UWB base station coordinate automatic calibration method and device

Abstract

This invention discloses an automatic calibration method and apparatus for UWB base station coordinates. The method includes: acquiring ranging data between base stations in a base station set and prior information about the base stations; constructing a constraint dataset based on the ranging data and prior information; constructing a nonlinear optimization model with base station coordinates as variables based on the constraint dataset, and performing robust optimization on the nonlinear optimization model to obtain an initial solution for the base station coordinates; calculating the uncertainty information of each base station coordinate based on the initial solution, determining the target base station to be fixed and the target coordinate dimension based on the uncertainty information, generating corresponding prior constraints based on the target coordinate dimension, and updating the constraint dataset based on the prior constraints; and re-executing the robust optimization solution based on the updated constraint dataset to obtain the final solution for the base station coordinates. This invention solves the technical problem of inaccurate base station coordinate determination in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of wireless positioning and geometric calibration technology, and more specifically, to a method and apparatus for automatic calibration of UWB base station coordinates. Background Technology

[0002] UWB (Ultra-Wideband) positioning systems typically consist of multiple base stations and tags. When performing tag positioning, it is necessary to obtain the three-dimensional coordinate information of each base station in a unified coordinate system beforehand. The accuracy of the base station coordinates directly affects the overall accuracy of the positioning system.

[0003] In traditional deployment methods, base station coordinates are typically measured manually point by point, or a coordinate system is established using external measuring tools such as total stations and laser rangefinders. These methods have the following problems: high labor costs, long deployment cycles, susceptibility to human measurement errors, and difficulty in achieving high-precision measurements in environments with many obstructions or complex spatial structures, such as factories and warehouses.

[0004] To reduce the degree of manual intervention, existing technologies have proposed various base station self-calibration methods, which are mainly based on one or a combination of the following approaches: First, using only automatically measured pairwise distance information between base stations, a nonlinear least squares optimization model is constructed with base station coordinates as variables, and solved using an iterative algorithm. To eliminate the non-uniqueness of the coordinate system, it is usually necessary to manually select some base stations as reference benchmarks, such as fixing the three-dimensional coordinates of some base stations or fixing the coordinate dimensions or distance relationships of some base stations. Second, the base station coordinates are initialized or solved globally using multi-dimensional scaling (MDS) or graph optimization methods, and then fine-tuned using nonlinear optimization methods. Third, outliers in the ranging data are removed by setting distance thresholds, packet reception rate thresholds, etc., or by using robust kernel functions to suppress abnormal ranging. Overall, the above-mentioned existing technical solutions typically use the distance constraints between base stations as the main information source, supplemented by a small number of fixed points or fixed dimensions to determine the coordinate system.

[0005] However, in large-scale deployments or complex environments, the above-mentioned technical solutions still have the following shortcomings: First, insufficient constraints can lead to underdetermined or ill-conditioned problems. When there are few ranging edges between base stations, sparse network topology, or insufficient fixed references, relying solely on distance constraints is insufficient to uniquely determine the solution, easily resulting in unstable solutions or convergence to distorted solutions. Second, they are sensitive to abnormal ranging data. UWB ranging is easily affected by multipath effects, occlusion, and non-line-of-sight (NLOS) factors, generating a large number of outliers. Relying solely on simple threshold removal or single robust optimization is insufficient to effectively handle structurally abnormal data, potentially causing a shift in the overall solution. Third, the nonlinear optimization process is prone to getting trapped in local optima. When there are many base stations and high dimensionality of optimization variables, the optimization result is highly dependent on the initial values, and a single solution can easily get trapped in local optima, thus affecting calibration accuracy. In addition, in actual deployments, base station heights are usually quite similar, resulting in relatively weak observation information in the Z direction. Without effective height constraints, height drift can easily occur, further affecting overall positioning accuracy.

[0006] Therefore, how to achieve stable and accurate automatic calibration of base station coordinates under the conditions of ranging anomalies, insufficient constraints, and inadequate observations has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] This invention provides a method and apparatus for automatic calibration of UWB base station coordinates, which at least solves the technical problem of inaccurate base station coordinate calculation in the prior art.

[0008] According to one aspect of the present invention, an automatic calibration method for UWB base station coordinates is provided, comprising: acquiring ranging data between base stations in a base station set and prior information of the base stations, and constructing an original dataset based on the ranging data and the prior information, wherein the prior information includes at least some coordinate information or distance information of the base stations; performing outlier removal and validity screening on the ranging data in the original dataset to obtain preprocessed ranging data, and constructing a constraint dataset based on the preprocessed ranging data and the prior information; constructing a nonlinear optimization model with base station coordinates as variables based on the constraint dataset, and performing robust optimization on the nonlinear optimization model to obtain an initial solution for the base station coordinates; calculating the uncertainty information of each base station coordinate based on the initial solution for the base station coordinates, determining the target base station to be fixed and the target coordinate dimension according to the uncertainty information, generating corresponding prior constraints based on the target coordinate dimension and updating the constraint dataset based on the prior constraints; and re-executing the robust optimization solution based on the updated constraint dataset until a preset termination condition is met to obtain the final solution for the base station coordinates.

