A positioning method and apparatus, an electronic device, and a storage medium

Abstract

This application discloses a positioning method and apparatus. The method includes: constructing a wide-area coordinate system for multiple local coordinate systems, and determining the transformation relationship between each local coordinate system and the wide-area coordinate system; obtaining second position data of the target object in the wide-area coordinate system based on first position data of the target object in one or more local coordinate systems and the transformation relationship; determining at least one position prediction data of the target object in the wide-area coordinate system; and determining the positioning position of the target object in the wide-area coordinate system based on the at least one position prediction data and the second position data. This application can achieve the unification of local coordinate system positioning data to wide-area coordinate system positioning data by establishing the transformation relationship between local coordinate systems and wide-area coordinate systems. By performing further position data prediction and fusion under the unified coordinate system, more accurate positioning can be achieved.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a positioning method, device, electronic device, and storage medium. Background Technology

[0002] With the rapid development of mobile communication technology and smart mobile devices, the number of devices requiring high-speed positioning is increasing. Traditional outdoor / wide-area positioning technologies include the Global Positioning System (GPS), BeiDou Navigation Satellite System (BDS), GLONASS, Galileo, and telecommunications base station location services; indoor / regional positioning technologies include Ultra Wide Band (UWB) positioning technology, wireless network communication technologies (i.e., Wi-Fi), Bluetooth, small base stations, light-emitting diode (LED) visible light, and laser positioning.

[0003] In practical applications, it is often necessary to select appropriate positioning equipment and technologies based on different business and application scenarios. Currently, satellite navigation systems serve various sectors of the national economy, especially differential GPS / BeiDou positioning technologies, which have been widely used in high-precision positioning fields such as surveying, transportation, ports, and aviation. Other positioning technologies, such as Wi-Fi, Bluetooth, small base stations, and UWB, mainly meet the needs of indoor positioning and navigation. Among them, Bluetooth positioning technology has been widely used in indoor buildings such as shopping malls, hospitals, airports, and museums.

[0004] This demonstrates that large-scale positioning platforms established by various enterprises and institutions employ diverse types of positioning terminals and methods. Furthermore, these large-scale positioning platforms are geographically widespread, with various sub-positioning platforms overlapping and intersecting. Therefore, achieving data unification and unified presentation on a wide-area positioning platform, and conducting further location data prediction and fusion under a unified coordinate system, presents a new challenge for large-scale positioning platforms. Summary of the Invention

[0005] To address the aforementioned technical problems, embodiments of this application provide a positioning method, apparatus, electronic device, and storage medium.

[0006] This application provides a positioning method, the method comprising:

[0007] Construct a wide-area coordinate system for multiple local coordinate systems, and determine the transformation relationship between each local coordinate system and the wide-area coordinate system.

[0008] The second position data of the target object in the wide-area coordinate system is obtained based on the first position data of the target object in one or more local coordinate systems among the multiple local coordinate systems and the transformation relationship;

[0009] Determine at least one predicted position of the target object in the wide-area coordinate system;

[0010] The target object's location in the wide-area coordinate system is determined based on the at least one location prediction data and the second location data.

[0011] In one optional embodiment of this application, the construction of a wide-area coordinate system for multiple local coordinate systems includes:

[0012] A multi-branch tree structure is established based on the hierarchical relationship between multiple local coordinate systems; wherein, each node in the multi-branch tree structure represents a positioning coordinate system, and the root node of the multi-branch tree structure corresponds to the wide-area coordinate system; the child nodes of each node in the multi-branch tree structure are the child positioning coordinate systems of the positioning coordinate system corresponding to that node, and the parent node of that node is the parent positioning coordinate system of the positioning coordinate system corresponding to that node.

[0013] In an optional embodiment of this application, determining the transformation relationship between each local coordinate system in the plurality of local coordinate systems and the wide-area coordinate system includes:

[0014] For each of the multiple local coordinate systems, determine the shortest path between the node corresponding to that local coordinate system in the multi-branch tree structure and the root node in the multi-branch tree structure. Based on the shortest path, obtain the path node vector corresponding to the shortest path. Based on the coordinate system transformation relationship between each adjacent node in the path node vector, determine the transformation relationship between the local coordinate system and the wide-area coordinate system.

[0015] In an optional embodiment of this application, determining the transformation relationship between each local coordinate system in the plurality of local coordinate systems and the wide-area coordinate system includes:

[0016] For each of the multiple local coordinate systems, the transformation equation between the local coordinate system and the wide area coordinate system is established using the Mologinsky seven-parameter model.

[0017] Based on the location data of the target object at multiple identical locations in the local coordinate system and the wide-area coordinate system, determine the coordinate values ​​of each location of the target object at the multiple identical locations in the local coordinate system and the wide-area coordinate system.

[0018] The seven transformation parameters in the Mologinsky seven-parameter model are determined based on the coordinate values ​​of the target object at each of the multiple identical locations in the local coordinate system and the wide-area coordinate system.

[0019] The coordinate transformation formula between the local coordinate system and the wide-area coordinate system is determined based on the seven transformation parameters.

[0020] In one optional embodiment of this application, the at least one position prediction data includes one or more of the following position prediction data: first position prediction data determined based on the target object's inertial navigation history data, second position prediction data determined based on the target object's historical position data, third position prediction data determined based on the target object's motion direction data, and fourth position prediction data determined based on the target object's historical trajectory.

[0021] In one optional embodiment of this application, determining at least one location prediction data of the target object in the wide-area coordinate system includes:

[0022] The first acceleration vector in the inertial navigation data at the first moment is obtained, and the acceleration components of the first acceleration vector at the first moment are calculated in the coordinate system corresponding to the acceleration sensor of the target object. The local observation value observed by the acceleration sensor is transformed into the second acceleration vector in the local coordinate system of the target using the attitude vector.

[0023] Acquire historical acceleration data from multiple moments prior to the first moment, and construct a third acceleration vector from the historical acceleration data based on the chronological order of the historical acceleration data.

[0024] The acceleration fitting formula corresponding to the third acceleration vector and the first acceleration difference fitting formula are obtained by using the polynomial fitting method.

