A real-time dynamic high-precision positioning method based on tight coupling of Beidou and UWB

Abstract

The application relates to the cross-technology field of satellite navigation and wireless positioning technology, in particular to a real-time dynamic high-precision positioning method based on tight coupling of Beidou and UWB, solves the problem that error coupling in Beidou and UWB observation data is not fully solved in the prior art, and in particular under a high-speed dynamic scene, the mutual influence of Beidou receiver clock error, UWB base station clock error and multipath error leads to continuous accumulation of errors, and the positioning precision is significantly reduced. A real-time dynamic high-precision positioning method based on tight coupling of Beidou and UWB comprises a data preprocessing and space-time alignment module, an error joint estimation model, a rapid ambiguity fixing algorithm, a geometric constraint enhanced positioning engine module and a dynamic coordinate conversion and fusion output module. The method disclosed in the application adopts a highly integrated multi-model Kalman filter architecture to realize synchronous and accurate compensation of Beidou receiver clock error, UWB base station time delay and multipath effect.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of satellite navigation and wireless positioning, and in particular to a real-time dynamic high-precision positioning method based on the tight coupling of BeiDou and UWB. Background Technology

[0002] In today's positioning technology field, the tightly coupled positioning method of BeiDou Navigation Satellite System (BDS) and Ultra-Wideband (UWB) technology occupies an important position due to its unique advantages. As a key component in achieving high-precision, real-time dynamic positioning, the tightly coupled technology of BeiDou and UWB aims to provide more accurate and reliable positioning services by integrating the advantages of two different positioning technologies. Especially in dynamic scenarios requiring high-precision positioning, such as autonomous driving, intelligent logistics, and indoor navigation, this tightly coupled technology has broad application prospects.

[0003] However, existing tightly coupled BeiDou and UWB positioning methods have gradually revealed a series of obvious limitations and technical problems. Specifically, existing technologies have not effectively solved the coupling problem of clock skew and multipath error in BeiDou and UWB observation data. In high-speed dynamic scenarios, the accumulation of clock skew and multipath error leads to a significant decrease in positioning accuracy, making it difficult to meet the requirements of high-precision positioning.

[0004] Furthermore, the traditional LAMBDA algorithm does not fully utilize the geometric constraints of UWB ranging when ambiguity is fixed. This results in a large ambiguity search space, high computational complexity, excessively long fixing time, and insufficient success rate. In applications requiring real-time positioning, this lengthy fixing process and low success rate severely restrict the real-time performance of positioning technology.

[0005] When the number of BeiDou satellites is insufficient due to signal blockage, existing tightly coupled methods cannot effectively compensate for geometric constraints using UWB ranging. This leads to abrupt changes in positioning results and insufficient system robustness. In complex indoor and outdoor environments, these abrupt changes and insufficient robustness can severely affect the reliability and stability of positioning technology.

[0006] Meanwhile, the method of relying on signal strength thresholds for indoor-outdoor switching also has shortcomings. Since signal strength is affected by various factors, relying solely on signal strength thresholds can easily lead to misjudgments and inaccurate positioning. Furthermore, the lack of a real-time dynamic transformation model between the BeiDou geocentric coordinate system (ECEF) and the UWB local coordinate system results in an uneven positioning trajectory in transition areas, with errors reaching the decimeter level. This unevenness and significant error severely affect the continuity and accuracy of positioning technology. Therefore, to address these shortcomings, we urgently need an innovative, tightly coupled real-time dynamic high-precision positioning method using BeiDou and UWB to solve these problems. Summary of the Invention