[0009] According to another aspect of the present invention, an automatic UWB base station coordinate calibration device is also provided, comprising: a dataset construction module configured to acquire ranging data between base stations in a base station set and prior information of the base stations, and construct an original dataset based on the ranging data and the prior information, wherein the prior information includes at least some coordinate information or distance information of the base stations; a constraint construction module configured to perform outlier filtering and validity filtering on the ranging data in the original dataset to obtain preprocessed ranging data, and construct a constraint dataset based on the preprocessed ranging data and the prior information; and an initial solution solving module configured to... To construct a nonlinear optimization model with base station coordinates as variables based on the constraint dataset, and to perform robust optimization on the nonlinear optimization model to obtain an initial solution for the base station coordinates; a constraint update module is configured to calculate the uncertainty information of each base station coordinate based on the initial solution of the base station coordinates, determine the target base station and target coordinate dimension to be fixed according to the uncertainty information, generate corresponding prior constraints based on the target coordinate dimension, and update the constraint dataset based on the prior constraints; a final solution solving module is configured to re-execute the robust optimization solution based on the updated constraint dataset until a preset termination condition is met to obtain the final solution for the base station coordinates.

[0010] In this embodiment of the invention, the above solution solves the technical problem of inaccurate base station coordinate calculation in the prior art. Attached Figure Description

[0011] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0012] Figure 1 This is a flowchart of an optional automatic UWB base station coordinate calibration method according to an embodiment of the present invention;

[0013] Figure 2 This is a flowchart of another optional automatic UWB base station coordinate calibration method according to an embodiment of the present invention;

[0014] Figure 3 This is a flowchart of an optional preprocessing method for eliminating constant distance edges according to an embodiment of the present invention;

[0015] Figure 4 This is a flowchart of an optional solution method according to an embodiment of the present invention;

[0016] Figure 5 This is an optional flowchart based on uncertainty-based dimension-by-dimensional fixation according to an embodiment of the present invention;

[0017] Figure 6 This is a schematic diagram of an optional automatic UWB base station coordinate calibration device according to an embodiment of the present invention;

[0018] Figure 7 A schematic diagram of the structure of a computer device suitable for implementing embodiments of the present disclosure is shown. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0020] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] Terminology Explanation

[0022] Anchor: A fixed node whose coordinates need to be calculated, with a unique ID.

[0023] Edge: Distance measurement between two base stations.

[0024] Packet reception rate (rxrate): This index represents the percentage of available measurements for a given ranging edge within a statistical period, reflecting the reliability of that edge.

[0025] Reference location (ref) x / ref y / ref z : The true value and standard deviation sigma of a base station along a certain axis, given manually or by known methods.

[0026] Reference distance (ref) dis ): The actual distance between two base stations measured manually and its standard deviation sigma.

[0027] Fixed dimension: A base station is considered "fixed" because it has been pinned to a certain coordinate axis by strong prior constraints.

[0028] Interior / Exterior Points: Confidential ranging edges (interior points) and abnormal ranging edges (exterior points) determined by the distance error threshold.

[0029] Tag stationary event: When a tag is stationary and located near or directly below a base station, the base station ID directly above it, the tag height measurement, and the distance between the tag and surrounding base stations at that moment can be obtained.

[0030] Label variable: Use the label coordinates of static events as the optimization variable q. i .

[0031] Directly below constraint: The label XY is unknown, but should be close to the XY of the base station directly above it, expressed as a soft constraint.

[0032] Credibility metrics: except for the package acceptance rate (rx) rate In addition, it may include ranging variance estimation, historical stability index, NLOS discrimination results, hardware output quality indicators such as RSSI / FP, or combinations thereof.

[0033] Statistics window: rx rate The reliability index can be obtained statistically within a preset time window or a preset number of packets window, such as the percentage of valid ranging times within T seconds or N ranging frames.

[0034] Overall concept

[0035] This invention proposes an automatic base station coordinate calibration method based on multi-source constraints, robust phased nonlinear optimization, and uncertainty-based dimensional fixation (gradual pinning). The method divides the data into: 1) Automatic ranging data: distance measurements between base stations and their reliability indicators (e.g., packet reception rate rx). rate ); 2) Manual prior knowledge: true coordinates of some base stations (may only be a part of x / y / z), single-axis distances of some base stations to reference planes such as walls, and true distances between some base stations measured manually. 3) Tag stationary point data: by moving and stopping the tag under the base station, the distance between the tag and the surrounding base stations and the tag height information are obtained as additional constraints; 4) Constraint boundaries: the coordinate range of the overall scene (or the range derived from the fixed base stations), and the reasonable range of distances, etc.

