve a constrained optimization problem based on current event observations to estimate the sensor's error parameters. Specifically, the parameter to be estimated is the acceleration bias error. Gyroscope zero bias error Under stationary events, calibration utilizes the physical constraint that the theoretical acceleration and angular velocity are zero at rest, acquiring multiple frames of raw measurements from the inertial measurement unit (IMU) during the stationary period. and Zero bias estimate of acceleration This can be obtained by calculating the average specific force deviation over that time period and compensating for the local gravitational field, i.e., by projection summation: ; Gyroscope zero bias estimate The average angular velocity within that segment can then be directly estimated using the measured value. The estimation formula is as follows: ; in, This represents the number of samples taken during a static time period. To estimate the attitude from the current position The corresponding rotation matrix from the unified construction coordinate system to the carrier coordinate system is used to calibrate the zero bias error of the inertial sensor directly using static physical constraints.

[0061] In recognizing events with known visual features, calibration utilizes the extremely high-precision absolute position constraints provided by visual feature points to determine the three-dimensional coordinates of successfully matched feature points. and the pixel coordinates of its feature points Considered as a set of super-strong observations, the fusion positioning from the previous moment is used as the initial value, and the coordinates of feature points are fixed to optimize the current position, attitude, and zero bias error of the inertial gyroscope of the carrier. and Zero bias error This is equivalent to performing a localization map optimization with visual absolute coordinates as anchor points, which can effectively correct cumulative errors and sensor parameter errors.

[0062] A new gyroscope bias estimate is obtained through calibration calculations. and zero bias estimate of acceleration Then, the state correction and output phase begins. These updated parameters are fed back into the state vector. In the next step, the updated zero-bias estimate is used to perform online compensation on the raw IMU data at the current and subsequent times, thereby improving the accuracy of inertial navigation calculations at the source. Simultaneously, based on the confidence level of the calibration results, a fine-tuning of the current vehicle correction positioning can be performed. Specifically, for two different calibration trigger events, the fine-tuning is typically performed as follows: In the case of a stationary vehicle event, the fine-tuning directly utilizes the strong physical constraints that the theoretical velocity and theoretical angular velocity are zero at rest. The most direct operation is to adjust the previous vehicle correction velocity estimate... Reset to zero to eliminate velocity drift. Simultaneously, the characteristic that the average specific force measurement during the stationary period points in the opposite direction of gravity in the carrier coordinate system can be used to correct the current carrier attitude. A quick refinement is performed to obtain the refined rotation matrix, which can be converted back to the calibrated pose obtained from quaternions. Furthermore, while the ideal position remains unchanged at rest, fine-tuning may involve significantly reducing the covariance of the position estimate to increase its confidence. In fine-tuning of visual feature point events, the adjustment of pose and velocity is achieved by solving for a vector containing the complete state vector. This is achieved through optimization. Its goal is to minimize the objective function. Within this framework, attitude parameters As a core variable, its update direction is directly determined by the gradient of the objective function, thus minimizing reprojection error most effectively. Velocity, as part of the state vector, adjusts synchronously with the iterative update of the entire state vector, eventually converging to an optimal estimate that is consistent with the refined pose and sensor parameters in the dynamic model. This yields the final calibrated positioning output. The final output of this process is the calibrated positioning of the carrier after calibration triggered by the carrier calibration mechanism. , If no carrier calibration mechanism is triggered, this output will be aligned with the carrier correction positioning of the input. same.

[0063] This invention provides a real-time positioning and calibration method for dynamic construction scenarios in engineering surveying. The method first deploys known-location base stations in the construction area and establishes a visual feature point coordinate library containing precise three-dimensional coordinates and visual feature descriptors. Simultaneously, it performs rigorous spatiotemporal calibration on a multi-sensor fusion terminal integrated on a carrier, providing known physical references and accurate observation models for subsequent positioning.

[0064] In the real-time positioning phase, the raw measurement data from the inertial measurement unit (IMU) is continuously read and integrated. Through inertial navigation mechanical orchestration calculations, starting from the previous optimal state, the predicted pose of the vehicle in a unified construction coordinate system is recursively obtained, including predicted attitude, predicted velocity, and predicted position. This step provides high-frequency continuous trajectory prediction, but it will drift over time. Next, observations from multiple sources, including GNSS pseudorange and carrier phase, UWB ranging, and industrial camera images, are simultaneously acquired. Real-time images are matched with a feature point coordinate library to obtain the pixel coordinates of the feature points and their corresponding known 3D coordinates. An error-state Kalman filter is run, using the aforementioned predicted vehicle pose as the nominal state, and the residuals of each raw observation are directly calculated in a tightly coupled manner. To suppress anomalous observations, a Huber kernel function is introduced to robustly weight the observation residuals. Finally, the correction amount for the predicted vehicle pose is calculated through optimal data fusion, outputting the fused and optimized high-precision corrected vehicle positioning and a quantitative assessment of its uncertainty, i.e., the confidence level of the positioning result.