[0007] The purpose of this invention is to provide a real-time dynamic high-precision positioning method based on tight coupling of BeiDou and UWB, which solves the problem of error coupling in BeiDou and UWB observation data that is not fully addressed in the existing technology. Especially in high-speed dynamic scenarios, the mutual influence of BeiDou receiver clock error, UWB base station clock error and multipath error leads to continuous error accumulation and a significant decrease in positioning accuracy. At the same time, traditional ambiguity fixing methods do not effectively utilize the geometric constraints of UWB ranging, resulting in a large search space and long computation time, which makes it difficult to meet real-time requirements.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] A real-time dynamic high-precision positioning method based on tight coupling of BeiDou and UWB includes a data preprocessing and spatiotemporal alignment module, an error joint estimation model, a fast ambiguity fixing algorithm, a geometric constraint-enhanced positioning engine module, and a dynamic coordinate transformation and fusion output module. In this tightly coupled BeiDou and UWB positioning method, data preprocessing and spatiotemporal alignment are performed first. The timestamps of the original BeiDou data and UWB ranging data are aligned using hardware PPS signals, and linear interpolation is used to compensate for sub-millisecond deviations, ensuring consistency of time references for multi-source data. Simultaneously, BeiDou positioning results and UWB base station coordinates are collected in open areas. The initial coordinate transformation matrix is ​​calculated using the least squares method, with calibration accuracy controlled within 2cm. Outliers are removed based on the 3σ criterion to provide high-quality input for deep fusion. Next, an error joint estimation model is constructed, and an adaptive Kalman filter algorithm is used to implement the model. The system estimates and compensates for BeiDou clock errors, UWB clock errors, and multipath errors, significantly improving estimation accuracy and suppressing error coupling in dynamic environments. Regarding ambiguity fixing, the LAMBDA algorithm is improved by employing a block search strategy and a ratio check mechanism to quickly fix ambiguities, overcoming efficiency bottlenecks and meeting the real-time requirements of high-dynamic scenarios. Geometric constraints enhance the positioning engine by introducing the UWB ranging equation as a supplementary constraint when the number of satellites is low, jointly solving for position increments, and using singular value decomposition to handle rank-deficient equations. A residual feedback mechanism is designed to suppress the influence of multipath and non-line-of-sight errors, maintaining positioning continuity. Finally, dynamic coordinate transformation and fusion output dynamically allocate weights based on the number of satellites and UWB signal quality, smoothly transitioning between indoor and outdoor areas through weighted fusion results, controlling switching errors to the centimeter level, improving trajectory smoothness, and achieving seamless positioning across all scenarios.

[0010] Preferably, data preprocessing and spatiotemporal alignment are performed. The raw BeiDou data, including pseudorange ρ and carrier phase φ, and the UWB ranging data d, first need to be timestamped using hardware PPS signals to ensure temporal consistency between BeiDou and UWB data, providing a foundation for subsequent data fusion. Since the hardware clock may have sub-millisecond deviations, linear interpolation is used to compensate for these subtle time differences. In open areas, BeiDou positioning results and UWB base station coordinates X are collected. UWB Using this data, the initial coordinate transformation matrices R and T are calculated using the least squares method. These transformation matrices are used to transform the local coordinates of UWB to the ECEF coordinate system of BeiDou, thereby achieving spatial alignment of the two positioning data sets. The calibration accuracy is controlled to ≤2cm. Furthermore, to eliminate the influence of outlier data on subsequent calculations, sliding window statistics are performed on both BeiDou pseudorange and UWB ranging data, and outliers are removed based on the 3σ criterion. This effectively avoids interference from outliers in the positioning results, providing high-quality input data for deep fusion. The initial coordinate transformation matrices R and T calculated using the least squares method are as follows:

[0011]

[0012] The preferred method is a joint error estimation model, designed to accurately estimate and compensate for the BeiDou clock bias b. GNSS UWB clock difference b UWB and multipath error e MP The state equations were constructed as follows:

[0013] x k+1 =Fx k +w k

[0014] z k =Hx k +v k

[0015] Where, x = [b GNSS ,b UWB ,e MP ] T Let F be the state vector, and H be the state transition matrix, which describes the change of state over time. H is the observation matrix, which maps the state vector to the observation space. w and v are the process noise and measurement noise, respectively, reflecting the uncertainty and observation error of the system. The adaptive Kalman filter algorithm iteratively updates the Kalman filter gain matrix, estimates and compensates for clock bias and multipath error in real time. By continuously adjusting the filter parameters, the algorithm can adapt to the noise characteristics under different environments, improve the accuracy of estimation, and achieve nanosecond-level accuracy in BeiDou clock bias estimation. It significantly suppresses error coupling phenomena under dynamic environments and improves the long-term stability of the system.

[0016] The preferred ambiguity fast fixing algorithm uses the UWB ranging equation to provide positional geometric constraints, compressing the integer ambiguity search dimension from 6 dimensions to less than 3 dimensions, and improving the LAMBDA algorithm:

[0017]

[0018] Where Q is the ambiguity covariance matrix, reflecting the correlation between ambiguities; a * The floating-point fuzzy solution is the starting point for fuzzy search. It adopts a block search strategy and a ratio test mechanism to prioritize the selection of the optimal fuzzy combination. Through block search, the search space can be divided into multiple small blocks for separate searches, thereby improving search efficiency. At the same time, the ratio test mechanism is used to evaluate the reliability of different fuzzy combinations to ensure that the selected fuzzy combination is optimal, breaking through the bottleneck of efficiency of fixed fuzzy values.

