RTK-based intelligent outdoor positioning method, device and system and storage medium

Abstract

The invention relates to an RTK-based intelligent outdoor positioning method, system and device and a storage medium. The method comprises the following steps: acquiring a carrier phase observation value, inertial measurement data and environmental data through a positioning terminal; carrying out anomaly diagnosis and correlation analysis on the carrier phase observation value based on the environment data, and generating a dynamic observation weight and a phase correction amount; performing tight coupling fusion solution on the inertial measurement data according to the dynamic observation weight and the phase correction to obtain fusion data; and performing direction constraint and carrier phase fixation on the fused data according to the environment data to obtain a positioning result. According to the invention, the technology spanning from passive receiving to active adaptation can be realized, and the precision and reliability of RTK positioning in a harsh outdoor scene are significantly improved.

Description

Technical Field

[0001] This invention relates to the technical field of outdoor positioning, and in particular to an intelligent outdoor positioning method, device, system and storage medium based on RTK. Background Technology

[0002] Real-time dynamic carrier phase differential (RTK) technology is currently the core method for achieving centimeter-level positioning outdoors, and it is widely used in surveying, autonomous driving, drones, and precision agriculture. However, as applications extend to complex scenarios such as urban canyons, under overpasses, and tree-lined roads, traditional RTK technology faces severe challenges: under conditions of severe signal obstruction, reflection, and other multipath interference, positioning results are prone to jumps, sharp drops in accuracy, or even loss of lock. Existing methods mainly rely on carrier phase observations under ideal conditions, lacking the ability to collaboratively perceive and intelligently process multi-source interference factors in complex environments. They fail to fully utilize the data from the terminal's built-in inertial and environmental sensors to form deep complementarity, and cannot construct a fusion mechanism that dynamically adjusts the reliability of observation data based on the real-time environment and actively compensates for system errors. This results in difficulties in continuously outputting stable and reliable high-precision positioning results under harsh outdoor conditions. Summary of the Invention

[0003] The main objective of this invention is to provide an intelligent outdoor positioning method, device, system, and storage medium based on RTK, which can achieve a technological leap from passive reception to active adaptation, and significantly improve the accuracy and reliability of RTK positioning in harsh outdoor scenarios.

[0004] To achieve the above objectives, the present invention provides an RTK-based intelligent outdoor positioning method, comprising: The positioning terminal acquires carrier phase observations, inertial measurement data, and environmental data. Based on the environmental data, anomaly diagnosis and correlation analysis are performed on the carrier phase observation values ​​to generate dynamic observation weights and phase correction values. The inertial measurement data is tightly coupled and fused based on the dynamic observation weights and the phase correction amount to obtain fused data; Based on the environmental data, the fused data is subjected to directional constraints and carrier phase fixation to obtain the positioning result.

[0005] Furthermore, the acquisition of carrier phase observations, inertial measurement data, and environmental data through the positioning terminal includes: The original carrier phase signal is obtained and extracted through the GNSS receiving module of the positioning terminal, and the original carrier phase signal is converted into the carrier phase observation value; The inertial measurement module of the positioning terminal acquires angular velocity and acceleration readings along three orthogonal axes, and integrates the angular velocity and acceleration readings to obtain the inertial measurement data. The environmental data is acquired through the environmental sensors of the positioning terminal.

[0006] Furthermore, the step of performing anomaly diagnosis and correlation analysis on the carrier phase observations based on the environmental data to generate dynamic observation weights and phase correction values ​​includes: Extract environmental feature parameters corresponding to the carrier phase observation values ​​from the environmental data; The carrier phase observation value is compared with a preset phase quality threshold. If the carrier phase observation value exceeds the phase quality threshold, the carrier phase observation value is marked as an observation value to be verified; otherwise, the carrier phase observation value is determined to be a normal observation value. If it is detected that both the observation to be verified and the environmental feature parameters meet the preset interference conditions, then the observation to be verified is determined to be an abnormal observation. The dynamic observation weights are obtained by performing confidence calculations and weight settings on the observations to be verified based on the environmental characteristic parameters. Based on the preset interference conditions, the abnormal observation values ​​are deviated and the deviated data is combined with the environmental feature parameters to perform phase compensation, thereby obtaining the phase correction amount.

[0007] Further, the step of performing tightly coupled fusion calculation on the inertial measurement data based on the dynamic observation weights and the phase correction amount to obtain fused data includes: The acquisition time period of the positioning terminal is obtained, and the motion state is recursively extrapolated from the inertial measurement data based on the acquisition time period by the state prediction unit of the state space model to obtain the predicted state and prediction error. The observation unit performs a weighted correction on the carrier phase observation value according to the dynamic observation weight and the phase correction amount to obtain a weighted correction observation value. The weighted corrected observation is compared with the predicted state by a tightly coupled fusion unit, and the obtained state comparison value is compared with the prediction error to obtain the state correction amount. The state correction amount and the predicted state are fused and updated by a tightly coupled fusion unit to obtain the fused data.