[0025] Determine the first coefficient corresponding to the acceleration fitting formula and the second coefficient corresponding to the first acceleration difference fitting formula, and obtain the first functional relationship corresponding to the acceleration fitting formula based on the first coefficient, and obtain the second functional relationship corresponding to the first acceleration difference fitting formula based on the second coefficient;

[0026] The second acceleration difference sequence is calculated using the first functional relationship;

[0027] Determine the first mean difference value corresponding to the first acceleration difference value sequence and the second mean difference value corresponding to the second acceleration difference value sequence;

[0028] The target acceleration fitting formula is calculated by selecting the smaller of the first and second mean difference acceleration sequences.

[0029] Based on the target acceleration fitting formula, the predicted first position data corresponding to the first moment is determined.

[0030] In one optional embodiment of this application, determining at least one location prediction data of the target object in the wide-area coordinate system includes:

[0031] Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system.

[0032] The Kalman filter method is used to predict the second location prediction data corresponding to the first time point based on the multiple historical location data.

[0033] In one optional embodiment of this application, determining at least one location prediction data of the target object in the wide-area coordinate system includes:

[0034] Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system.

[0035] Based on the historical location data, determine the acceleration and velocity at a time preceding the first time.

[0036] Based on the target object's motion direction, as well as the acceleration and velocity predictions, the third position prediction data corresponding to the first moment is obtained.

[0037] In one optional embodiment of this application, determining at least one location prediction data of the target object in the wide-area coordinate system includes:

[0038] Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system; each historical location data point is multi-dimensional data.

[0039] Based on the historical location data at the multiple times, a polynomial regression function equation is constructed to represent the change of the target object's position over time.

[0040] Based on the polynomial regression function equation, the fourth position prediction data corresponding to the first time point of the target object is determined.

[0041] In an optional embodiment of this application, determining the location of the target object in the wide-area coordinate system based on the at least one location prediction data and the second location data includes:

[0042] The weights corresponding to each location data in the at least one location data and the weights corresponding to the second location data are determined using a neural network model.

[0043] Based on each location prediction data in the at least one location prediction data, each weight corresponding to each location prediction data, the second location data, and the weight corresponding to the second location data, the location fusion data of the target object is obtained, and the location fusion data is determined as the positioning position of the target object in the wide-area coordinate system.

[0044] In an optional embodiment of this application, determining the weights corresponding to each location data in the at least one location data and the weights corresponding to the second location data using a neural network model includes:

[0045] N experimental samples are obtained, the N experimental samples include N third position data of the target object in the wide-area coordinate system and position prediction data corresponding to each of the N third position data; each of the N third position data and the position prediction data corresponding to that third position data corresponds to one experimental sample;

[0046] Construct an M-dimensional dataset based on the N experimental samples;

[0047] Calculate the covariance matrix corresponding to the M-dimensional data set, and determine the M-1 eigenvalues ​​corresponding to the covariance matrix;

[0048] Determine the M-1 eigenvalues ​​corresponding to the M-dimensional eigenvectors, and obtain a mapping matrix based on the M-dimensional eigenvectors; then, based on the mapping matrix, reduce the dimensionality of the M-dimensional data set to an M-1 dimensional data set.

[0049] The M-1 dimension dataset is normalized to obtain the target dataset;

[0050] The neural network model is trained using the target dataset to obtain the neural network model.

[0051] This application embodiment also provides a positioning device, the device comprising:

[0052] A construction unit is used to construct a wide-area coordinate system for multiple local coordinate systems and to determine the transformation relationship between each local coordinate system and the wide-area coordinate system.

[0053] The processing unit is configured to obtain the second position data of the target object in the wide-area coordinate system based on the first position data of the target object in one or more local coordinate systems among the plurality of local coordinate systems and the transformation relationship;

[0054] The first determining unit is used to determine at least one position prediction data of the target object in the wide-area coordinate system;

[0055] The second determining unit is used to determine the positioning position of the target object in the wide-area coordinate system based on the at least one position prediction data and the second position data.

[0056] In an optional embodiment of this application, the construction unit is specifically used to: establish a multi-branch tree structure based on the hierarchical relationship between multiple local coordinate systems; wherein, each node in the multi-branch tree structure represents a positioning coordinate system, the root node of the multi-branch tree structure corresponds to the wide-area coordinate system; the child nodes of each node in the multi-branch tree structure are the child positioning coordinate systems of the positioning coordinate system corresponding to that node, and the parent node of that node is the parent positioning coordinate system of the positioning coordinate system corresponding to that node.

[0057] In an optional embodiment of this application, the construction unit is further specifically used for: for each of the plurality of local coordinate systems, determining the shortest path between the node corresponding to the local coordinate system in the multi-branch tree structure and the root node in the multi-branch tree structure; obtaining a path node vector corresponding to the shortest path based on the shortest path; and determining the transformation relationship between the local coordinate system and the wide-area coordinate system based on the coordinate system transformation relationship between each adjacent node in the path node vector.

[0058] In an optional embodiment of this application, the construction unit is further specifically used to: for each of the multiple local coordinate systems, establish a transformation equation between the local coordinate system and the wide-area coordinate system using the Mologinski seven-parameter model;

[0059] Based on the location data of the target object at multiple identical locations in the local coordinate system and the wide-area coordinate system, determine the coordinate values ​​of each location of the target object at the multiple identical locations in the local coordinate system and the wide-area coordinate system.

[0060] The seven transformation parameters in the Mologinsky seven-parameter model are determined based on the coordinate values ​​of the target object at each of the multiple identical locations in the local coordinate system and the wide-area coordinate system.

[0061] The coordinate transformation formula between the local coordinate system and the wide-area coordinate system is determined based on the seven transformation parameters.

[0062] In one optional embodiment of this application, the at least one position prediction data includes one or more of the following position prediction data: first position prediction data determined based on the target object's inertial navigation history data, second position prediction data determined based on the target object's historical position data, third position prediction data determined based on the target object's motion direction data, and fourth position prediction data determined based on the target object's historical trajectory.