[0019] Preferably, a geometrically constrained enhanced positioning engine is used when the number of satellites N sat When the value is ≥4, the position is calculated using the BeiDou observation equation:

[0020] ρ i =‖XX i ‖+cb GNSS +e MP,i

[0021] N sat When the value is 2 or 3, the UWB ranging equation is introduced as a supplementary constraint to jointly solve for the position increment:

[0022] d j =‖XX j ||

[0023] In the solution process, singular value decomposition is used to handle the rank-deficient equation, avoiding the problem of matrix inversion failure. At the same time, it has a feedback mechanism that dynamically adjusts the observation weights according to the observation residuals, suppressing the impact of multipath and non-line-of-sight errors on the positioning results. This mechanism can maintain the continuity of positioning in occluded scenarios.

[0024] Preferably, the dynamic coordinate transformation and fusion output module needs to estimate the transformation parameters of the two coordinate systems in real time to achieve seamless fusion of BeiDou and UWB positioning results. It dynamically allocates weights based on the number of satellites and the quality of UWB signals to ensure accurate positioning results in different environments. When the number of satellites decreases, the weight of UWB is gradually increased from 20% to 80% to make full use of UWB positioning information. In the transition area between indoor and outdoor environments, we achieve a smooth transition through weighted fusion results, avoiding trajectory jump problems caused by relying on fixed thresholds. The switching error is controlled within the centimeter level, and the trajectory smoothness is significantly improved.

[0025] The present invention has at least the following beneficial effects:

[0026] This invention addresses the challenge of multi-source error coupling in complex environments by employing a highly integrated multi-model Kalman filter architecture to achieve synchronous and accurate compensation for BeiDou receiver clock errors, UWB base station delays, and multipath effects. This architecture not only considers the mutual influence between error sources but also establishes a detailed error propagation model in the time-frequency domain by constructing an advanced joint error estimation module. This model dynamically decouples error terms, effectively avoiding positioning drift caused by error accumulation in traditional solutions. Particularly in challenging scenarios with strong reflections, such as urban canyons and tunnels, the introduction of a time-varying noise covariance adjustment mechanism significantly enhances the system's anti-interference capability. This mechanism adjusts filter parameters in real time according to environmental changes, ensuring stable centimeter-level positioning performance even in dynamic scenarios, greatly improving positioning accuracy and reliability.

[0027] This invention innovatively integrates ambiguity search algorithms, cleverly embedding the geometric constraints of UWB ranging. By establishing a dimensionality compression model, the parameter search space is reconstructed, significantly reducing the computational complexity of the integer least squares problem. This improvement makes the algorithm more efficient when processing large-scale data. Simultaneously, an improved ratio verification mechanism is incorporated, and a parallel processing strategy enables rapid fixing of ambiguity parameters. This not only meets the stringent requirements for positioning update rates in high-speed mobile scenarios but also ensures the real-time performance and accuracy of positioning. Furthermore, the hierarchical computing architecture designed in the algorithm processing unit, optimized for hardware compatibility, allows these complex algorithms to be deployed on resource-constrained embedded platforms, broadening the system's application scope.

[0028] The present invention also has at least the following beneficial effects:

[0029] To achieve seamless conversion between the BeiDou and UWB coordinate systems, a dynamic calibration system based on the extended Kalman filter framework was constructed. This invention utilizes online parameter identification technology to adjust the transformation parameters between the coordinate systems in real time, ensuring the fusion effect of the two positioning technologies. In the fusion module, an adaptive weighting strategy is employed, dynamically constructing a weighting function based on the satellite visibility index and the quality parameters of the UWB signal. This strategy overcomes the performance bottleneck of traditional fixed-threshold fusion methods under sudden environmental changes, enabling the system to respond more flexibly to various environmental variations. A specially designed signal quality assessment circuit, through multi-dimensional feature extraction technology, provides the system with accurate environmental state perception capabilities. This allows the system to maintain a continuous and smooth positioning trajectory in indoor-outdoor transition areas, improving the user's positioning experience. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is a schematic diagram of the overall process of the present invention. Detailed Implementation

[0032] 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.