[0008] Furthermore, the state prediction unit using the state-space model recursively extrapolates the motion state of the inertial measurement data based on the acquisition time period to obtain the predicted state and prediction error, including: The state prediction unit extracts the angular velocity sequence and acceleration sequence within the acquisition time period from the inertial measurement data; Using the inertial measurement data corresponding to the start time of the acquisition time period as the recursive starting point, the position increment, velocity increment and attitude increment at each moment are obtained by sequentially calculating the time increment based on the angular velocity sequence and the acceleration sequence. Based on the position increment, the velocity increment, and the attitude increment, the state of the inertial measurement data is predicted to obtain the predicted state; The prediction error is obtained by calculating the deviation propagation of the predicted state based on the preset state deviation rules and the acquisition time period.

[0009] Further, the weighted corrected observation is compared with the predicted state by a tightly coupled fusion unit, and the obtained state comparison value is compared with the prediction error to obtain the state correction amount, including: The phase difference between the corrected carrier phase value in the weighted corrected observation and the predicted carrier phase value in the predicted state is calculated by the tightly coupled fusion unit to obtain the phase residual sequence. The pseudorange difference between the corrected pseudorange value in the weighted corrected observation and the predicted pseudorange value in the predicted state is calculated to obtain the pseudorange residual sequence. The phase residual sequence and the pseudorange residual sequence are integrated to obtain the observation residual vector; The state correction gain value is obtained by linearly transforming the observation residual vector with the state transition matrix and the observation matrix in the prediction error. The state correction amount is obtained by applying correction constraints to the state correction gain value based on the observed residual vector.

[0010] Further, the step of applying directional constraints and carrier phase fixation to the fused data based on the environmental data to obtain the positioning result includes: The current scene features and motion direction features are extracted from the environmental data, and the position increment sequence and phase floating-point solution are extracted from the fused data; Based on the current scene features, the corresponding scene constraint rules are selected from the preset constraint rule table. Based on the scene constraint rules and combined with the motion direction features, the position increment sequence is corrected by the direction component to obtain the constrained position increment sequence. The phase floating-point solution is calculated successively with a predefined set of integer candidates to obtain candidate integer residual values, and it is detected whether there are candidate integer residual values ​​that are less than a preset residual threshold. If there exists a candidate integer residual value that is less than the preset residual threshold, then the corresponding candidate integer is set as the fixed solution; otherwise, the phase floating-point solution is directly set as the fixed solution. The positioning result is obtained by iteratively calculating the constraint position increment sequence and the fixed solution.

[0011] The present invention also provides an RTK-based intelligent outdoor positioning device, applied to any of the above-described RTK-based intelligent outdoor positioning methods, comprising: The acquisition module is used to acquire carrier phase observations, inertial measurement data, and environmental data through a positioning terminal. The analysis module is used to perform anomaly diagnosis and correlation analysis on the carrier phase observations based on the environmental data, and generate dynamic observation weights and phase correction values; The association module is used to perform tightly coupled fusion calculation on the inertial measurement data according to the dynamic observation weight and the phase correction amount to obtain fused data; The processing module is used to perform directional constraints and carrier phase fixation on the fused data based on the environmental data to obtain the positioning result.

[0012] The present invention also provides an RTK-based intelligent outdoor positioning system, comprising: Memory, used to store programs; A processor is configured to execute the program to implement the various steps of the RTK-based intelligent outdoor positioning method described in any of the preceding claims.

[0013] The present invention also provides a storage medium storing computer instructions for causing a computer to perform any of the methods described above.

[0014] The present invention provides an RTK-based intelligent outdoor positioning method, device, system, and storage medium, which has the following beneficial effects: By introducing environmental data to perform anomaly diagnosis and correlation analysis on carrier phase observations, and generating dynamic observation weights and phase corrections, observation errors caused by complex environmental factors such as multipath effects and signal obstruction can be effectively identified and suppressed, thus significantly improving the accuracy and reliability of positioning in harsh outdoor scenarios. By tightly coupling and fusing weighted and corrected carrier phase observations with inertial measurement data in a state-space model, deep complementarity and synergy of multi-source sensor data are achieved, enhancing autonomous navigation and state inference capabilities during brief satellite signal interruptions or poor geometric configurations, ensuring the continuity of positioning. Using environmental data to impose directional constraints on the fusion results provides additional geometric or physical constraints for fixing carrier phase ambiguities, effectively reducing the ambiguity search space, improving the fixation success rate and the reliability of the fixed solution, thereby ensuring stable output of positioning results in various complex environments. Attached Figure Description

[0015] Figure 1 This is a flowchart of an RTK-based intelligent outdoor positioning method provided for this invention; Figure 2 This is a structural diagram of an RTK-based intelligent outdoor positioning device provided for this invention; Figure 3 This is a structural diagram of an RTK-based intelligent outdoor positioning system provided for this invention.

[0016] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0019] Reference Figure 1 As shown, the present invention provides an intelligent outdoor positioning method based on RTK, comprising: Step S1: Obtain carrier phase observations, inertial measurement data, and environmental data through the positioning terminal; Specifically, the GNSS receiving module of the positioning terminal continuously receives radio frequency signals from multiple navigation systems. After down-conversion, carrier stripping, and tracking loop processing, it demodulates the raw carrier phase signal measured in weeks and outputs this signal along with the corresponding satellite number, signal strength, and timestamp to form a carrier phase observation sequence. The inertial measurement unit built into the positioning terminal operates at a sampling frequency much higher than the GNSS update rate (100Hz or higher), measuring the angular velocity changes and linear acceleration of the carrier in real time along three orthogonal axes, and outputting angular velocity and acceleration readings. These readings are then calibrated and integrated into an inertial measurement data packet. The environmental perception part is accomplished by a group of miniature sensors on the positioning terminal, mainly collecting current atmospheric environmental parameters, such as obtaining absolute air pressure values ​​through a barometer and obtaining chip operating temperature through a temperature sensor. These readings together constitute environmental data.