[0063] In an optional embodiment of this application, the first determining unit is specifically configured to: acquire a first acceleration vector from the inertial navigation data at a first moment; calculate each acceleration component of the first acceleration vector at the first moment in the coordinate system corresponding to the accelerometer of the target object; convert the local observation value observed by the accelerometer into a second acceleration vector in the local coordinate system of the target object using an attitude vector; acquire historical acceleration data from multiple moments prior to the first moment; construct a third acceleration vector from the historical acceleration data based on the chronological order of the historical acceleration data; fit an acceleration fitting formula corresponding to the third acceleration vector and a first acceleration difference fitting formula using a polynomial fitting method; and determine the acceleration... The system calculates the first coefficient corresponding to the velocity fitting formula and the second coefficient corresponding to the first acceleration difference fitting formula, and obtains the first functional relationship corresponding to the acceleration fitting formula based on the first coefficient, and the second functional relationship corresponding to the first acceleration difference fitting formula based on the second coefficient; it calculates the second acceleration difference sequence using the first functional relationship; it determines the first mean difference value corresponding to the first acceleration difference sequence and the second mean difference value corresponding to the second acceleration difference sequence; it selects the smaller mean difference value acceleration sequence between the first mean difference value and the second mean difference value to calculate the target acceleration fitting formula; and it determines the first position prediction data corresponding to the first time point based on the target acceleration fitting formula.

[0064] In an optional embodiment of this application, the first determining unit is specifically used to: acquire historical position data of multiple times prior to the first time; the historical position data is the historical position data of the target object in the wide-area coordinate system; and use the Kalman filter method to predict second position prediction data corresponding to the first time based on the multiple historical position data.

[0065] In an optional embodiment of this application, the first determining unit is specifically used for: acquiring historical position data of multiple times prior to the first time; the historical position data being the historical position data of the target object in the wide-area coordinate system; determining the acceleration and velocity of a time prior to the first time based on the historical position data; and predicting third position prediction data corresponding to the first time based on the motion direction of the target object and the acceleration and velocity.

[0066] In an optional embodiment of this application, the first determining unit is specifically used for:

[0067] Historical position data from multiple time points prior to the first time point are acquired; the historical position data refers to the historical position data of the target object in the wide-area coordinate system; each historical position data is multidimensional data; a polynomial regression function equation for the position of the target object over time is constructed based on the historical position data from the multiple time points; and a fourth position prediction data for the target object corresponding to the first time point is determined based on the polynomial regression function equation.

[0068] In an optional embodiment of this application, the second determining unit is specifically used to: determine the weights corresponding to each location data in the at least one location data and the weights corresponding to the second location data using a neural network model; obtain the location fusion data of the target object based on each location prediction data in the at least one location prediction data, each weight corresponding to each location prediction data, the second location data and the weights corresponding to the second location data, and determine the location fusion data as the positioning position of the target object in the wide-area coordinate system.

[0069] In an optional embodiment of this application, the second determining unit is specifically configured to: obtain N experimental samples, wherein the N experimental samples include N third position data of the target object in the wide-area coordinate system and position prediction data corresponding to each of the N third position data; each of the N third position data and the position prediction data corresponding to that third position data corresponds to one experimental sample; construct an M-dimensional data set based on the N experimental samples; calculate the covariance matrix corresponding to the M-dimensional data set and determine the M-1 eigenvalues ​​corresponding to the covariance matrix; determine the M-1 eigenvalues ​​corresponding to the M-1 eigenvalues ​​and obtain a mapping matrix based on the M-1 eigenvalues; reduce the dimensionality of the M-dimensional data set to an M-1 dimensional dataset based on the mapping matrix; normalize the M-1 dimensional dataset to obtain a target dataset; and train the neural network model to be trained using the target dataset to obtain the neural network model.

[0070] This application also provides an electronic device, which includes a memory and a processor. The memory stores computer-executable instructions, and the processor can implement the methods described in the above embodiments when it executes the computer-executable instructions in the memory.

[0071] This application also provides a computer storage medium storing executable instructions that, when executed by a processor, implement the methods described in the above embodiments.

[0072] The technical solution of this application embodiment constructs a wide-area coordinate system for multiple local coordinate systems and determines the transformation relationship between each local coordinate system and the wide-area coordinate system; based on the first position data of the target object in one or more local coordinate systems and the transformation relationship, a second position data of the target object in the wide-area coordinate system is obtained; at least one position prediction data of the target object in the wide-area coordinate system is determined; and based on the at least one position prediction data and the second position data, the positioning position of the target object in the wide-area coordinate system is determined. Thus, by establishing the transformation relationship between local and wide-area coordinate systems, the local coordinate system positioning data can be unified to the wide-area coordinate system positioning data. Furthermore, by performing further position data prediction and fusion under the unified coordinate system, a more accurate positioning can be achieved. Attached Figure Description

[0073] Figure 1 This is a schematic diagram of the components of a positioning system;

[0074] Figure 2 A schematic diagram illustrating the overall concept of the positioning method provided in the embodiments of this application;

[0075] Figure 3 A flowchart illustrating the positioning method provided in an embodiment of this application;

[0076] Figure 4 This is a schematic diagram of a multi-branch tree structure provided in an embodiment of this application;

[0077] Figure 5 This is a schematic diagram of the structural composition of the positioning device provided in the embodiments of this application;

[0078] Figure 6 This is a schematic diagram of the structural composition of the electronic device provided in the embodiments of this application. Detailed Implementation

[0079] In order to gain a more detailed understanding of the features and technical content of the embodiments of this application, the implementation of the embodiments of this application will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for reference and illustration only and are not intended to limit the embodiments of this application.

[0080] Figure 1 This is a schematic diagram of the components of a positioning system, such as... Figure 1 As shown, the positioning system includes: a map data construction unit, an indoor ultra-wideband positioning and navigation unit, an outdoor BD / GPS positioning and navigation unit, an indoor and outdoor positioning data fusion unit, a central server unit, and a terminal display unit. The indoor ultra-wideband positioning and navigation unit is used for indoor spatial positioning using a ranging mode, the outdoor BD / GPS positioning and navigation unit is used for outdoor spatial positioning using the BeiDou satellite navigation / GPS satellite navigation system, and the indoor and outdoor positioning data fusion unit is used to determine whether the positioning tag is indoors or outdoors, and intelligently switch the positioning mode according to the determination result, thus fusing indoor and outdoor positioning methods. Figure 1 The positioning system shown can solve the problems of difficulty in locating and navigating people and objects in a large area and low accuracy. The use of ranging mode can avoid positioning deviation caused by large signal interference and mutual interference between signals.

[0081] Figure 1 The positioning system shown mainly utilizes map data construction units to construct map data by loading outdoor electronic maps and using 3D modeling to construct indoor 3D map scenes. It does not yet take into account the map layering structure and the conversion between map coordinate systems on large positioning platforms.