[0033] Example 1

[0034] Reference Figure 1 The system includes a data preprocessing and spatiotemporal alignment module, a joint error estimation model, a fast ambiguity fixing algorithm, a geometric constraint-enhanced positioning engine module, and a dynamic coordinate transformation and fusion output module. In the tightly coupled BeiDou and UWB positioning method, data preprocessing and spatiotemporal alignment are performed first. The timestamps of the original BeiDou data and UWB ranging data are aligned using hardware PPS signals, and linear interpolation is used to compensate for sub-millisecond deviations, ensuring consistency of the time reference for multi-source data. Simultaneously, BeiDou positioning results and UWB base station coordinates are collected in open areas. The initial coordinate transformation matrix is ​​calculated using the least squares method, with calibration accuracy controlled within 2cm. Outliers are removed based on the 3σ criterion, providing high-quality input for deep fusion. Next, a joint error estimation model is constructed, and an adaptive Kalman filter algorithm is used to estimate and compensate for BeiDou clock errors and UWB clock bias in real time. The WB clock bias and multipath error are significantly improved to enhance estimation accuracy and suppress error coupling in dynamic environments. Regarding ambiguity fixing, the LAMBDA algorithm is improved by employing a block search strategy and a ratio check mechanism to quickly fix ambiguity, overcoming efficiency bottlenecks and meeting the real-time requirements of high-dynamic scenarios. Geometric constraints enhance the positioning engine by introducing the UWB ranging equation as a supplementary constraint when the number of satellites is low, jointly solving for position increments, and using singular value decomposition to handle rank-deficient equations. A residual feedback mechanism is designed to suppress the influence of multipath and non-line-of-sight errors, maintaining positioning continuity. Finally, dynamic coordinate transformation and fusion output dynamically allocate weights based on the number of satellites and UWB signal quality, smoothly transitioning between indoor and outdoor areas through weighted fusion results. Switching errors are controlled at the centimeter level, improving trajectory smoothness and achieving seamless positioning across all scenarios.

[0035] First, the data preprocessing and spatiotemporal alignment module ensures the consistency of the time base between the BeiDou raw data and the UWB ranging data, and compensates for sub-millisecond deviations through linear interpolation. Next, using data collected in open areas, the initial coordinate transformation matrix is ​​calculated using the least squares method, with calibration accuracy controlled within 2cm, and outliers are removed based on the 3σ criterion. Then, a joint error estimation model is constructed, and an adaptive Kalman filter algorithm is used to estimate and compensate for various errors in real time. An improved LAMBDA algorithm is then used to quickly fix ambiguities. The geometric constraint enhancement positioning engine introduces the UWB ranging equation as a supplementary constraint when the number of satellites is small. Finally, the dynamic coordinate transformation and fusion output module dynamically allocates weights based on the number of satellites and the quality of the UWB signal to achieve seamless positioning across all scenarios. This achieves the effects of improving positioning accuracy, suppressing error coupling in dynamic environments, meeting the real-time requirements of highly dynamic scenarios, maintaining positioning continuity, and realizing seamless positioning across all scenarios.

[0036] Example 2

[0037] Reference Figure 1 Data preprocessing and spatiotemporal alignment are performed. The raw BeiDou data, including pseudorange and carrier phase, is compared with the UWB ranging data. First, the raw BeiDou data includes pseudorange ρ and carrier phase φ, while the raw UWB data is the ranging value d. To ensure a unified time reference for the two types of data during fusion computation, a hardware PPS (Pulse Per Second) signal triggering mechanism is used to sample and align the timestamps of the BeiDou and UWB modules respectively. Considering the sub-millisecond clock skew between device hardware, a linear interpolation method is further used to compensate and correct the data from different sources. Let the BeiDou data timestamp at a certain moment be t. BDS UWB data timestamp is t UWB When Δt=∣t BDS -t UWB When |≤1ms, the compensation value d is estimated by linear interpolation based on two neighboring UWB sampling points. interp (t BDS This allows for the acquisition of UWB ranging results that are strictly aligned with BeiDou data.

[0038] To achieve spatial alignment of the two positioning data, coordinates (local coordinate system) of multiple UWB base stations were simultaneously acquired in open areas with GNSS signals. Corresponding BeiDou output coordinates (ECEF coordinate system) at the corresponding time Selecting n≥3 pairs of sample points, the following coordinate transformation model is established:

[0039]

[0040] Where R is the rotation matrix, satisfying the orthogonality constraint R T R = I, T is the translation vector ∈ iThe rotation matrix R and translation vector T are derived for the residual terms using the least squares method. The specific steps are as follows:

[0041] 1. Calculate the centroids of two sets of points:

[0042] 2. Decentralization at each point:

[0043] 3. Construct the covariance matrix:

[0044] 4. Perform singular value decomposition on H: H = UΣV T

[0045] 5. Calculate the rotation matrix: R = VU T

[0046] If det(R) < 0, then take the negative of the last column of V to ensure that the rotation matrix satisfies the right-hand rule constraint.