[0036] In the optimization process, the original ranging measurements are first preprocessed and outliers are removed. Then, multiple random initial values ​​are used for parallel solution to improve global accuracy. A robust loss function, which is used to optimize in stages from coarse to fine, is then used to suppress outliers. Subsequently, based on the covariance / standard deviation estimates obtained from the solution, the most reliable unfixed dimension is iteratively selected and gradually transformed into a strong prior (fixed), thereby improving solvability and stability round by round until it can no longer be fixed or the preset termination condition is reached.

[0037] In ultra-large-scale deployment scenarios (e.g., coverage areas up to 1000m × 1000m, and the number of base stations up to 1000 or more), this invention can also adopt a hierarchical solution approach of block-based decomposition, inter-block alignment, and global refinement to reduce the scale of a single solution and improve computational efficiency and stability. The core idea is to divide the entire network ranging map into several subgraphs (blocks) with overlapping nodes or cross-block connecting edges, solve the base station coordinates within each block separately, then use shared base station / cross-block reference distances between blocks for coordinate system alignment, and finally perform one or more joint optimization refinements globally.

[0038] According to an embodiment of the present invention, a method embodiment for automatic calibration of UWB base station coordinates is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0039] Figure 1 This is an automatic coordinate calibration method for UWB base stations according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0040] Step S102: Obtain ranging data between base stations in the base station set and prior information of the base stations, and construct an original dataset based on the ranging data and the prior information, wherein the prior information includes at least some coordinate information or distance information of the base stations.

[0041] Step S104: Perform outlier removal and validity screening on the distance measurement data in the original dataset to obtain preprocessed distance measurement data, and construct a constraint dataset based on the preprocessed distance measurement data and the prior information.

[0042] For example, based on the reliability index of the ranging data, the ranging data is filtered, and ranging data with a reliability lower than a preset threshold is removed to obtain first filtered data; based on the reasonable range of the ranging values, the first filtered data is range-filtered to remove ranging data outside the preset range to obtain second filtered data; based on the triangular consistency constraint, the second filtered data is judged for consistency to remove ranging data that does not satisfy the triangular relationship, to obtain the preprocessed ranging data. Then, a constraint dataset is constructed based on the preprocessed ranging data and the prior information.

[0043] Step S106: Construct a nonlinear optimization model with base station coordinates as variables based on the constraint dataset, and perform robust optimization on the nonlinear optimization model to obtain the initial solution for the base station coordinates.

[0044] First, distance constraints between base stations are constructed based on the preprocessed ranging data to obtain a set of ranging residuals; position constraints and reference distance constraints are constructed based on the prior information to obtain a set of prior residuals; the set of ranging residuals and the set of prior residuals are fused to construct a unified objective function to obtain the nonlinear optimization model.

[0045] Next, the nonlinear optimization model is initialized based on multiple sets of random initial values ​​to obtain multiple initial solutions. For each initial solution, a robust kernel function with high tolerance is used to optimize the nonlinear optimization model in the first stage to obtain a coarse optimization result. Based on the coarse optimization result, the tolerance of the robust kernel function is gradually reduced in subsequent stages to perform fine optimization on the nonlinear optimization model to obtain a fine optimization result. The result with the optimal objective function value is selected from the fine optimization result as the initial solution for the base station coordinates.

[0046] Step S108: Calculate the uncertainty information of each base station coordinate based on the initial solution of the base station coordinates, determine the target base station to be fixed and the target coordinate dimension based on the uncertainty information, generate corresponding prior constraints based on the target coordinate dimension, and update the constraint dataset based on the prior constraints.

[0047] First, the covariance matrix corresponding to the initial solution is calculated, and the covariance matrix is ​​processed to obtain the standard deviation of each base station in each coordinate dimension, which is used as the uncertainty information of the base station coordinates. Based on the preset topology admission rules, a set of candidate base stations is selected from the base stations that are not completely fixed. Based on the uncertainty information of the coordinates of each base station in the set of candidate base stations, the target base station is determined, and the coordinate dimension with the smallest uncertainty is selected as the target coordinate dimension among the unfixed coordinate dimensions of the target base station.

[0048] Next, the coordinate values ​​of the target coordinate dimension are used as prior values, and prior weights are determined based on the uncertainty information of the target coordinate dimension; the prior values ​​and the prior weights are used to construct new prior constraints and added to the constraint dataset to obtain the updated constraint dataset.

[0049] Step S110: Re-execute the robust optimization solution based on the updated constraint dataset until the preset termination condition is met, and obtain the final solution for the base station coordinates.

[0050] The final output includes the 3D coordinates (pos) of each base station, the standard deviation (noise) of each axis for each base station (obtained by taking the square root of the covariance diagonal), and the distance residual (dis) for each ranging edge. residual And evaluation information (such as total RMSE, interior point RMSE, interior point proportion, maximum interior point error, optimization convergence summary, etc.).