[0020] Step S2: Based on the environmental data, perform anomaly diagnosis and correlation analysis on the carrier phase observation values ​​to generate dynamic observation weights and phase correction values; Specifically, characteristic parameters closely related to the current positioning scenario and signal propagation conditions are extracted from environmental data. For example, short-term trends in air pressure can be used to indirectly infer whether the terminal is in motion or in a closed environment, and temperature data can be used to assist in assessing sensor noise levels. These environmental characteristic parameters, combined with the physical properties of the carrier phase observations themselves (such as instantaneous rate of change and epoch continuity), constitute the basis for diagnosis. Each carrier phase observation is compared with a phase quality threshold; observations exceeding the threshold are marked as observations to be verified. Correlation analysis is initiated to check whether these observations to be verified are associated with specific environmental interference conditions (such as a sudden drop in signal strength when air pressure is stable, suggesting entry into an obstructed area). When an observation to be verified is confirmed to conform to a preset interference pattern, it is ultimately determined to be an abnormal observation. Based on the diagnostic results, confidence calculation is performed for each observation, outputting a dynamic observation weight between 0 and 1. The weight value directly reflects the reliability of the observation in the current environment. For observations that are judged to be abnormal but have a specific error pattern, a phase compensation value, namely the phase correction amount, is calculated based on the degree to which they deviate from the normal trend and in combination with the current environmental characteristics.

[0021] Step S3: Perform tight-coupled fusion calculation on the inertial measurement data according to the dynamic observation weight and the phase correction amount to obtain fused data; Specifically, using a state-space model as a framework, the model defines the system's state vector by combining the position, velocity, and attitude of the positioning terminal with the error parameters of the inertial sensors. Based on the previous optimal state estimate and the inertial measurement data of the current acquisition period, the predicted value of the system state at the current epoch and the corresponding prediction error covariance are obtained by recursively extrapolating the motion state of the carrier. Each original observation value is algebraically superimposed with the corresponding phase correction, and different confidence levels are assigned to the superimposed observation values ​​according to their dynamic observation weights. This information is directly used to configure the corresponding elements of the observation noise matrix in the state-space model. The weighted and corrected carrier phase observation value itself (rather than its calculated position) is used as the observation and compared with the corresponding theoretical observation value derived from the state prediction value to generate the observation residual. Based on the prediction error covariance and the configured observation noise matrix, the Kalman gain is calculated, and this gain is used to optimally weight the observation residual, thereby correcting and updating the state prediction value. This process is completed in a recursive filtering loop, and the final output is a fused data that integrates satellite observation geometric information and inertial motion dynamics information.

[0022] Step S4: Based on the environmental data, perform directional constraints and carrier phase fixation on the fused data to obtain the positioning result.

[0023] Specifically, information that can be used to enhance geometric constraints is extracted from environmental data, such as absolute altitude information provided by barometers. The altitude component in the fused data is compared with the absolute altitude reference value converted from barometric pressure. If the difference exceeds a vertical tolerance threshold set based on sensor accuracy, barometric altitude is used as a strong constraint to directly replace or forcefully pull together the altitude component in the fused solution, thereby effectively suppressing vertical drift caused by poor satellite geometry. Using the position information optimized by directional constraints as initial values, double-difference carrier phase observations between all common-view satellite pairs are calculated. Within the integer search space centered on this initial value, all integer ambiguity candidate combinations are traversed. For each candidate combination, its corresponding baseline vector and theoretical double-difference observation are calculated, and the residuals are obtained by comparing them with the actual observations. Among all candidate combinations, the one with the smallest sum of squared residuals is selected as the optimal candidate. The ratio of the residual of the optimal candidate to the residual of the second-best candidate is calculated. When this ratio is greater than a preset fixed threshold, the ambiguity is determined to be successfully fixed. By substituting the fixed integer ambiguity value into the double-difference observation equation, the calculated three-dimensional relative coordinates are combined with the base station coordinates to obtain the positioning result.

[0024] This invention provides an RTK-based intelligent outdoor positioning method. By introducing environmental data to perform anomaly diagnosis and correlation analysis on carrier phase observations, and generating dynamic observation weights and phase corrections, it can effectively identify and suppress observation errors caused by complex environmental factors such as multipath effects and signal obstruction, thereby significantly improving the accuracy and reliability of positioning in harsh outdoor scenarios. By tightly coupling and fusing the weighted and corrected carrier phase observations with inertial measurement data in a state-space model, deep complementarity and synergy of multi-source sensor data are achieved, enhancing autonomous navigation and state inference capabilities during brief satellite signal interruptions or poor geometric configurations, ensuring the continuity of positioning. Using environmental data to impose directional constraints on the fusion results provides additional geometric or physical constraints for fixing carrier phase ambiguities, effectively reducing the ambiguity search space, improving the fixation success rate and the reliability of the fixed solution, thus ensuring stable output of positioning results in various complex environments.