[0047] 6. Calculate the translation vector: The final coordinate transformation relationship is obtained as follows:

[0048] P BDS =R·P UWB +T

[0049] The method was validated on a real-world dataset with a calibration accuracy better than 2 cm, meeting the high-precision requirement for coordinate consistency in subsequent fusion positioning. To further improve the quality of the input data, a sliding window statistical method was introduced for both BeiDou pseudorange data and UWB ranging data. With a window size of w, the mean μ and standard deviation σ were calculated within each window, and outliers were removed based on the 3σ criterion. i -μ∣>3σ then x i Outliers are removed. The data preprocessing and spatiotemporal alignment module ensures time synchronization between BeiDou and UWB data through hardware alignment and interpolation compensation, achieves spatial registration through coordinate transformation matrix calculation based on the least squares method, and ensures the quality of fused input through an outlier removal mechanism, providing the prerequisites for high-precision and stable operation of the system.

[0050] Example 3

[0051] Reference Figure 1 To accurately estimate and compensate for BeiDou clock errors, UWB clock errors, and multipath errors, a joint error estimation model was constructed, which included a state equation:

[0052] x k =F k x k-1 +w k

[0053] z k =Hk x k +v k

[0054] Wherein, the state vector xk=[b GNSS ,b UWB ,e MP ] T This represents the BeiDou clock bias, UWB clock bias, and multipath error; F k H is the identity matrix (i.e., assuming the error state changes slowly over a short period of time). k For the unit observation matrix, w k ~N(0,Q) k ), v k ~N(0,R k The noise values ​​are system noise and observation noise, respectively, reflecting the uncertainty and observation error of the system. The adaptive Kalman filter algorithm iteratively updates the Kalman filter gain matrix, estimates and compensates for clock bias and multipath error in real time. By continuously adjusting the filter parameters, the algorithm can adapt to the noise characteristics under different environments, improve the accuracy of estimation, and achieve nanosecond-level accuracy in BeiDou clock bias estimation. It significantly suppresses error coupling in dynamic environments and improves the long-term stability of the system.

[0055] To achieve adaptive filtering capability, the filter parameter Q is dynamically adjusted. k R k The residual covariance fitting method is adopted, that is, the R-squared is adaptively updated based on the variance variation trend between the actual observed residuals and the predicted residuals. k :

[0056]

[0057] in The residual is the observation value, and a∈[0,1] is the smoothing coefficient. This mechanism enhances the filter's adaptability to different time-varying errors and improves clock error compensation accuracy.

[0058] First, a state equation is constructed to describe the change of state over time and to map the state vector to the observation space. Then, an adaptive Kalman filter algorithm is used to iteratively update the Kalman filter gain matrix, estimating and compensating for clock errors and multipath errors in real time. This achieves the effects of accurately estimating and compensating for various errors, adapting to noise characteristics under different environments, improving estimation accuracy, significantly suppressing error coupling phenomena under dynamic environments, and enhancing the long-term stability of the system.

[0059] Example 4

[0060] Reference Figure 1A fast ambiguity fixation algorithm, using UWB ranging equations to provide positional geometric constraints, compresses the integer ambiguity search dimension from 6 dimensions to less than 3 dimensions, and improves the LAMBDA algorithm:

[0061]

[0062] Where Q is the ambiguity covariance matrix, reflecting the correlation between ambiguities; a * The floating-point fuzzy solution is the starting point for fuzzy search. It adopts a block search strategy and a ratio test mechanism to prioritize the selection of the optimal fuzzy combination. Through block search, the search space can be divided into multiple small blocks for separate searches, thereby improving search efficiency. At the same time, the ratio test mechanism is used to evaluate the reliability of different fuzzy combinations to ensure that the selected fuzzy combination is optimal, breaking through the bottleneck of efficiency of fixed fuzzy values.

[0063] First, the UWB ranging equation provides positional geometric constraints, compressing the search dimension for integer ambiguity. Then, the LAMBDA algorithm is improved by employing a block search strategy and a ratio check mechanism to prioritize the selection of optimal ambiguity combinations. This achieves the effects of improving ambiguity search efficiency, evaluating the reliability of different ambiguity combinations, ensuring that the selected ambiguity combination is optimal, and overcoming the efficiency bottleneck of fixed ambiguity.

[0064] To address the low search efficiency of the original LAMBDA algorithm during the rapid fixation of ambiguity, a block search and ratio test mechanism is introduced. The block search strategy involves decomposing the search space of the ambiguity solution into principal components based on relevance, prioritizing the search of highly relevant dimensions to reduce the dimensionality of each search and decrease computational load. For example, the covariance matrix Q of the ambiguity vector a... a Perform Cholesky or LDL decomposition to extract principal axis directions. Ratio verification mechanism: For candidate integer fuzzy solutions... Define the discriminant ratio:

[0065]

[0066] When r < 0.25, the solution is considered unique and reliable, and the ambiguity is fixed; otherwise, it is judged as unsolvable. This mechanism effectively avoids the problem of misfixation and improves overall stability.