[0051] In this embodiment, when there are a large number of outlier ranging values, the network scale is large, and some base stations can only obtain some prior information (such as height, single-axis distance against the wall, and a small amount of manual ranging), the technical problem of how to stably, automatically, and scalably solve the three-dimensional coordinates of the base station is solved, and error and uncertainty indices that can be used for subsequent quality assessment are provided.

[0052] This embodiment provides an automatic coordinate calibration method for UWB base stations. Based on multi-source constraints and robust nonlinear least squares optimization, this method automatically solves for the three-dimensional coordinates of the base station and outputs uncertainty and evaluation information. The method is as follows... Figure 2 As shown, it includes the following steps:

[0053] Step S202, data input and parameter configuration.

[0054] Obtain the input data. The input data should include at least: 1) a set of base stations and their pairwise ranging data: each data point contains distance values ​​dis and rx. rate ;2) Optional: refs for some base stations x / ref y / ref z 3) Optional: Reference distance between some base stations dis 4) Optional: Scene range (minimum and maximum values ​​of x / y / z); 5) Parameter: Distance within a reasonable range min / dis max Basic ranging noise sigmabase Random initialization count startcount Number of iterations in a single LM cycle count Interior point error threshold dis errorthreshold wait.

[0055] Optionally, it also includes tag stationary point data: 1) List of stationary events: each event contains the base station ID directly above and the tag height measurement value (tag). z And a list of distance measurements from that time tag to surrounding base stations. ;2) : The known vertical offset from the tag reference point to the base station reference point.

[0056] Combinations of manual priors include, but are not limited to: base station height (z) priors, i.e., base station height is usually easier to measure during deployment and can provide references for all or some base stations. z Precise coordinates of corner points or reference points, such as scene corner points or key locations where base stations provide complete (x,y,z) or at least (x,y) reference positions; precise coordinates of partial dimensions, such as wall-mounted base stations providing x or y single-axis reference positions; and some manual distances between base stations, used for blind spot filling, cross-region connectivity, inter-block alignment, or improving overall accuracy.

[0057] Step S204, abnormal ranging edge removal preprocessing.

[0058] Perform the following on the raw ranging data: Figure 3 The following processing is shown:

[0059] Step S2042, packet reception rate filtering.

[0060] Delete rx rate The ranging edge is below a threshold value, which can be 0.1 to 0.5, typically 0.3.

[0061] Step S2044, distance range filtering.

[0062] Delete distance less than dis min or greater than dis max The ranging edge, where dis max The setting ranges from 50m to 1000m depending on the scenario, and from 1000m to 3000m in very large scenarios; for slightly negative distances, the value is truncated to 0 or a small positive number.

[0063] Step S2046, triangular consistency filtering.

[0064] For any edge If there exists a common neighbor k such that Then determine the edge These are considered abnormal and removed. This represents the ranging edge to be inspected, i.e., the node whose anomaly needs to be determined. With nodes The distance between them; Common neighbor nodes refer to nodes that are simultaneously connected to each other. and nodes There is a third node in the distance measurement relationship in all of them; and Representing nodes respectively With nodes ,node With nodes The distance between them; It is the ranging edge of the target to be inspected; This is a tolerance parameter, ranging from 0.1m to 2m, with a typical value of 0.5m. Triangular consistency filtering can effectively remove excessively large outlier ranging edges, especially effective for distance elongation anomalies caused by NLOS.

[0065] Step S206: Construct a multi-source constraint optimization model.

[0066] With each base station coordinate For the variable to be estimated; optionally, for each label static event Introducing label coordinates: Construct the following residual terms:

[0067] (1) Base station-to-base station ranging residual (for all base station pairs (i,j) with ranging observations):

[0068]

[0069] in Indicates base station With base station The distance measurement residual between; and Base stations and base stations The three-dimensional coordinate variables; It is a base station and The Euclidean distance between them; For base stations and Automatic distance measurement values ​​(observations) between them; For ranging weights, and rx rate Proportional.

[0070] (2) Reference position residual (can be uniaxial only, for all (i,a), a∈{x,y,z} with prior knowledge):

[0071] in, Indicates base station On the coordinate axes Positional residuals on; For base stations On the axis The coordinate components to be estimated on (e.g.) , or ); It is a base station On the axis Reference coordinates (prior values) on the reference coordinates; Indicates the direction of the coordinate axes.

[0072] (3) Reference distance residuals for all base station pairs (i,j) with artificial reference distances:

[0073]

[0074] in, Indicates base station and The artificial reference distance residual; The Euclidean distance between base stations (estimated value); It is a base station and The manually measured reference distance between them.

[0075] (4) Tag stationary point constraint (optional): Let event t correspond to base station i=a(t) directly above it. This event provides tag height measurement. Ranging from the tag to the surrounding base station j .