[0025] In one embodiment, acquiring carrier phase observations, inertial measurement data, and environmental data via a positioning terminal includes: The original carrier phase signal is obtained and extracted through the GNSS receiving module of the positioning terminal, and the original carrier phase signal is converted into the carrier phase observation value; The inertial measurement module of the positioning terminal acquires angular velocity and acceleration readings along three orthogonal axes, and integrates the angular velocity and acceleration readings to obtain the inertial measurement data. Specifically, the inertial measurement module completes self-testing and initialization upon terminal startup. Its internal three-axis gyroscope and three-axis accelerometer sample at a constant high frequency (e.g., 100 Hz or 200 Hz). During each sampling, the gyroscope senses the carrier's angular motion and outputs an analog voltage signal proportional to the instantaneous angular velocity. This signal is converted into a raw digital reading via an analog-to-digital converter. Similarly, the accelerometer outputs a raw digital reading reflecting specific force (including kinetic acceleration and gravitational acceleration). A pre-stored temperature compensation coefficient is applied, and based on real-time feedback from the module's built-in temperature sensor, the drift caused by temperature changes in the readings is corrected. A scaling factor matrix and a zero-bias matrix are applied to calibrate the readings to eliminate inconsistencies in sensitivity across axes and static biases. The corrected angular velocity and acceleration readings are assigned precise timestamps from the module's internal clock or synchronized with the GNSS receiver module, and then constructed into a data frame. This data frame sequentially includes: a timestamp, corrected X / Y / Z three-axis angular velocity values, corrected X / Y / Z three-axis acceleration values, and data validity and quality indicators. This data frame is continuously output at a high rate and with low latency, forming inertial measurement data.

[0026] The environmental data is acquired through the environmental sensors of the positioning terminal.

[0027] The method provided in this embodiment ensures that carrier phase observations, inertial measurement data, and environmental data have a unified time reference through the collaborative work and synchronous acquisition of the GNSS receiving module, inertial measurement module, and environmental sensors within the positioning terminal. This provides a multi-source data foundation with strict spatiotemporal alignment for subsequent tightly coupled fusion, solving the problem of decreased fusion accuracy caused by asynchronous data in traditional methods. By performing real-time error correction, formatted encapsulation, and structured integration on the raw sensor signals, standard data units that can be directly used for high-precision calculations are generated, effectively improving the reliability and availability of the data source.

[0028] In one embodiment, the step of performing anomaly diagnosis and correlation analysis on the carrier phase observations based on the environmental data to generate dynamic observation weights and phase correction values ​​includes: Extract environmental feature parameters corresponding to the carrier phase observation values ​​from the environmental data; The carrier phase observation value is compared with a preset phase quality threshold. If the carrier phase observation value exceeds the phase quality threshold, the carrier phase observation value is marked as an observation value to be verified; otherwise, the carrier phase observation value is determined to be a normal observation value. If it is detected that both the observation to be verified and the environmental feature parameters meet the preset interference conditions, then the observation to be verified is determined to be an abnormal observation. Among them, the preset interference conditions clearly describe the correlation between specific types of signal anomaly patterns (such as multipath effect, short-term blockage, and ionospheric scintillation) and specific environmental characteristic parameter change patterns.

[0029] Specifically, for each observation to be verified and its associated environmental feature parameter vector, pattern matching is performed between the observation and each preset interference condition. The abnormal pattern of the observation and the pattern of the environmental features are checked to see if they are temporally synchronized and logically consistent. For example, the diagnostic logic might analyze: a verification observation exhibiting slow, periodic fluctuations, if its corresponding environmental feature parameters indicate that the current temperature is stable but the air pressure changes drastically (potentially meaning the terminal is moving between tall buildings, causing multiple signal reflections), then the abnormal pattern of this observation matches the dynamic multipath interference condition. A verification observation exhibiting a momentary drop in the signal-to-noise ratio, if its environmental features indicate that the air pressure data undergoes a small step change at the same moment (potentially suggesting the terminal is rapidly passing through a doorway or tunnel entrance), then this pattern matches the momentary signal obstruction interference condition.

[0030] An observation is considered an anomalous only when all its anomalous properties match the environmental and signal characteristics required by a given interference condition. Observations that do not match any interference condition are considered random noise or unmodeled errors and will be handled using a conservative strategy.

[0031] The dynamic observation weights are obtained by performing confidence calculations and weight settings on the observations to be verified based on the environmental characteristic parameters. Based on the preset interference conditions, the abnormal observation values ​​are deviated and the deviated data is combined with the environmental feature parameters to perform phase compensation, thereby obtaining the phase correction amount.

[0032] Specifically, the theoretical rate of change of carrier phase, calculated based on satellite ephemeris and carrier prediction status, is used as a reference model. The actual short-term trend of abnormal observations is compared with this model, and the average trend deviation and fluctuation characteristics of the phase values ​​are quantified through differencing or fitting methods to form deviation data. Phase compensation is then performed: based on the interference type corresponding to the abnormal observation and a pre-defined compensation model, the model coefficients are modulated in real time in conjunction with current environmental characteristic parameters, and a specific phase correction is calculated and output.