[0067] Example 5

[0068] Reference Figure 1 Geometric constraint-enhanced positioning engine, when the number of satellites N sat When the value is ≥4, the position is calculated using the BeiDou observation equation:

[0069] ρ i =‖XX i ‖+cbGNSS +e MP,i

[0070] N sat When the value is 2 or 3, the UWB ranging equation is introduced as a supplementary constraint to jointly solve for the position increment:

[0071] d j =‖XX j ||

[0072] In the solution process, singular value decomposition is used to handle the rank-deficient equation, avoiding the problem of matrix inversion failure. At the same time, it has a feedback mechanism that dynamically adjusts the observation weights according to the observation residuals, suppressing the impact of multipath and non-line-of-sight errors on the positioning results. This mechanism can maintain the continuity of positioning in occluded scenarios.

[0073] When the number of satellites N sat When N ≥ 4, the position is calculated using the BeiDou observation equation; when N sat When the number of satellites is 2 or 3, the UWB ranging equation is introduced as a supplementary constraint to jointly solve for the position increment. During the solution process, singular value decomposition is used to handle the rank-deficient equation, and a residual feedback mechanism is designed. This achieves the effects of maintaining positioning accuracy when the number of satellites is small, avoiding the problem of matrix inversion failure, dynamically adjusting observation weights based on observation residuals, suppressing the impact of multipath and non-line-of-sight errors on positioning results, and maintaining the continuity of positioning in occluded scenarios.

[0074] To address the positioning instability caused by insufficient satellite count, this embodiment introduces UWB constraints into the observation equations and designs a dynamic weight adjustment mechanism based on residual feedback. A joint observation matrix G = [G...] is constructed. BDS G UWB When the matrix rank is insufficient, SVD decomposition is used for stable solution to avoid failure of direct inversion. The dynamic update rules for the observation weight matrix W are as follows:

[0075] 1. Calculate the current residual vector

[0076] 2. For each observation, the residual e i Calculate the standardized residual r i =e i / σ i , where σ i This corresponds to the standard deviation of the observation error;

[0077] 3. Update the weights to γ is an adjustment parameter, typically ranging from [0.5, 2].

[0078] 4. Construct a weighted least squares system The final localization solution is obtained through iterative solutions.

[0079] By dynamically adjusting weights through a residual feedback mechanism, the interference of multipath and non-line-of-sight errors on the results is suppressed, and the positioning continuity is maintained in occluded scenarios.

[0080] Example 6

[0081] Reference Figure 1 This embodiment provides a dynamic coordinate transformation and fusion output module to achieve seamless fusion of BeiDou and UWB positioning results, ensuring positioning continuity and trajectory smoothness. It is particularly suitable for complex application scenarios such as indoor-outdoor transition areas and semi-open environments. The module comprehensively considers the number of BDS satellites (N). sat By dynamically adjusting the fusion weights based on UWB signal quality indicators, an adaptive and smooth transition of multi-source fusion positioning results can be achieved.

[0082] Specifically, to estimate the transformation parameters between the two coordinate systems in real time, the system first solves for the initial coordinate transformation matrix based on the least squares method, and then unifies the positioning results of BeiDou and UWB to the same coordinate system. Subsequently, based on environmental perception information, the system dynamically allocates the fusion weights of the BDS and UWB positioning results to ensure that the system can obtain stable, continuous, and accurate positioning results in different environments.

[0083] To determine UWB signal quality, the system extracts the following key indicators for comprehensive evaluation: Received Signal Strength Indicator (RSSI); First Path Index (FBI); and Received Energy (Rx Energy). When the RSSI is greater than -80 dBm, the FBI exceeds a set threshold, and the received signal is stable with no packet loss, the UWB signal quality is considered good. Furthermore, a sliding window residual statistical model is constructed to analyze the mean and standard deviation of continuous UWB ranging residuals. If the standard deviation σ is less than 0.1 m, the ranging fluctuation is considered small and the error stable, suitable for weighted enhancement.