[0076] Soft constraint directly below:

[0077]

[0078] in, This represents the residual deviation between the tag and the base station directly above it in the XY plane. Index for tagged static events; Indicates an event The corresponding base station number directly above; For the event The corresponding label coordinate variables; These are the coordinates of the label in the XY direction; It is a base station Coordinates in the XY direction; The weight (or corresponding standard deviation) of this soft constraint. ).

[0079] Prior knowledge for label height measurement:

[0080]

[0081] in, Indicates the label height residual; This is an estimate of the label in the Z direction; It is a measured (observed) value of the label height; Weights for height measurement constraints.

[0082] Vertical geometry of base station-tag (Δz is a known vertical offset):

[0083]

[0084] in, This represents the vertical relationship residual between the base station and the tag; For base stations Height (Z coordinate); This is an estimated label height. The known vertical offset from the tag reference point to the base station reference point.

[0085] Tag - Base station ranging residual (for all (t,j) with this observation):

[0086]

[0087] in, Indicates tags and base stations The ranging residual; Label coordinate variables (corresponding to events) ); It is a base station The coordinates; The distance between the tag and the base station is in Euclidean form. It is a tag to a base station The distance measurement observation value.

[0088] Range constraints. Set upper and lower bounds for each coordinate axis, for example: The range can be input by the user or derived from the coordinates of a fixed base station.

[0089] Coordinate system uniqueness constraint. When relying solely on pairwise distance constraints, the solution often exhibits non-uniqueness due to global translation, global rotation (and possibly mirroring / scale in 3D). This invention eliminates or suppresses this non-uniqueness by introducing at least one of the following methods, making the solution unique and stable in an engineering sense: 1) Absolute coordinate constraints of at least one or more base stations (ref x / ref y / ref z ); 2) Partial dimensional coordinate constraints (fixed on a single axis); 3) Reference distance constraints (ref dis ) and / or scenario range constraints; 4) The dimension-wise fixing mechanism based on uncertainty gradually increases strong priors during the iteration process.

[0090] Construct the objective function:

[0091]

[0092] in For robust kernel functions (such as Cauchy / Huber / Tukey). Represents the coordinates of all base stations and label coordinates Optimize; Robust kernel functions (such as Cauchy, Huber, Tukey) are used to suppress outliers; summation symbol This represents the summation of the ranging edges over all base stations. This indicates a priori summation over all positions. This represents the summation of all labeled static events; It is the square norm of the vector residual.

[0093] In this embodiment, the three-dimensional coordinates of the base station to be located are used. For the variable to be estimated, optionally for each label static event. Introducing label coordinates The objective function is constructed based on at least one of the following: base station-to-base station ranging residual, reference position residual, reference distance residual, soft constraint residual directly below the tag, prior residual of tag height measurement, base station-tag vertical geometric relationship residual, and tag-to-base station ranging residual. A robust kernel function is then applied to the objective function. By suppressing outliers and solving for each variable to be estimated through least squares optimization, the base station coordinates can be calculated more accurately.

[0094] Step S208: Solve the problem.

[0095] like Figure 4 As shown, the solution process includes the following steps:

[0096] Step S2082: Solve multiple random initial values ​​in parallel.

[0097] Within the range constraints, random initial values ​​are assigned to the unfixed dimensions, while the fixed dimensions retain their reference values. Multiple sets of initial values ​​(10 to 500 sets, typically 100 sets) are generated, and optimization is performed in parallel.

[0098] Step S2084: Phased robust optimization.

[0099] For each set of initial values, multiple rounds of optimization are performed using kernel function parameters ranging from strong robustness and tolerance to weak robustness and strict tolerance (e.g., the Cauchy kernel scale decreases from large to small) to achieve coarse alignment followed by fine convergence.

[0100] Step S2086: Select the optimal result.

[0101] The solution with the smallest objective function and successful convergence is selected from all parallel results as the best solution for this round; at the same time, the interior point RMSE, interior point ratio, etc. can be output as quality indicators.

[0102] Step S210, dimension-wise fixation based on uncertainty.

[0103] To address the instability caused by underdetermined / weak constraints, this invention proposes a dimension-by-dimensional fixing mechanism. First, the estimated coordinates and uncertainty information of each base station obtained in the current round of solution are acquired. Based on preset topology admission rules, a candidate base station set is determined from the base stations that are not fully fixed. These topology admission rules include at least the following: at least two base stations in the candidate base station set and the fixed base station set have ranging constraints or other equivalent connection constraints. For the candidate base station set, the target base station to be fixed and the target coordinate dimension are determined, based at least on the uncertainty information of each candidate base station. The estimated value of the target coordinate dimension is used to generate prior constraints and added to the optimization problem. The weight of the prior constraints is related to the uncertainty of that dimension or is a preset strong constraint weight. After adding the prior constraints, the next round of optimization is executed until a termination condition is met. The termination condition includes: the candidate set is empty, a preset maximum number of rounds is reached, or all base station coordinate dimensions are fixed. An optional rollback strategy is that if adding a strong prior causes solution failure or a significant deterioration in quality indicators, the fixing can be revoked, the next candidate dimension can be selected, or the prior sigma can be relaxed, and the solution can be recalculated.