[0033] The method provided in this embodiment utilizes environmental characteristic parameters to jointly diagnose carrier phase observations. By combining physical scene information such as air pressure and temperature, it can identify signal anomalies, thereby improving the accuracy of judging complex interferences such as multipath effects and transient blockages, and avoiding misjudgments based on a single signal indicator. By introducing preset interference conditions to determine the correlation of the observations to be verified, it can effectively distinguish error patterns from different sources and assign appropriate initial confidence levels to different anomaly types, thus laying the foundation for subsequent differentiated processing.

[0034] In one embodiment, the step of performing tightly coupled fusion calculation on the inertial measurement data based on the dynamic observation weights and the phase correction amount to obtain fused data includes: The acquisition time period of the positioning terminal is obtained, and the motion state is recursively extrapolated from the inertial measurement data based on the acquisition time period by the state prediction unit of the state space model to obtain the predicted state and prediction error. The observation unit performs a weighted correction on the carrier phase observation value according to the dynamic observation weight and the phase correction amount to obtain a weighted correction observation value. Specifically, the observation unit converts the observation value into a specific observation noise variance value using a preset mapping function, based on the dynamic observation weights carried by the observation value. Higher weights result in smaller noise variance values, indicating a more reliable observation value, which will be given greater importance in subsequent fusion calculations; conversely, lower weights correspond to larger noise variances, and their influence will be suppressed. Finally, the observation unit outputs a structured weighted corrected observation data packet, which includes at least: the satellite number, the phase-corrected carrier phase value, and the observation noise variance value determined according to the dynamic observation weights.

[0035] The weighted corrected observation is compared with the predicted state by a tightly coupled fusion unit, and the obtained state comparison value is compared with the prediction error to obtain the state correction amount. The state correction amount and the predicted state are fused and updated by a tightly coupled fusion unit to obtain the fused data.

[0036] The method provided in this embodiment, through the state prediction unit of the state-space model, recursively extrapolates the motion state based on inertial measurement data, enabling the prediction of carrier attitude, velocity, and position within the GNSS observation interval, thus ensuring the continuity of the positioning process. By using the observation unit to differentially weight and directly correct the carrier phase observation values ​​according to dynamic observation weights and phase correction amounts, the results of front-end intelligent diagnosis can be transformed into specific input parameters for the fusion algorithm. Through a tightly coupled fusion unit, the weighted and corrected observation values ​​are optimally compared and corrected with the predicted state, achieving statistically optimal fusion of multi-source information based on real-time changes in prediction errors and observation uncertainties. The final fusion update operation outputs the optimal state estimate after observation information correction and its corresponding error covariance, resulting in more accurate and smoother fused positioning data, forming a continuously self-optimizing high-precision positioning closed loop.

[0037] In one embodiment, the state prediction unit using the state-space model performs motion state recursion on the inertial measurement data based on the acquisition time period to obtain the predicted state and prediction error, including: The state prediction unit extracts the angular velocity sequence and acceleration sequence within the acquisition time period from the inertial measurement data; Using the inertial measurement data corresponding to the start time of the acquisition time period as the recursive starting point, the position increment, velocity increment and attitude increment at each moment are obtained by sequentially calculating the time increment based on the angular velocity sequence and the acceleration sequence. Specifically, the fused data is read, which includes the optimally estimated carrier position, velocity, attitude, and sensor error state at the start of the acquisition period (i.e., the previous fusion epoch). These state variables are fully loaded into the recursive algorithm as the initial values ​​for all calculations. Incremental calculations are performed within a loop, traversing each pair of adjacent sampling points in the angular velocity and acceleration sequences. For each pair of adjacent sampling point intervals, the attitude increment is calculated: using the carrier's current attitude corresponding to the start point of this time interval, the angular velocity readings acquired during this period are integrated. The integration employs a quaternion-based conical compensation algorithm to handle the non-commutative error caused by the change in the angular velocity vector direction during high-speed carrier rotation, thereby accurately calculating the quaternion change in carrier attitude within this short time interval. The velocity and position increments are calculated: using the calculated new attitude, the accelerometer specific force readings acquired within the same time interval are rotated from the carrier coordinate system to the navigation coordinate system. The rotated specific force readings need to be subtracted from the local gravitational acceleration vector to obtain the motion acceleration. Integrate the acceleration of motion with respect to time to obtain the velocity increment within that interval; then integrate this velocity increment with respect to time to obtain the position increment within that interval.

[0038] Based on the position increment, the velocity increment, and the attitude increment, the state of the inertial measurement data is predicted to obtain the predicted state; Specifically, starting from the beginning of the acquisition period, iterations are performed chronologically. For the first inertial sampling interval, the attitude at the beginning is combined with the first attitude increment, and updated using the direction cosine matrix to obtain the predicted attitude at the end of the first interval. Using this predicted attitude, the first velocity increment is added to the velocity vector at the beginning to obtain the predicted velocity at that moment. The first position increment is added to the position coordinates at the beginning to obtain the predicted position at that moment. This predicted attitude, predicted velocity, and predicted position constitute the complete state vector for the first high-frequency moment. This state then becomes the new starting point for the recursion of the next sampling interval, and is combined with the next increment pair to calculate the state for the next moment. This iterative cycle continues until all inertial sampling intervals are processed and the predicted state is generated.

[0039] The prediction error is obtained by calculating the deviation propagation of the predicted state based on the preset state deviation rules and the acquisition time period.