[0084] The system employs the following dynamic weight allocation strategy based on the number of available BDS satellites (Nsat) and the UWB signal quality:

[0085] If N sat If the value is ≥6 and the UWB signal quality is average, then the weight W of the BDS result is... GNSS =0.8, UWB result weight W UWB =0.2; if N sat If the Nsat value is ≤3 and the UWB signal quality is good, the BDS weight is reduced to 0.2 and the UWB weight is increased to 0.8. In other cases, linear interpolation is used to smoothly transition the fusion weights based on the Nsat value and the UWB quality scoring factor.

[0086] Final fusion localization result X fusion Calculated using the following weighted average:

[0087] X fusion =W BDS ·X BDS +W UWB ·X UWB

[0088] Among them, X GNSS X UWB W represents the coordinates estimated by the BeiDou system and the UWB system at the current moment, respectively. GNSS W UWB For the corresponding fusion weights, satisfy W GNSS +W UWB =1 To avoid trajectory jumps caused by sudden weight changes, this system is designed with a weight gradual change mechanism. That is, in each fusion cycle, the weight change of BDS and UWB shall not exceed 10%, ensuring the continuity and stability of the weight change process.

[0089] Furthermore, to further improve the continuity and robustness of the positioning trajectory, the fusion result X obtained in each cycle is... fusion By performing sliding window weighted averaging, the impact of occasional single-point errors on the overall trajectory is effectively suppressed, achieving centimeter-level error control and improved trajectory smoothness.

[0090] Example 7 also includes the following alternatives:

[0091] Alternatives to the joint error estimation module:

[0092] Alternative Solution 1: Replace the multi-model Kalman filter with a deep learning-based error prediction network, train the network parameters using historical error data, and achieve dynamic compensation for clock errors and multipath errors.

[0093] Alternative Solution 2: Adopt a federated filtering architecture, design independent filters for BeiDou clock error and UWB base station delay, and then output joint estimation results through information fusion algorithm.

[0094] Alternatives to the ambiguity fixation algorithm:

[0095] Alternative Solution 1: Introduce lattice basis reduction techniques into the LAMBDA algorithm to further reduce computational complexity by transforming the search space.

[0096] Alternative Solution 2: Use a fuzzy clustering algorithm to classify floating-point fuzzy solutions, prioritizing the search for integer combinations corresponding to cluster centers.

[0097] Alternatives to coordinate transformation and fusion:

[0098] Alternative Solution 1: Replace least squares calibration with a robust registration algorithm based on RANSAC to improve calibration accuracy by removing outliers.

[0099] Alternative Solution 2: Introduce signal multipath features into dynamic weight allocation to construct a multi-dimensional weight decision model.

[0100] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A real-time dynamic high-precision positioning method based on tight coupling of BeiDou and UWB, comprising a data preprocessing and spatiotemporal alignment module, an error joint estimation model, a fast ambiguity fixing algorithm, a geometric constraint enhanced positioning engine module, and a dynamic coordinate transformation and fusion output module. In the tight coupling positioning method of BeiDou and UWB, data preprocessing and spatiotemporal alignment are first performed. The timestamps of the original BeiDou data and UWB ranging data are aligned using hardware PPS signals, and linear interpolation is used to compensate for sub-millisecond deviations, ensuring consistency of the time reference for multi-source data. Simultaneously, BeiDou positioning results and UWB base station coordinates are collected in open areas. The initial coordinate transformation matrix is ​​calculated using the least squares method, with calibration accuracy controlled within 2cm. Outliers are removed based on the 3σ criterion to provide high-quality input for deep fusion. Next, an error joint estimation model is constructed, utilizing an adaptive Kalman filter algorithm. Real-time estimation and compensation of BeiDou clock bias, UWB clock bias, and multipath errors significantly improve estimation accuracy and suppress error coupling in dynamic environments. Regarding ambiguity fixing, the LAMBDA algorithm is improved by employing a block search strategy and a ratio check mechanism to quickly fix ambiguities, overcoming efficiency bottlenecks and meeting the real-time requirements of high-dynamic scenarios. Geometric constraints enhance the positioning engine by introducing the UWB ranging equation as a supplementary constraint when the number of satellites is low, jointly solving for position increments, and using singular value decomposition to handle rank-deficient equations. A residual feedback mechanism is designed to suppress the influence of multipath and non-line-of-sight errors, maintaining positioning continuity. Finally, dynamic coordinate transformation and fusion output dynamically allocate weights based on the number of satellites and UWB signal quality, smoothly transitioning between indoor and outdoor areas through weighted fusion results, controlling switching errors to the centimeter level, improving trajectory smoothness, and achieving seamless positioning across all scenarios.