[0104] Specifically, such as Figure 5 As shown, the method includes the following steps:

[0105] Step S2102, candidate base station screening.

[0106] Let F be the set of currently fixed nodes. For a node u that is not fully fixed, if it has a distance-measuring edge with at least two fixed nodes in set F (or satisfies a preset topology condition), then u is added to the candidate set C. This rule is used to avoid introducing unreliable fixing constraints based on a single fixed neighbor.

[0107] Step S2104, uncertainty assessment.

[0108] For the optimal solution in this round, calculate the covariance matrix of each base station coordinate, and take the square root of the diagonal to obtain the standard deviation of each axis. This represents the uncertainty in that dimension.

[0109] Step S2106, Fixed dimension selection.

[0110] In the candidate set C, the base station with the lowest overall uncertainty is selected first (e.g., selecting...). (The smallest), and then select the dimension with the smallest standard deviation among the dimensions that are not yet fixed to fix it.

[0111] Step S2108: Generate strong priors and proceed to the next round.

[0112] Use the estimated value of the selected dimension as the new reference position (and assign a smaller sigma). Add the candidate node (in meters) to the constraint set and return to S208 to continue iteration. The termination condition is when the candidate set is empty, all three dimensions of all nodes are fixed, or the maximum number of iterations is reached, and the final result is output.

[0113] Dimensional fixing is not simply fixing integer points, but rather using covariance-driven, progressively increasing constraint strength dimension by dimension, balancing stability and scalability. Let the set of fixed nodes be... For a node u that is not completely fixed, if it is related to the set If at least two nodes have a ranging edge, then add it to the candidate set. .

[0114] Step S212, quality assessment and optional internal point statistics.

[0115] To facilitate deployment acceptance and automatic diagnosis, the following can be calculated: 1) Total RMSE: Root mean square of all distance measurement edge errors (largely affected by outside points, for reference only); 2) Interior point RMSE: Only edges with errors not exceeding the threshold T (e.g., 0.2m~2m, typically 1m); 3) Interior point ratio: The proportion of interior point edges; 4) Maximum interior point error: The maximum error in the set of interior points.

[0116] The above indicators can be used to automatically determine the reliability of the calibration and to suggest the need for additional manual constraints or threshold adjustments. Optional failure diagnosis and rollback suggestions: 1) If the optimization is underdetermined / non-convergent: suggest adding absolute coordinates / single-axis coordinates / manual distance priors, or expanding candidate admission connectivity, or using block-based solution. 2) If the proportion of interior points is low or the interior point RMSE is high: suggest increasing the preprocessing intensity (e.g., increasing rx). rate Thresholds, tightened triangulation consistency tolerance, or phased robust kernel parameters), or increasing cross-regional artificial distances to improve topology connectivity. 3) If the standard deviation in the Z direction is significantly large: suggest supplementing base station height priors (ref z Alternatively, set a reasonable range for Z.

[0117] The application of this method will be described below using the block-based solution of ultra-large-scale networks as an example.

[0118] When the number of base stations is large (e.g., more than 1000) or the coverage area is large (e.g., up to 1000m × 1000m), resulting in excessive computational load for a single global solution, or when the ranging map contains weak connectivity / long links, the following block-based solution process can be adopted (compatible with S1~S10, and can be used as its upper-layer framework). To avoid unnecessary limitations, this invention does not limit the specific number of blocks, the number of nodes per block, or the details of the alignment algorithm.

[0119] 1) Construction in blocks.

[0120] Based on installation area information, spatial grid, ranging graph connectivity / community structure, or k-hop neighborhood centered on a fixed base station, the entire network base stations and ranging edges are constructed into multiple sub-graph blocks; and for each block, the following are recorded: the set of nodes within the block, the set of edges within the block, the set of overlapping nodes between blocks (shared base stations), and the set of cross-block connecting edges.

[0121] 2) In-block calibration.

[0122] For each sub-block, S302 to S312 are executed independently based on the data within the block to obtain the coordinate solution and uncertainty within the block; optionally, to improve the convergence within the block, a tighter coordinate range constraint is set within the block or the height / corner point / single axis priors available within the block are introduced first.

[0123] 3) Inter-block alignment (coordinate system one).

[0124] For each sub-plot, based on the shared base station set and / or cross-plot constraints between the plot and the aligned plot, calculate the alignment transformation of the plot from the local coordinate system to the global coordinate system, and transform the plot coordinates to the global coordinate system.

[0125] One possible way to align shared base stations is as follows: if there are at least three non-collinear shared base stations between block A and block B (or sufficient spatial distribution in three dimensions), the square error of the shared base station coordinates can be minimized by solving the rigid body transformation (rotation and translation) through least squares fitting; this fitting can be achieved using numerical methods such as SVD.