[0040] The method provided in this embodiment accurately extracts the angular velocity and acceleration sequences within the acquisition period from the time-synchronized inertial data stream through a state prediction unit. This ensures that the recursive calculation is based on high-frequency, continuous data strictly aligned with the GNSS epoch, thus providing reliable input for subsequent incremental calculations. By sequentially accumulating and reconstructing all time increments and interpolating at the GNSS epoch, the complete predicted attitude, velocity, and position of the carrier at the target time can be obtained. By calculating the deviation propagation of the predicted state according to a preset error propagation model, the state uncertainty caused by sensor noise and initial errors during the inertial recursion process can be quantitatively assessed.

[0041] In one embodiment, the weighted corrected observation is compared with the predicted state using a tightly coupled fusion unit, and the obtained state comparison value is compared with the prediction error to obtain a state correction amount, including: The phase difference between the corrected carrier phase value in the weighted corrected observation and the predicted carrier phase value in the predicted state is calculated by the tightly coupled fusion unit to obtain the phase residual sequence. Specifically, the tightly coupled fusion unit associates each weighted corrected observation with the current epoch and its corresponding signal source based on the timestamp and signal source identifier. For each successfully matched signal source, the calculation process for the predicted observation is initiated. Using the position coordinates in the predicted state, combined with the precise spatial coordinates of the signal source at the signal transmission and reception times obtained from external data, the geometric distance between the two is calculated. This geometric distance is converted to the corresponding week number or metric unit according to the observation type (phase or pseudorange), and a comprehensive correction term consisting of clock error, atmospheric delay, etc., is added to finally generate the predicted observation for that signal source. Difference calculation is then performed: the corrected observation is read from the corresponding weighted corrected observation and algebraically subtracted from it from the calculated predicted observation. All available signal sources are processed, and the above calculation and subtraction operations are repeated. Each observation difference is stored in an array according to a predetermined index, and this array constitutes the observation residual sequence.

[0042] The pseudorange difference between the corrected pseudorange value in the weighted corrected observation and the predicted pseudorange value in the predicted state is calculated to obtain the pseudorange residual sequence. Specifically, the tightly coupled fusion unit reads the prediction error covariance matrix, which quantifies the uncertainty of the predicted state in various dimensions. Simultaneously, based on the current predicted state and the spatial geometric distribution of all signal sources, the observation matrix is ​​calculated in real time. This matrix is ​​constructed based on geometric relationships such as the direction cosine vector between the signal sources and the predicted positions, mapping changes in the state space (position, velocity, etc.) to the rate of change in the observation space (distance or phase change). The prediction error covariance matrix is ​​multiplied by the observation matrix and its transpose to obtain an intermediate matrix. The observation noise matrix is ​​reconstructed from the information carried by the weighted correction observations. This is a diagonal matrix whose diagonal elements are mapped by the dynamic weights of the observations from each signal source; lower weights result in larger noise variance. The intermediate matrix is ​​added to this observation noise matrix, and the result is inverted. The prediction error covariance matrix is ​​multiplied by the transpose of the observation matrix, and then multiplied by the result of the previous inversion step to finally output the state correction gain value.

[0043] The phase residual sequence and the pseudorange residual sequence are integrated to obtain the observation residual vector; The state correction gain value is obtained by linearly transforming the observation residual vector with the state transition matrix and the observation matrix in the prediction error. The state correction amount is obtained by applying correction constraints to the state correction gain value based on the observed residual vector.

[0044] The method provided in this embodiment calculates the difference between the weighted corrected observations and the predicted state through a tightly coupled fusion unit. This allows for a direct and accurate comparison between the actual observation data, after intelligent diagnosis and correction, and the inertial prediction model, generating an observation residual sequence that reflects subtle deviations between the two. This provides a high-precision input benchmark for state correction. By performing a linear transformation based on optimal estimation theory on the observation residual sequence, prediction error, and observation matrix, the optimal state correction gain value can be dynamically calculated according to the uncertainty of the predicted state and the real-time reliability of each observation, ensuring that the fusion process is statistically the most reasonable. Further correction constraints on the state correction gain value based on the observation residual sequence prevent overcorrection in cases of sudden increases in observation noise or temporary model mismatch, thus ensuring the robustness and reliability of the state correction amount. Ultimately, this ensures that the tightly coupled fusion result remains stable and accurate even in complex environments.

[0045] In one embodiment, the step of applying directional constraints and carrier phase fixation to the fused data based on the environmental data to obtain a positioning result includes: The current scene features and motion direction features are extracted from the environmental data, and the position increment sequence and phase floating-point solution are extracted from the fused data; The current scene feature is a quantitative identifier obtained by performing specific analysis on the environmental data, used to describe the type of environment in which the carrier is located, such as category labels and their corresponding confidence levels for static stable scenes, dynamic open scenes, or bridge and tunnel obstruction scenes.