2. The real-time dynamic high-precision positioning method based on tight coupling of BeiDou and UWB as described in claim 1, characterized in that, The data preprocessing and spatiotemporal alignment involved aligning the raw BeiDou data (including pseudorange ρ and carrier phase φ) with the UWB ranging data d. First, timestamp alignment was performed using a hardware PPS signal to ensure temporal consistency between BeiDou and UWB data, laying the foundation for subsequent data fusion. Since the hardware clock may have sub-millisecond deviations, linear interpolation was used to compensate for these subtle time differences. In open areas, BeiDou positioning results and UWB base station coordinates X were collected. UWB Using this data, the initial coordinate transformation matrices R and T are calculated using the least squares method. These transformation matrices are used to transform the local coordinates of UWB to the ECEF coordinate system of BeiDou, thereby achieving spatial alignment of the two positioning data sets. The calibration accuracy is controlled to ≤2cm. Furthermore, to eliminate the influence of outlier data on subsequent calculations, sliding window statistics are performed on both BeiDou pseudorange and UWB ranging data, and outliers are removed based on the 3σ criterion. This effectively avoids interference from outliers in the positioning results, providing high-quality input data for deep fusion. The initial coordinate transformation matrices R and T calculated using the least squares method are as follows:

3. The real-time dynamic high-precision positioning method based on tight coupling of BeiDou and UWB as described in claim 1, characterized in that, The joint error estimation model is designed to accurately estimate and compensate for the BeiDou clock error b. GNSS UWB clock difference b UWB and multipath error e MP The state equations were constructed as follows: x k+1 =Fx k +w k z k =Hx k +v k in, Let F be the state vector, and H be the state transition matrix, which describes the change of state over time. H is the observation matrix, which maps the state vector to the observation space. w and v are the process noise and measurement noise, respectively, reflecting the uncertainty and observation error of the system. The adaptive Kalman filter algorithm iteratively updates the Kalman filter gain matrix, estimates and compensates for clock bias and multipath error in real time. By continuously adjusting the filter parameters, the algorithm can adapt to the noise characteristics under different environments, improve the accuracy of estimation, and achieve nanosecond-level accuracy in BeiDou clock bias estimation. It significantly suppresses error coupling phenomena under dynamic environments and improves the long-term stability of the system.

4. The real-time dynamic high-precision positioning method based on tight coupling of BeiDou and UWB as described in claim 1, characterized in that, The proposed fast ambiguity fixing algorithm uses the UWB ranging equation to provide positional geometric constraints, compressing the integer ambiguity search dimension from 6 dimensions to less than 3 dimensions, and improving the LAMBDA algorithm. Where Q is the ambiguity covariance matrix, reflecting the correlation between ambiguities; a * The floating-point fuzzy solution is the starting point for fuzzy search. It adopts a block search strategy and a ratio test mechanism to prioritize the selection of the optimal fuzzy combination. Through block search, the search space can be divided into multiple small blocks for separate searches, thereby improving search efficiency. At the same time, the ratio test mechanism is used to evaluate the reliability of different fuzzy combinations to ensure that the selected fuzzy combination is optimal, breaking through the bottleneck of efficiency of fixed fuzzy values.

5. The real-time dynamic high-precision positioning method based on tight coupling of BeiDou and UWB as described in claim 1, characterized in that, The geometric constraint-enhanced positioning engine, when the number of satellites N sat When the value is ≥4, the position is calculated using the BeiDou observation equation: ρ i =‖X-X i ‖+cb GNSS +e MP,i N sat When the value is 2 or 3, the UWB ranging equation is introduced as a supplementary constraint to jointly solve for the position increment: d j =‖X-X j ‖ In the solution process, singular value decomposition is used to handle the rank-deficient equation, avoiding the problem of matrix inversion failure. At the same time, it has a feedback mechanism that dynamically adjusts the observation weights according to the observation residuals, suppressing the impact of multipath and non-line-of-sight errors on the positioning results. This mechanism can maintain the continuity of positioning in occluded scenarios.

6. The real-time dynamic high-precision positioning method based on tight coupling of BeiDou and UWB as described in claim 1, characterized in that, To achieve seamless fusion of BeiDou and UWB positioning results, the dynamic coordinate transformation and fusion output module needs to estimate the transformation parameters of the two coordinate systems in real time. It dynamically allocates weights based on the number of satellites and the quality of UWB signals to ensure accurate positioning results in different environments. When the number of satellites decreases, the weight of UWB is gradually increased from 20% to 80% to fully utilize UWB positioning information. In the transition area between indoor and outdoor environments, we achieve a smooth transition through weighted fusion results, avoiding trajectory jumps caused by relying on fixed thresholds. The switching error is controlled within centimeters, and the trajectory smoothness is significantly improved.

Images