[0065] A carrier calibration mechanism is established based on the confidence level of the positioning results and the carrier's motion state. The calibration process is automatically triggered when two specific events are detected: the carrier is detected to be stationary, or a known visual feature point is identified. This process utilizes the strong physical or geometric constraints provided by the events to estimate and update the acceleration of the inertial measurement unit (IMU) and the zero-bias error of the gyroscope online. New... The partial estimate is fed back in real time to compensate for the original inertial measurement unit (IMU) data, thereby suppressing error drift at the source and obtaining the final online-calibrated carrier positioning.

[0066] This invention also protects a real-time positioning and calibration system for dynamic construction scenarios in engineering surveying, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0067] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, the phrase "comprising an element defined as..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0068] 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 likenesses.

Claims

1. A real-time positioning and calibration method for dynamic construction scenarios in engineering surveying, characterized in that, Includes the following steps: Step S1: Deploy the composite base station and measure its coordinates, and establish a visual feature point coordinate library; install and calibrate the sensor fusion terminal to obtain the calibrated sensor fusion terminal; Step S2: Read the measurement data of the vehicle-mounted inertial measurement unit in the calibrated sensor fusion terminal, calculate the measurement data, and obtain the carrier predictive positioning; Step S3: Based on the calibrated sensor fusion terminal, collect terminal observation values, and combine the terminal observation values ​​with the visual feature point coordinate library to correct the carrier's predicted positioning, thereby obtaining the carrier's corrected positioning and the confidence level of the positioning result; Step S4: Based on the confidence level of the positioning result, establish a carrier calibration mechanism, and calibrate the carrier correction positioning through the carrier calibration mechanism to obtain the carrier calibration positioning.

2. The real-time positioning and calibration method for dynamic construction scenarios in engineering surveying according to claim 1, characterized in that, Deploying composite base stations and measuring their coordinates includes: The composite base station consists of an ultra-wideband (UWB) base station module, a power supply unit, and a wireless network communication module. It is a fixed beacon that can transmit wireless signals for terminal ranging. The specific deployment work needs to be planned according to the scope of the construction area, the terrain, and the main work routes. After the composite base station is installed, external high-precision measuring equipment is used to measure the absolute coordinates of each base station. Through resection mapping, the precise three-dimensional coordinates of each base station in a unified construction coordinate system are directly determined.

3. The real-time positioning and calibration method for dynamic construction scenarios in engineering surveying according to claim 2, characterized in that, Establish a visual feature point coordinate library, including: The process involves collecting and establishing a visual feature point coordinate library. Pre-deployed, precisely located, and stable reflective points are used as feature points. A tilting camera is used to collect data from the construction area. Multi-view photogrammetry and feature extraction techniques are employed to obtain the 3D coordinates of each feature point and its corresponding high-resolution multi-view image. Scale- and rotation-invariant visual feature descriptors are extracted from these multi-view images to encode the appearance, texture, and local geometric features of the feature points. The 3D coordinates of each feature point are associated with its visual feature descriptor and identifier, and stored in a structured manner to establish the visual feature point coordinate library.

4. The real-time positioning and calibration method for dynamic construction scenarios in engineering surveying according to claim 3, characterized in that, The sensor fusion terminal is installed and spatiotemporally calibrated to obtain a calibrated sensor fusion terminal, including: The sensor fusion terminal is a hardware suite that integrates multiple high-precision sensors. Its core components include providing motion sensing data of the carrier and outputting real-time triaxial angular velocity measurements in the carrier coordinate system. Compared with triaxial force measurement value An inertial measurement unit (IMU) is used to receive satellite signals to provide global positioning observations and output pseudorange. phase with carrier A Global Navigation Satellite System (GNSS) receiver providing raw observations; used for wireless ranging with known coordinate base stations to provide local relative distance observations and output real-time ranging. Ultra-wideband (UWB) tags; used to capture environmental visual images, provide relative positioning observations by matching with a visual feature point coordinate library, and output feature point pixel coordinates. Industrial cameras; Spatiotemporal calibration is specifically divided into three levels: spatial, temporal, and intrinsic parameter calibration, and outputs a calibrated sensor fusion terminal.