[0046] Based on the current scene features, the corresponding scene constraint rules are selected from the preset constraint rule table. Based on the scene constraint rules and combined with the motion direction features, the position increment sequence is corrected by the direction component to obtain the constrained position increment sequence. Specifically, the current scene features are used as query keys to search a pre-defined constraint rule table. This rule table is stored in key-value pairs, where the key is the scene feature identifier and the value is the corresponding set of constraint parameters, such as the horizontal constraint strength coefficient, the vertical constraint strength coefficient, and the variance threshold for constraint effectiveness. Upon successful matching, the corresponding scene constraint rule parameters are loaded. Directional features are introduced to refine the constraints. The dominant motion direction vector provided in the motion direction features is analyzed and used as a reference axis to establish the constraint coordinate system. For example, in a straight-line motion scenario within a tunnel, the constraint rules might require that position increments along the tunnel direction be relatively large, while changes perpendicular to the tunnel direction should be strictly limited. The correction operation is applied to each increment vector in the position increment sequence. For each increment, it is projected onto the constraint coordinate system defined by the scene rules and motion direction (e.g., decomposed into three components: along the motion direction, perpendicular to the motion direction, and perpendicular to the ground direction).

[0047] Based on the predefined constraint strength coefficients (ranging from 0 to 1, where 0 represents full constraint and 1 represents no constraint) for each directional component in the rules, the increment value in that direction is scaled. The scaled components are then back-projected back onto the original East-North-Sky coordinate system to obtain the corrected individual position increments. This process is repeated for all increments in the sequence, ultimately generating a completely new sequence of constrained position increments.

[0048] The phase floating-point solution is calculated successively with a predefined set of integer candidates to obtain candidate integer residual values, and it is detected whether there are candidate integer residual values ​​that are less than a preset residual threshold. If there exists a candidate integer residual value that is less than the preset residual threshold, then the corresponding candidate integer is set as the fixed solution; otherwise, the phase floating-point solution is directly set as the fixed solution. The positioning result is obtained by iteratively calculating the constraint position increment sequence and the fixed solution.

[0049] The method provided in this embodiment extracts scene and motion direction features, as well as position increments and phase floating-point solutions from environmental data and fused data. This deeply integrates environmental perception with high-precision fusion, thereby enhancing the adaptability of the final positioning process to complex scenes and the reliability of the processing basis. By dynamically invoking constraint rules based on scene features and physically correcting the position increment sequence in conjunction with the motion direction, prior environmental and motion knowledge can be transformed into strong constraints on short-term relative trajectories, effectively suppressing inertial accumulation errors and position drift caused by signal obstruction or reflection in specific scenes (such as bridges, tunnels, and forests). By performing successive residual calculations and threshold detection on the phase floating-point solution and integer candidate set, and intelligently selecting fixed integer solutions or retaining floating-point solutions based on the detection results, rapid and reliable ambiguity fixation can be achieved when the observation quality is good.

[0050] Reference Figure 2 As shown, the present invention also provides an RTK-based intelligent outdoor positioning device, applied to any of the above-described RTK-based intelligent outdoor positioning methods, comprising: The acquisition module is used to acquire carrier phase observations, inertial measurement data, and environmental data through a positioning terminal. The analysis module is used to perform anomaly diagnosis and correlation analysis on the carrier phase observations based on the environmental data, and generate dynamic observation weights and phase correction values; The association module is used to perform tightly coupled fusion calculation on the inertial measurement data according to the dynamic observation weight and the phase correction amount to obtain fused data; The processing module is used to perform directional constraints and carrier phase fixation on the fused data based on the environmental data to obtain the positioning result.

[0051] Reference Figure 3 As shown, the present invention also provides an RTK-based intelligent outdoor positioning system, comprising: Memory, used to store programs; A processor is configured to execute the program to implement the various steps of the RTK-based intelligent outdoor positioning method described in any of the preceding claims.

[0052] In this embodiment, the processor and memory can be connected via a bus or other means. The memory may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive. The processor may be a general-purpose processor, such as a central processing unit, digital signal processor, application-specific integrated circuit, or one or more integrated circuits configured to implement embodiments of the present invention.

[0053] The present invention also provides a storage medium storing computer instructions for causing a computer to perform any of the methods described above.

[0054] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the system and each module described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0055] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. An intelligent outdoor positioning method based on RTK, characterized in that, include: The positioning terminal acquires carrier phase observations, inertial measurement data, and environmental data. Based on the environmental data, anomaly diagnosis and correlation analysis are performed on the carrier phase observation values ​​to generate dynamic observation weights and phase correction values. The inertial measurement data is tightly coupled and fused based on the dynamic observation weights and the phase correction amount to obtain fused data; Based on the environmental data, the fused data is subjected to directional constraints and carrier phase fixation to obtain the positioning result.

2. The RTK-based intelligent outdoor positioning method according to claim 1, characterized in that, The acquisition of carrier phase observations, inertial measurement data, and environmental data through the positioning terminal includes: The original carrier phase signal is obtained and extracted through the GNSS receiving module of the positioning terminal, and the original carrier phase signal is converted into the carrier phase observation value; The inertial measurement module of the positioning terminal acquires angular velocity and acceleration readings along three orthogonal axes, and integrates the angular velocity and acceleration readings to obtain the inertial measurement data. The environmental data is acquired through the environmental sensors of the positioning terminal.