[0126] Enhanced cross-block constraints: When the number of shared base stations is insufficient, cross-block ranging edges or manual reference distances can be introduced as alignment constraints; during alignment, shared base stations or cross-block edges can be weighted according to uncertainty / confidence.

[0127] Global Refinement: Using the aligned coordinates of the entire network as initial values, construct a joint robust optimization problem or a hierarchical joint optimization problem covering the entire network (you can choose to retain high-confidence edges, cross-block key edges, and manual priors), perform one or more robust optimizations, and output the final coordinates of the entire network base stations, the standard deviation of each axis, residuals, and interior point statistics.

[0128] Block-based solutions can significantly reduce computational and memory pressure, and suppress inter-block drift through "shared base stations / cross-block connections," making them suitable for ultra-large-scale deployments. At the same time, the block results can serve as high-quality initial values ​​for global refinement, improving the final convergence success rate.

[0129] This application also provides an automatic UWB base station coordinate calibration device, including: 1) a data acquisition module: acquiring inter-base station ranging data (including distance values ​​and packet reception rate / confidence), and optional manual measurement data (single-point coordinates, single-axis coordinates, manual ranging, base station height). 2) a data preprocessing module: filtering / correcting ranging edges according to packet reception rate threshold, reasonable distance range threshold, triangulation consistency rules, etc. 3) a constraint construction module: generating position constraints (absolute coordinates / single-axis coordinates), manual distance constraints, automatic distance constraints (with weighted / noise models), and scene range constraints. 4) a solution module: using robust nonlinear least squares (LM) for phased optimization, and supporting parallel solution with multiple random initial values. 5) an iterative fixing module: selecting candidate base stations and their most reliable unfixed dimensions based on the covariance / standard deviation evaluation results, converting them into strong prior constraints, and entering the next round of solution. 6) a result evaluation module: outputting quality indicators such as base station coordinates, standard deviation / confidence of each dimension, distance residuals, inlier ratio, and inlier RMSE. Optionally, it also includes (especially suitable for ultra-large-scale networks): 1) Block management module: Divides the entire network base stations and ranging edges into multiple sub-blocks, and maintains the set of overlapping nodes and cross-block connection edges between blocks. 2) Inter-block alignment module: Based on shared base stations, cross-block ranging edges or manual reference distances, calculates the alignment transformation of each block's local coordinate system to the global coordinate system, and generates the aligned initial values ​​of the entire network.

[0130] This application also provides another automatic coordinate calibration device for UWB base stations, such as... Figure 6As shown, it includes: a dataset construction module 62, configured to acquire ranging data between base stations in a base station set and prior information of the base stations, and construct an original dataset based on the ranging data and the prior information, wherein the prior information includes at least some coordinate information or distance information of the base stations; a constraint construction module 64, configured to perform outlier filtering and validity filtering on the ranging data in the original dataset to obtain preprocessed ranging data, and construct a constraint dataset based on the preprocessed ranging data and the prior information; and an initial solution solving module 66, configured to construct an initial solution based on the constraint dataset and the base station... A nonlinear optimization model with coordinates as variables is used, and robust optimization is performed on the nonlinear optimization model to obtain an initial solution for the base station coordinates. A constraint update module 68 is configured to calculate the uncertainty information of each base station coordinate based on the initial solution of the base station coordinates, determine the target base station and target coordinate dimension to be fixed according to the uncertainty information, generate corresponding prior constraints based on the target coordinate dimension, and update the constraint dataset based on the prior constraints. A final solution solving module 70 is configured to re-execute the robust optimization solution based on the updated constraint dataset until a preset termination condition is met to obtain the final solution for the base station coordinates.

[0131] It should be noted that the UWB base station coordinate automatic calibration device provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the UWB base station coordinate automatic calibration device and the UWB base station coordinate automatic calibration method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0132] Figure 7 A schematic diagram of a computer device suitable for implementing embodiments of the present disclosure is shown. It should be noted that... Figure 7 The computer device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments disclosed herein.

[0133] like Figure 7 As shown, the computer device includes a central processing unit (CPU) 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage section 1008 into a random access memory (RAM) 1003. The RAM 1003 also stores various programs and data required for system operation. The CPU 1001, ROM 1002, and RAM 1003 are interconnected via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0134] The following components are connected to I / O interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to I / O interface 1005 as needed. A removable medium 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 1010 as needed so that computer programs read from it can be installed into storage section 1008 as needed.