[0082] and, Figure 1 The positioning system, while using indoor ultra-wideband positioning data and outdoor BD / GPS data for data fusion, has not yet considered the fusion of multi-dimensional positioning data such as inertial navigation, Bluetooth, and UWB, which has certain limitations.

[0083] also, Figure 1 The positioning system uses building boundaries to switch between indoor and outdoor positioning methods, but this method does not fully consider the accuracy of various positioning data. It simply switches positioning data based on building boundaries, which does not meet the requirements of high-precision positioning.

[0084] Furthermore, Figure 1 The positioning system in the text only generates a single fusion positioning result of indoor and outdoor positioning data, without further analyzing the characteristics of travel trajectory, spatial characteristics, temporal characteristics, and path characteristics, and without using historical data for data fusion to further improve positioning accuracy.

[0085] In this embodiment, the construction of a wide-area, large-scale, high-precision positioning platform first requires solving the high-precision conversion from local coordinate systems to wide-area coordinate systems for various positioning data, involving inertial navigation coordinate systems, GPS coordinate systems, and the local coordinate systems of various sub-platforms. The technical solution of this embodiment can plan and establish the hierarchical relationship between various coordinate systems according to a multi-branch tree structure, and provides a method for high-precision conversion between various coordinate systems. Furthermore, how to use a unified wide-area coordinate system and historical data for corresponding position prediction, and then perform a second fusion of position data to further improve positioning accuracy, is also a challenge faced by wide-area positioning platforms. The technical solution of this embodiment utilizes historical data, proposes multiple position prediction methods, and employs a weighted average method for the second fusion of positioning data. Then, a neural network is used to optimize and iterate the weight values ​​of each positioning data point, effectively improving the positioning accuracy of the wide-area positioning platform. Figure 2 This is a block diagram illustrating the implementation of the positioning method provided in the embodiments of this application.

[0086] Below, based on Figure 2 The implementation block diagram of the positioning method shown is illustrated below. The positioning method provided in the embodiments of this application is described as follows.

[0087] Figure 3 A flowchart illustrating the positioning method provided in the embodiments of this application is shown below. Figure 3 As shown, the positioning method provided in this application includes the following steps:

[0088] Step 301: Construct a wide-area coordinate system for multiple local coordinate systems, and determine the transformation relationship between each local coordinate system and the wide-area coordinate system.

[0089] In a wide-area positioning platform, various real-world applications may employ their own defined coordinate systems. Furthermore, the platform incorporates multiple high-precision positioning devices, including Bluetooth AOA (Angle of Arrival) / AOD (Angle of Departure), UWB, and Real-time Kinematic (RTK), achieving centimeter-level positioning accuracy. Therefore, the construction of a large-scale, wide-area high-precision positioning platform first requires solving the high-precision conversion between local coordinate systems and wide-area coordinate systems for various positioning data. The local coordinate systems involve inertial navigation coordinate systems, GPS coordinate systems, and the local coordinate systems of each sub-platform.

[0090] In one optional embodiment of this application, the step of constructing a wide-area coordinate system for multiple local coordinate systems can be implemented in the following way:

[0091] A multi-branch tree structure is established based on the hierarchical relationship between multiple local coordinate systems; wherein, each node in the multi-branch tree structure represents a positioning coordinate system, and the root node of the multi-branch tree structure corresponds to the wide-area coordinate system; the child nodes of each node in the multi-branch tree structure are the child positioning coordinate systems of the positioning coordinate system corresponding to that node, and the parent node of that node is the parent positioning coordinate system of the positioning coordinate system corresponding to that node.

[0092] The technical solution of this application embodiment plans and establishes the layered relationship between various coordinate systems according to the multi-branch tree structure, and determines the high-precision conversion method between various coordinate systems based on the layered relationship.

[0093] In this embodiment, for a wide-area positioning platform, the first step is to establish a coordinate system data structure based on coordinate system partitioning. Specifically, this application can establish a multi-branch tree structure according to the hierarchical relationship between each positioning platform, and describe the hierarchical relationship of the coordinate systems of each positioning platform according to the tree structure. Each positioning platform coordinate system is located at a node in the tree structure, and its directly subordinate child positioning platform coordinate systems are its child nodes; the positioning platform coordinate systems of its directly connected superior nodes are its parent nodes. The wide-area positioning platform coordinate system, i.e., the wide-area coordinate system, is the root node and is a globally unified coordinate system; the directly subordinate positioning platform coordinate systems of the root node are first-level nodes. Figure 4 This is a schematic diagram of a multi-branch tree structure provided in an embodiment of this application. Figure 4 In the multi-branch tree structure shown, the root node is the wide-area positioning platform coordinate system, or simply the wide-area coordinate system, and all its child nodes are local positioning platform coordinate systems, or simply local coordinate systems.

[0094] In this embodiment, the coordinate system layering relationship of each positioning platform is described by a multi-branch tree structure, which effectively solves the technical problems of wide geographical distribution of positioning sub-platforms, mutual intersection and layering between various platforms, etc. The problem of coordinate system transformation of complex structure is transformed into the problem of coordinate transformation between nodes of coordinate system tree structure, and coordinate system management, coordinate system transformation and coordinate system unification are efficiently realized.

[0095] The technical solution of this application embodiment addresses the current situation of numerous positioning platforms with unclear and complex hierarchical structures. Based on the stacking relationship of positioning platforms, a multi-branch tree structure is established. The tree structure is used to describe the stacking relationship of the coordinate systems of each positioning platform. This can solve the problem of numerous sub-platforms and complex stacking relationships in wide-area large-scale positioning platforms, forming a standardized coordinate system architecture. It is an efficient data storage structure for coordinate system search and calculation.

[0096] In an optional embodiment of this application, the step of determining the transformation relationship between each local coordinate system in the plurality of local coordinate systems and the wide-area coordinate system can be implemented in the following way:

[0097] For each of the multiple local coordinate systems, determine the shortest path between the node corresponding to that local coordinate system in the multi-branch tree structure and the root node in the multi-branch tree structure. Based on the shortest path, obtain the path node vector corresponding to the shortest path. Based on the coordinate system transformation relationship between each adjacent node in the path node vector, determine the transformation relationship between the local coordinate system and the wide-area coordinate system.