3. The RTK-based intelligent outdoor positioning method according to claim 1, characterized in that, The step of performing anomaly diagnosis and correlation analysis on the carrier phase observations based on the environmental data, and generating dynamic observation weights and phase correction values, includes: Extract environmental feature parameters corresponding to the carrier phase observation values ​​from the environmental data; The carrier phase observation value is compared with a preset phase quality threshold. If the carrier phase observation value exceeds the phase quality threshold, the carrier phase observation value is marked as an observation value to be verified; otherwise, the carrier phase observation value is determined to be a normal observation value. If it is detected that both the observation to be verified and the environmental feature parameters meet the preset interference conditions, then the observation to be verified is determined to be an abnormal observation. The dynamic observation weights are obtained by performing confidence calculations and weight settings on the observations to be verified based on the environmental characteristic parameters. Based on the preset interference conditions, the abnormal observation values ​​are deviated and the deviated data is combined with the environmental feature parameters to perform phase compensation, thereby obtaining the phase correction amount.

4. The RTK-based intelligent outdoor positioning method according to claim 1, characterized in that, The step of performing tightly coupled fusion calculation on the inertial measurement data based on the dynamic observation weights and the phase correction amount to obtain fused data includes: The acquisition time period of the positioning terminal is obtained, and the motion state is recursively extrapolated from the inertial measurement data based on the acquisition time period by the state prediction unit of the state space model to obtain the predicted state and prediction error. The observation unit performs a weighted correction on the carrier phase observation value according to the dynamic observation weight and the phase correction amount to obtain a weighted correction observation value. The weighted corrected observation is compared with the predicted state by a tightly coupled fusion unit, and the obtained state comparison value is compared with the prediction error to obtain the state correction amount. The state correction amount and the predicted state are fused and updated by a tightly coupled fusion unit to obtain the fused data.

5. The RTK-based intelligent outdoor positioning method according to claim 4, characterized in that, The state prediction unit using the state-space model performs motion state recursion on the inertial measurement data based on the acquisition time period to obtain the predicted state and prediction error, including: The state prediction unit extracts the angular velocity sequence and acceleration sequence within the acquisition time period from the inertial measurement data; Using the inertial measurement data corresponding to the start time of the acquisition time period as the recursive starting point, the position increment, velocity increment and attitude increment at each moment are obtained by sequentially calculating the time increment based on the angular velocity sequence and the acceleration sequence. Based on the position increment, the velocity increment, and the attitude increment, the state of the inertial measurement data is predicted to obtain the predicted state; The prediction error is obtained by calculating the deviation propagation of the predicted state based on the preset state deviation rules and the acquisition time period.

6. The RTK-based intelligent outdoor positioning method according to claim 4, characterized in that, The weighted corrected observation is compared with the predicted state using a tightly coupled fusion unit. The resulting state comparison value is then compared with the prediction error to obtain a state correction amount, including: The phase difference between the corrected carrier phase value in the weighted corrected observation and the predicted carrier phase value in the predicted state is calculated by the tightly coupled fusion unit to obtain the phase residual sequence. The pseudorange difference between the corrected pseudorange value in the weighted corrected observation and the predicted pseudorange value in the predicted state is calculated to obtain the pseudorange residual sequence. The phase residual sequence and the pseudorange residual sequence are integrated to obtain the observation residual vector; The state correction gain value is obtained by linearly transforming the observation residual vector with the state transition matrix and the observation matrix in the prediction error. The state correction amount is obtained by applying correction constraints to the state correction gain value based on the observed residual vector.

7. The RTK-based intelligent outdoor positioning method according to claim 1, characterized in that, The step of performing directional constraints and carrier phase fixation on the fused data based on the environmental data to obtain the positioning result includes: The current scene features and motion direction features are extracted from the environmental data, and the position increment sequence and phase floating-point solution are extracted from the fused data; Based on the current scene features, the corresponding scene constraint rules are selected from the preset constraint rule table. Based on the scene constraint rules and combined with the motion direction features, the position increment sequence is corrected by the direction component to obtain the constrained position increment sequence. The phase floating-point solution is calculated successively with a predefined set of integer candidates to obtain candidate integer residual values, and it is detected whether there are candidate integer residual values ​​that are less than a preset residual threshold. If there exists a candidate integer residual value that is less than the preset residual threshold, then the corresponding candidate integer is set as the fixed solution; otherwise, the phase floating-point solution is directly set as the fixed solution. The positioning result is obtained by iteratively calculating the constraint position increment sequence and the fixed solution.

8. An intelligent outdoor positioning device based on RTK, characterized in that, The RTK-based intelligent outdoor positioning method applied to any one of claims 1-7 includes: The acquisition module is used to acquire carrier phase observations, inertial measurement data, and environmental data through a positioning terminal. The analysis module is used to perform anomaly diagnosis and correlation analysis on the carrier phase observations based on the environmental data, and generate dynamic observation weights and phase correction values; The association module is used to perform tightly coupled fusion calculation on the inertial measurement data according to the dynamic observation weight and the phase correction amount to obtain fused data; The processing module is used to perform directional constraints and carrier phase fixation on the fused data based on the environmental data to obtain the positioning result.

9. An RTK-based intelligent outdoor positioning system, characterized in that, include: Memory, used to store programs; A processor is configured to execute the program to implement the steps of the RTK-based intelligent outdoor positioning method as described in any one of claims 1-7.

10. A storage medium, characterized in that, The computer contains computer instructions for causing the computer to perform the method according to any one of claims 1 to 7.

Images