[0135] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for automatic coordinate calibration of a UWB base station, characterized in that, include: Obtain ranging data between base stations in the base station set and prior information of the base stations, and construct an original dataset based on the ranging data and the prior information, wherein the prior information includes at least some coordinate information or distance information of the base stations; Outlier removal and validity screening are performed on the distance measurement data in the original dataset to obtain preprocessed distance measurement data, and a constraint dataset is constructed based on the preprocessed distance measurement data and the prior information. Based on the constrained dataset, a nonlinear optimization model with base station coordinates as variables is constructed, and the nonlinear optimization model is robustly optimized to obtain the initial solution for the base station coordinates. The uncertainty information of each base station coordinate is calculated based on the initial solution of the base station coordinates. The target base station to be fixed and the target coordinate dimension are determined according to the uncertainty information. The corresponding prior constraints are generated based on the target coordinate dimension and the constraint dataset is updated based on the prior constraints. The robust optimization solution is re-executed based on the updated constraint dataset until the preset termination condition is met, and the final solution for the base station coordinates is obtained.

2. The method according to claim 1, characterized in that, The ranging data in the original dataset undergoes outlier removal and validity screening to obtain preprocessed ranging data, including: Based on the reliability index of the ranging data, the ranging data is filtered, and ranging data with a reliability lower than a preset threshold is removed to obtain the first filtered data; Based on a reasonable range of distance measurements, the first set of filtered data is filtered to remove distance measurements that are not within the preset range, thus obtaining the second set of filtered data. Based on the triangular consistency constraint, the second screening data is judged for consistency, and the ranging data that does not satisfy the triangular relationship is removed to obtain the preprocessed ranging data.

3. The method according to claim 1, characterized in that, Based on the constrained dataset, a nonlinear optimization model with base station coordinates as variables is constructed, including: Based on the preprocessed ranging data, a distance constraint term between base stations is constructed to obtain a set of ranging residuals. Based on the prior information, position constraint terms and reference distance constraint terms are constructed to obtain the prior residual set; The ranging residual set and the prior residual set are fused to construct a unified objective function, thus obtaining the nonlinear optimization model.

4. The method according to claim 1, characterized in that, Robust optimization solution of the nonlinear optimization model includes: The nonlinear optimization model is initialized based on multiple sets of random initial values ​​to obtain multiple sets of initial solutions; For each initial solution, a robust kernel function with high tolerance is used to optimize the nonlinear optimization model in the first stage to obtain a coarse optimization result. Based on the coarse optimization result, the tolerance of the robust kernel function is gradually reduced in subsequent stages to perform fine optimization on the nonlinear optimization model to obtain a fine optimization result. The result with the optimal objective function value is selected from the optimization results and used as the initial solution for the base station coordinates.

5. The method according to claim 1, characterized in that, Based on the initial solution of the base station coordinates, the uncertainty information of each base station coordinate is calculated, and the target base station to be fixed and the target coordinate dimension are determined according to the uncertainty information, including: Based on the initial solution, the corresponding covariance matrix is ​​calculated, and the covariance matrix is ​​processed to obtain the standard deviation of each base station in each coordinate dimension, which serves as the uncertainty information of the base station coordinates. Based on preset topology admission rules, a set of candidate base stations is selected from base stations that are not completely fixed. Based on the uncertainty information of the coordinates of each base station in the set of candidate base stations, a target base station is determined. Among the unfixed coordinate dimensions of the target base station, the coordinate dimension with the smallest uncertainty is selected as the target coordinate dimension.

6. The method according to claim 1, characterized in that, Generate corresponding prior constraints based on the target coordinate dimensions and update the constraint dataset based on the prior constraints, including: The coordinate values ​​of the target coordinate dimension are used as prior values, and the prior weights are determined based on the uncertainty information of the target coordinate dimension. The prior values ​​and prior weights are used to construct new prior constraints and added to the constraint dataset to obtain the updated constraint dataset.

7. An automatic coordinate calibration device for a UWB base station, characterized in that, include: The dataset construction module is configured to acquire ranging data between base stations in the base station set and prior information of the base stations, and construct an original dataset based on the ranging data and the prior information, wherein the prior information includes at least some coordinate information or distance information of the base stations; The constraint construction module is configured to perform outlier removal and validity screening on the distance measurement data in the original dataset to obtain preprocessed distance measurement data, and construct a constraint dataset based on the preprocessed distance measurement data and the prior information. The initial solution solving module is configured to construct a nonlinear optimization model with base station coordinates as variables based on the constraint dataset, and to perform robust optimization solving on the nonlinear optimization model to obtain the initial solution for the base station coordinates; The constraint update module is configured to calculate the uncertainty information of each base station coordinate based on the initial solution of the base station coordinates, determine the target base station to be fixed and the target coordinate dimension based on the uncertainty information, generate corresponding prior constraints based on the target coordinate dimension, and update the constraint dataset based on the prior constraints. The final solution module is configured to re-execute the robust optimization solution based on the updated constraint dataset until a preset termination condition is met, thereby obtaining the final solution for the base station coordinates.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 6.

9. A computer device, characterized in that, include: Memory and processor The memory stores computer programs; The processor is configured to execute a computer program stored in the memory, wherein when the computer program is executed, the processor performs the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images