[0098] Specifically, in the established multi-branch tree structure of the coordinate system, for paths between non-adjacent nodes N and M, the Tarjan algorithm can be used to search for the shortest path, thereby forming a path node vector {N, a1, a2, ..., a...}. k , M}, where a i (i = 1, 2, ..., k) are intermediate path nodes. In the subsequent coordinate system transformation process, the coordinate system transformation between adjacent nodes is first realized, and then the local coordinate system transformation can be carried out step by step using the path node vector, finally completing the coordinate transformation between N and M.

[0099] This application utilizes the Tarjan algorithm to calculate the shortest path between nodes, which can efficiently solve the problem of coordinate transformation between stacked local coordinate systems.

[0100] In an optional embodiment of this application, the step of determining the transformation relationship between each local coordinate system and the wide-area coordinate system can be implemented in the following way:

[0101] For each of the multiple local coordinate systems, the transformation equation between the local coordinate system and the wide area coordinate system is established using the Mologinsky seven-parameter model.

[0102] Based on the location data of the target object at multiple identical locations in the local coordinate system and the wide-area coordinate system, determine the coordinate values ​​of each location of the target object at the multiple identical locations in the local coordinate system and the wide-area coordinate system.

[0103] The seven transformation parameters in the Mologinsky seven-parameter model are determined based on the coordinate values ​​of the target object at each of the multiple identical locations in the local coordinate system and the wide-area coordinate system.

[0104] The coordinate transformation formula between the local coordinate system and the wide-area coordinate system is determined based on the seven transformation parameters.

[0105] Specifically, in a wide-area positioning platform, the direction of the coordinate axes of the wide-area coordinate system is set to determine the wide-area coordinate system. The wide-area coordinate system is located at the root node of the coordinate system's multi-branch tree structure. Each level of node directly subordinate to the root node uses its own local coordinate system. For each local coordinate system, a transformation equation between the wide-area coordinate system and the Molodinsky seven-parameter model is established. By obtaining coordinate values ​​in different coordinate systems at the same locations using measurements from at least three identical locations in both the local and wide-area coordinate systems, the coordinate values ​​at those locations are obtained. Then, using the Gauss-Newton method, a nonlinear least squares regression fitting method is performed. An iterative algorithm is used to solve for the seven transformation parameters of the Molodinsky seven-parameter model, thereby establishing the coordinate transformation formula from the local coordinate system to the wide-area coordinate system, realizing the transformation from the local coordinate system to the wide-area coordinate system.

[0106] Specifically, the Mologinsky seven-parameter model is a three-dimensional affine transformation. If the local coordinate system is defined as the source coordinate system and the wide-area coordinate system as the target coordinate system, for a certain location point, the coordinates of that location point in the local coordinate system are (X... A ,Y A Z A This coordinate value is called the source coordinate; the coordinates of this location point in the wide-area coordinate system are (X... B ,Y B Z B This coordinate value is called the target coordinate. A transition point P is introduced between the source and target coordinates; the coordinates of point P are (X...). P ,Y P Z P Specifically, we can take the coordinates of point P as the geocentric origin (0,0,0), and first translate the source coordinates to the transition point P. Then, the spatial rectangular coordinate system with point P as the origin is translated, rotated, and scaled to match the target coordinate system. The specific formula is as follows:

[0107]

[0108] In the above formula, if ω X ω Y ω Z Since all angles are small, sinω≈ω and cosω≈1. Thus, we can obtain the following formula:

[0109]

[0110]

[0111] In the above formula, T X T Y T Z ω is the translation parameter. X ω Y ωZ is the rotation parameter, and m is the scale parameter.

[0112] The translation, rotation, and scaling parameters in the above formula are the seven parameters in the Mologinsky seven-parameter model. By obtaining these seven parameters, the coordinate transformation formula between the local coordinate system and the wide-area coordinate system can be obtained.

[0113] In this embodiment, for the inertial navigation vehicle, its accelerometer has its own coordinate system, and during the motion and rotation of the vehicle, its own coordinate system will continuously change relative to the local coordinate system. This embodiment first uses the angular velocity measured by the gyroscope and calculates the attitude vector R using the quaternion method; using the attitude vector R, the local observation values ​​observed by the accelerometer are transformed into the local coordinate system; then, using Newton's kinematics principle, the velocity and position update equations are calculated, thereby iteratively calculating the real-time position of the inertial navigation vehicle based on the local coordinate system, realizing the transformation between the inertial navigation coordinate system and the local coordinate system; finally, the Mologinsky seven-parameter model is used to realize the transformation from the local coordinate system to the wide-area coordinate system.

[0114] For the GPS coordinate system, the Gauss-Kruger orthographic projection method can be used to project the latitude and longitude of GPS onto a plane rectangular coordinate system to form its own local coordinate system. Then, using the Mologinsky seven-parameter model, the seven transformation parameters from the GPS coordinate system to the wide area coordinate system can be calculated, thus realizing the transformation from the GPS coordinate system to the wide area coordinate system.

[0115] Step 302: Based on the first position data of the target object in one or more local coordinate systems among the multiple local coordinate systems and the transformation relationship, obtain the second position data of the target object in the wide-area coordinate system.

[0116] In this embodiment of the application, after obtaining the transformation relationship between the local coordinate system and the wide-area coordinate system, the positioning data of the target object in the wide-area coordinate system can be determined by using the determined transformation relationship between the local coordinate system and the wide-area coordinate system, based on the measured positioning data of the target object in a certain local coordinate system.

[0117] In this application, the first location data is the positioning data of the target object in a local coordinate system at the first moment, and the second location data is the positioning data of the target object in a wide-area coordinate system at the first moment. Here, the first moment is the current moment.

[0118] Step 303: Determine at least one predicted location of the target object in the wide-area coordinate system.

[0119] In one optional embodiment of this application, the at least one position prediction data includes one or more of the following position prediction data: first position prediction data determined based on the target object's inertial navigation history data, second position prediction data determined based on the target object's historical position data, third position prediction data determined based on the target object's motion direction data, and fourth position prediction data determined based on the target object's historical trajectory.