5. The real-time positioning and calibration method for dynamic construction scenarios in engineering surveying according to claim 4, characterized in that, Reading measurement data from the vehicle-mounted inertial measurement unit in the calibrated sensor fusion terminal, and performing calculations on the measurement data to obtain the carrier's predicted positioning, includes: Continuously read the measurement data output in real time from the inertial measurement unit (IMU) within the calibrated sensor fusion terminal, including triaxial angular velocity measurements in the carrier coordinate system. Compared with triaxial force measurement value ; The body posture is achieved by rotating quaternions from the carrier coordinate system to the unified construction coordinate system. Describe; the carrier at any time Predicted posture The posture of the previous moment The velocity prediction value is obtained by combining angular velocity recursion. The predicted velocity is obtained by performing discrete-time numerical integration on the continuous-time velocity differential equation; Next, the predicted location of the carrier. It is obtained by integrating the velocity; The predicted motion state of the output vehicle at the current moment, calculated by inertial navigation, is collectively referred to as vehicle predictive positioning. , .

6. The real-time positioning and calibration method for dynamic construction scenarios in engineering surveying according to claim 5, characterized in that, Based on the calibrated sensor fusion terminal, the terminal observations are collected, including: Real-time observations from calibrated sensor fusion terminals are collected, including pseudorange data from GNSS receivers. phase with carrier , and known coordinate base station Ultra-wideband (UWB) ranging ; and real-time images captured by industrial cameras.

7. The real-time positioning and calibration method for dynamic construction scenarios in engineering surveying according to claim 6, characterized in that, The predicted positioning of the carrier is corrected by combining the terminal observations and the visual feature point coordinate library to obtain the corrected positioning of the carrier and the confidence level of the positioning result, including: Predicting attitude using the carrier Prediction speed With predicted location This serves as the nominal state at the current moment, and the error state vector is recursively propagated based on the noise statistics of the inertial measurement unit (IMU). Prediction error covariance matrix A tightly coupled strategy is adopted to directly calculate the observation residuals based on terminal observations. To suppress abnormal observations caused by multipath effects, non-line-of-sight propagation, and visual mismatches, the Huber kernel function is introduced in the update process to robustly weight the aforementioned observation residuals. Subsequently, the filter uses this adjusted robust observation noise variance matrix. Calculate Kalman gain Then perform a state update to obtain the optimal estimate of the error state. Optimal estimate of error state This is then inverted to the nominal state to complete the predictive positioning of the carrier. Corrections; The output contains two distinct results: one is the carrier-corrected positioning after fusion and correction. The first result is the final positioning output at the current moment; the second is the confidence level of the positioning result, which is directly derived from the updated error state estimation covariance matrix. .

8. The real-time positioning and calibration method for dynamic construction scenarios in engineering surveying according to claim 7, characterized in that, Based on the confidence level of the positioning results, a carrier calibration mechanism is established, including: The carrier calibration mechanism is based on a logical OR judgment of the following two events, and the judgment logic deeply integrates the assessment of the confidence level of the positioning result: The first step is carrier static event detection, which measures the angular velocity modulus of the carrier over several consecutive time periods. Continuously below the preset angular velocity static threshold And velocity modulus The speed remains below the preset static threshold. At this point, it is initially determined that the carrier is stationary; based on this initial determination, the positional uncertainty of the current positioning result is further examined; and the error state estimation covariance matrix is ​​extracted. Mid-position error The corresponding 3x3 submatrix and calculate its trace ;like Exceeding the preset uncertainty threshold for static trigger position If so, the calibration process is officially triggered; Secondly, there are known visual feature recognition events, where images captured in real-time by an industrial camera mounted on a carrier are successfully matched with a visual feature database, and one or more known precise three-dimensional coordinates are identified. This event is triggered when a feature point is found; after a successful match, the covariance of the reprojection error of all matched feature points is calculated, or a local observation confidence index is calculated based on the number and spatial distribution of matched points. ; Obtain the overall confidence index of the current fusion positioning; If the local confidence index of visual observation If the overall confidence level is significantly higher than the current level, the calibration process will be strongly triggered; otherwise, if the visual observation confidence level is still low, only a routine state update may be performed, without rushing to calibrate the sensor parameters.

9. The real-time positioning and calibration method for dynamic construction scenarios in engineering surveying according to claim 8, characterized in that, Carrier calibration and positioning, including: The final output is the carrier calibration and positioning after the calibration process is triggered by the carrier calibration mechanism. If no carrier calibration mechanism is triggered, this output will be aligned with the carrier correction positioning of the input. same.

10. A real-time positioning and calibration system for dynamic construction scenarios in engineering surveying, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-9.

Images