[0120] In this embodiment, during the position data prediction process, under a unified coordinate system, the inherent characteristics, kinematic laws, spatial features of motion, and geometric features of the trajectory of the inertial navigation device are comprehensively utilized. Based on historical data, various mathematical tools such as data interpolation and data regression fitting are employed to predict the position point. This mainly includes: using velocity and acceleration in inertial navigation to predict the direction of motion of the inertial navigation device; using Kalman filtering to predict spatial position; calculating and fitting velocity and direction using historical data, and using kinematic laws to predict the position point; and predicting the latest geometric trajectory point through multivariate quadratic regression of the geometric trajectory. Subsequently, based on these four methods of position prediction, comprehensive calculations are performed from the perspectives of the inertial navigation device, time, space, and trajectory to fit the latest position data, thereby improving the accuracy of position calculation for the target object.

[0121] In an optional embodiment of this application, the first position prediction data can be determined through the following steps:

[0122] The first acceleration vector in the inertial navigation data at the first moment is obtained, and the acceleration components of the first acceleration vector at the first moment are calculated in the coordinate system corresponding to the acceleration sensor of the target object. The local observation value observed by the acceleration sensor is transformed into the second acceleration vector in the local coordinate system of the target using the attitude vector.

[0123] Acquire historical acceleration data from multiple moments prior to the first moment, and construct a third acceleration vector from the historical acceleration data based on the chronological order of the historical acceleration data.

[0124] The acceleration fitting formula corresponding to the third acceleration vector and the first acceleration difference fitting formula are obtained by using the polynomial fitting method.

[0125] Determine the first coefficient corresponding to the acceleration fitting formula and the second coefficient corresponding to the first acceleration difference fitting formula, and obtain the first functional relationship corresponding to the acceleration fitting formula based on the first coefficient, and obtain the second functional relationship corresponding to the first acceleration difference fitting formula based on the second coefficient;

[0126] The second acceleration difference sequence is calculated using the first functional relationship;

[0127] Determine the first mean difference value corresponding to the first acceleration difference value sequence and the second mean difference value corresponding to the second acceleration difference value sequence;

[0128] The target acceleration fitting formula is calculated by selecting the smaller of the first and second mean difference acceleration sequences.

[0129] Based on the target acceleration fitting formula, the predicted first position data corresponding to the first moment is determined.

[0130] In this embodiment of the application, the first position prediction data is the prediction of the position of the target object using inertial navigation direction.

[0131] For user inertial navigation data, the acceleration vector 'a' represents the various acceleration components in its own inertial frame at the current moment. Using the attitude vector R, the local observations from the accelerometer are first transformed into a local coordinate system. The transformation equation is A = R * a, where 'a' is the acceleration vector measured by the inertial navigation vehicle using its own accelerometer, and A is the acceleration vector in the local coordinate system.

[0132] Extract the historical data of acceleration A from the N time steps before the current time step to form an acceleration vector A arranged in time series. T ={A1,A2...A N The acceleration curve A(t) is calculated using polynomial fitting. However, polynomial fitting can result in underfitting and overfitting, leading to significant errors in the fitted curve. Therefore, in this embodiment, two polynomial fitting formulas are first used: one for acceleration and one for the acceleration itself. And for the acceleration difference △A T ={A2-A1,A3-A2...A N -A N-1 Fitting formula for} Where n is the polynomial order, t is time, and a i b i Let be the polynomial coefficients. The polynomial coefficients 'a' are calculated using fitting. i b i This yields the functional relationships between A1(t) and ΔA1(t). Further, A1(t) is used to calculate A... T ={A1,A2...A N The difference between two adjacent time points is ΔA2(t) = A1(t+T) - A(t); then the differences between ΔA1(t) and ΔA2(t) at time A are calculated. T ={A1,A2...AN Let avg(|△A1(t)|) and avg(|△A2(t)|) be the mean absolute values ​​of} at N time points. A fitting method with a larger mean indicates a steeper slope of the fitted curve at the sampling points, faster data change, and lower accuracy. Therefore, a fitting method with a smaller mean absolute value can be selected for subsequent calculations. When the A1(t) fitting method is selected, the acceleration formula is A(t) = A1(t); when the △A2(t) fitting method is selected, the acceleration formula is A(t) = A(tT) + △A2(t).

[0133] By substituting A(t) into the current time T, the acceleration at the current time point can be calculated. Therefore, using Newton's equations of motion, the current position can be predicted as X1 = X0 + v0ΔT + A. T △T 2 / 2, where X0 and v0 are the position and velocity of the previous time point, and △T is the time interval between the current time point and the previous time point.

[0134] For inertial navigation direction prediction, this application starts with the traditional polynomial fitting method. First, two polynomial fitting methods are used to calculate the fitting curve. Then, based on the change in the curve slope, the fitting method with higher fitting quality and accuracy is selected, thereby avoiding underfitting and overfitting and improving the accuracy of data prediction.

[0135] In an optional embodiment of this application, the second location prediction data can be determined through the following steps:

[0136] Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system.

[0137] The Kalman filter method is used to predict the second location prediction data corresponding to the first time point based on the multiple historical location data.

[0138] In this embodiment of the application, the second location prediction data is the prediction of the current location of the target object using historical location data.

[0139] Specifically, in this embodiment, historical location data prior to the current time point is used, along with a Kalman filter method and spatial prediction, to predict the location data at the current time. The Kalman filter time update equation is X2 = F t x t-1 +B t u t-1 P t|t-1 =F t P t-1 F t T+Q t Where X2 is the location prediction result, x t-1 F represents the optimal prediction result from the previous moment. t Let F be the state transition matrix. t T For F t Matrix transpose, B t u t-1 Parameters introduced for the system parameter model; P t|t-1 Let P be the covariance matrix of the prediction result at time t based on time t-1. t-1 Let Q be the covariance matrix at time t-1. t This is the covariance matrix resulting from system process disturbances.

[0140] To perform iterative calculations of the Kalman equations, the Kalman filter state update equations need to be established as follows:

[0141] x t =x t|t-1 +K t (w t -H t x t|t-1 ), P t =(IK t H t )P t|t-1

[0142] Where K t =P t|t-1 H t T / (H t P t|t-1 H t T +R t ) represents the Kalman gain, w t The measured value is the position calculated at time t, H. t Let R be the transformation matrix. t Let I be the covariance matrix of the measured values, and let I be the identity matrix.

[0143] In the calculation process, the Kalman gain K is calculated first. t Using K t As weights, a weighted average is calculated between the predicted and observed results to obtain the state estimate for the current time step. The calculated x... t P t Then, the process returns to step one for iterative calculation to predict the next position, thus iteratively generating the current time point prediction result X2 after Kalman filtering.

[0144] In an optional embodiment of this application, the third position prediction data can be determined through the following steps:

[0145] Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system.

[0146] Based on the historical location data, determine the acceleration and velocity at a time preceding the first time.

[0147] Based on the target object's motion direction, as well as the acceleration and velocity predictions, the third position prediction data corresponding to the first moment is obtained.

[0148] In this embodiment of the application, the third location prediction data is the location prediction data obtained by predicting the movement direction of the target object using historical location data.

[0149] Specifically, this embodiment of the application uses N historical position data points prior to the current time to calculate the acceleration and velocity at the previous time point, and uses the direction of motion to predict the position data at the current time point. During the prediction process, historical position data from the N points prior to the current time point are first extracted to form a position vector X. N ={X1, X2, ..., X} N}, calculate the velocity of two adjacent points to form a velocity vector V = {v i =(X i+1 -X i ) / T i}, and then calculate the acceleration vector A = {a i =(V i+1 -V i ) / T i For the velocity and acceleration vectors, piecewise curve fitting using Hermite interpolation is employed to generate smooth velocity and acceleration curves, thereby enabling the prediction of the position data at the current time point. The calculation steps are as follows:

[0150] For piecewise fitting of velocity V, the velocity vector V is used, and for two adjacent data points V... i V i+1 Perform two-point cubic Hermite piecewise interpolation, using a cubic function H. i (t)=a0+a1t+a2t 2 +a3t 3 Where t is the time variable. In the Hermite piecewise interpolation process, H... i The function (t) satisfies:

[0151] 1. V i V i+1 In function H i (t)

[0152] 2. To satisfy the smoothness of the function, at the endpoint V i Above, satisfying H i (t) derivative and H i-1 The derivatives of (t) are equal at the endpoint V. i+1 Above, satisfying H i (t) derivative and H i+1 The derivatives of (t) are equal, thus satisfying the smoothness requirement of the fitted function.

[0153] Using the two types of conditions mentioned above, i.e., obtaining four known conditions, we substitute them into the cubic function H. i (t)=a0+a1t+a2t 2 +a3t 3 The undetermined coefficients a are obtained by solving the problem piecewise. i (i=0,1,2,3), then (V1,V2)......(V N-1 V N The cubic polynomial fitting curve H between points i (t)(i=1,2,....N-1). Here, V has not yet been obtained. N Predicted curve H after point N (t), continue the prediction calculation, H N The condition that (t) satisfies is:

[0154] 1. V N-2 V N-1 V N Three coordinate values, on curve H N (t) on;

[0155] 2. H N-1 (t), H N (t) has the same derivative value at the Nth point, which satisfies the smoothness requirement of the function.

[0156] Substitute the above four conditions into H N (t)=a0+a1t+a2t 2 +a3t 3 H can then be solved. N The polynomial coefficients a of (t) i (i=0,1,2,3), thus realizing the determination of the parameters of the velocity prediction equation.

[0157] Substitute the current time T into H. N (t) polynomial, and then calculate the velocity prediction value V at the Nth point. T =H N (T). Using the same calculation method, the predicted acceleration value A at the Nth point can also be calculated. T .

[0158] According to Newton's kinematics formulas, the current position coordinates X3 = X0 + V can be predicted and calculated. T △T+A T △T 2 / 2, where X0 is the position of the previous time point, and △T is the time difference between the current time point and the previous time point.

[0159] In an optional embodiment of this application, the fourth position prediction data can be determined through the following steps:

[0160] Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system; each historical location data point is multi-dimensional data.

[0161] Based on the historical location data at the multiple times, a polynomial regression function equation is constructed to represent the change of the target object's position over time.

[0162] Based on the polynomial regression function equation, the fourth position prediction data corresponding to the first time point of the target object is determined.

[0163] In this embodiment of the application, the fourth location prediction data is the prediction of the location of the target object using geometric trajectory regression prediction.

[0164] Specifically, in this embodiment, using N historical position data points preceding the current point, a discrete geometric trajectory can be depicted in two-dimensional or three-dimensional space. Considering that the changes along each axis of the two-dimensional or three-dimensional coordinate system in the trajectory description possess both independence and correlation, the correlation between the coordinate axes during object movement must be considered when establishing the geometric trajectory regression model. Furthermore, the new position point and historical data points also have a certain correlation during the object's movement. This embodiment comprehensively considers the coordinate changes of each coordinate axis itself, as well as the correlations between coordinate axes and between historical data points, using a three-dimensional multivariate polynomial quadratic regression model to achieve geometric trajectory regression prediction.

[0165] Set the object's three-dimensional coordinates as x i (i = 1, 2, 3), the polynomial regression function equation for the position coordinates changing with time is: Where △T is the time interval between data points, a i,k b i,k For the regression coefficients to be solved, The regression part is its own coordinate axis. This is the regression component that represents the interrelationship between historical points and coordinate axes. In the regression formula, f... k Since f(t) is an unknown function, we first need to calculate f using regression.k (t), whose polynomial regression equation is: in, To traverse all 1+r2+r3＝r The combination The product term represents the interrelationship between the various coordinate axes, c r,s f represents the coefficients of the polynomial to be determined. k (t) can be rewritten as Among them, A s For coordinate product term, d s Let f be the polynomial coefficients to be regressed. k (t) has been transformed into polynomial form, so the coefficients d can be calculated using a polynomial regression model. s Then calculate c r,s Complete f k Regression calculation of (t).

[0166] f k (t) Substitute The polynomial coefficients 'a' can be calculated using a polynomial regression model and quadratic regression. i,k b i,k Complete x i The multiple regression calculation of (t) is performed. Using the regression function, the predicted position of the current point X4={x1(t),x2(t),x3(t)} is calculated.

[0167] In the geometric trajectory regression prediction, this application establishes a 3D regression prediction model and uses the 3D model for rapid calculation to obtain prediction parameters, effectively improving the speed and accuracy of geometric trajectory prediction.

[0168] Step 304: Determine the location of the target object in the wide-area coordinate system based on the at least one location prediction data and the second location data.

[0169] In this embodiment, based on the initial fusion of positioning data from multiple sensors already performed by the positioning platform, multiple methods of location data prediction are performed using the initial fusion data from multiple historical location points under a unified coordinate system. This enables a secondary fusion of the initial fused location data and the predicted data from multiple methods to calculate and obtain a more accurate location for the user.

[0170] In a wide-area positioning platform, location data from various sensors and location acquisition devices are distributed, forming a data mart with multiple data sources, such as inertial navigation, Bluetooth, Wi-Fi, UWB, GPS / BeiDou, and carrier base station data. For the same user, multiple positioning devices are also deployed, and the positioning data from each device is uploaded to the wide-area positioning platform. Multi-sensor data fusion, known as primary data fusion, typically employs methods such as Kalman filtering and particle filtering to fuse the multi-sensor data.

[0171] The technical solution of this application embodiment, based on the initial fusion of positioning data from multiple sensors by the positioning platform, utilizes the initial fusion data from multiple historical location points to perform location data prediction in various ways. Then, it performs a secondary data fusion of the initially fused location data and the predicted data from multiple methods to calculate and obtain a more accurate user location. During the secondary data fusion, a neural network is used to provide feedback correction to the weights used in the prediction process, thereby significantly improving positioning accuracy.

[0172] For the weighted location data secondary fusion process, in the wide-area positioning platform, the current location X of the object is calculated by first fusing multi-sensor data. Simultaneously, based on the historical location data of the previous N points, the predicted locations X1, X2, X3, and X4 of the current point are calculated. Then, a weighted average method is used to perform secondary fusion of the data to form the final location data. Where a is the weight of the current solution position, a i The weights for each predicted position.

[0173] When performing weighted location data fusion, different weights are applied to the currently collected location data and four types of predicted data. The weight values ​​can be set based on experience and the accuracy of various predictions. Considering the requirements of high-precision positioning platforms, this embodiment employs a neural network model to correct each weight value, achieving dynamic adjustment of the weight values, thereby effectively improving the positioning accuracy of the wide-area positioning platform.

[0174] Specifically, in an optional embodiment of this application, step 304 above can be implemented through the following steps:

[0175] The weights corresponding to each location data in the at least one location data and the weights corresponding to the second location data are determined using a neural network model.

[0176] Based on each location prediction data in the at least one location prediction data, each weight corresponding to each location prediction data, the second location data, and the weight corresponding to the second location data, the location fusion data of the target object is obtained, and the location fusion data is determined as the positioning position of the target object in the wide-area coordinate system.

[0177] The technical solution of this application embodiment can use historical data to predict the inertial navigation direction, spatial direction, motion direction, and geometric trajectory of the latest location. It can also use the predicted data and the currently collected location data to perform secondary data fusion using a weighted average method, which effectively improves the positioning accuracy of the wide-area positioning platform.

[0178] In an optional embodiment of this application, the weights corresponding to each location data in at least one location data set and the weights corresponding to the second location data set can be determined in the following manner:

[0179] N experimental samples are obtained, the N experimental samples include N third position data of the target object in the wide-area coordinate system and position prediction data corresponding to each of the N third position data; each of the N third position data and the position prediction data corresponding to that third position data corresponds to one experimental sample;

[0180] Construct an M-dimensional dataset based on the N experimental samples;

[0181] Calculate the covariance matrix corresponding to the M-dimensional data set, and determine the M-1 eigenvalues ​​corresponding to the covariance matrix;

[0182] Determine the M-1 eigenvalues ​​corresponding to the M-dimensional eigenvectors, and obtain a mapping matrix based on the M-dimensional eigenvectors; then, based on the mapping matrix, reduce the dimensionality of the M-dimensional data set to an M-1 dimensional data set.

[0183] The M-1 dimension dataset is normalized to obtain the target dataset;

[0184] The neural network model is trained using the target dataset to obtain the neural network model.

[0185] Specifically, a neural network model is used to adjust each weight value, and the dynamic adjustment process of the weight values ​​is as follows:

[0186] Create a three-layer backpropagation (BP) neural network model, including an input layer, a hidden layer, and an output layer. The input layer has 4 neurons, and the output layer has 5 neurons.

[0187] Given the current positioning data and four types of positioning prediction data, N experimental samples are formed into a calibrated five-dimensional dataset S. The first dimension is the positioning data after one data fusion; the second dimension is the inertial navigation direction prediction data; the third dimension is the spatial prediction data; the fourth dimension is the motion direction prediction data; and the fifth dimension is the geometric trajectory fitting prediction data. Using the five-dimensional dataset S, the covariance matrix K of S is calculated, and the four eigenvalues ​​of K are solved and arranged in descending order as γ1, γ2, γ3, and γ4. Then, the corresponding eigenvalues ​​γ of the covariance matrix K are solved. i The five-dimensional normalized eigenvector α i (i = 1, 2, 3, 4), forming a mapping matrix F = {α1, α2, α3, α4}. This mapping matrix allows the five-dimensional dataset S to be reduced in dimension and mapped to the labeled four-dimensional dataset I, i.e., I = S * F. Then, dataset I is normalized to obtain a dataset J with a mean of 0 and a variance of 1. The resulting dataset J can then be input into the neural network model for training and testing.

[0188] Using Gaussian distributed random numbers, initialize all values ​​and thresholds of the neural network model, initialize the learning rate to 0.01, and set the training minimum mean square error target to 10. -3 The minimum performance gradient during training is set to 10. -6 The maximum number of training iterations is 100,000, and the training algorithm chosen is Bayesian regularization. The neural network model is trained using real-world data to determine all values ​​and thresholds that provide generalization capability that meets practical requirements. By adjusting the error and conducting multiple experiments, the neural network model is trained to improve accuracy while avoiding overfitting. The activation function used is... The training of a neural network model is completed when any one of the following conditions is met: the training mean square error reaches the minimum training mean square error target, the training performance gradient reaches the minimum training performance gradient, or the number of training iterations reaches the maximum number of training iterations.

[0189] By using a neural network model that has been trained and optimized multiple times, the weights of each predicted data can be iteratively optimized.

[0190] The technical solution of this application, based on a multi-branch tree structure, plans and establishes the hierarchical relationship between coordinate systems, and provides high-precision conversion methods from various local coordinate systems to wide-area coordinate systems. Secondly, utilizing historical data, four location prediction methods are proposed, and a weighted average method is used for secondary fusion of positioning data. A neural network model is then used to optimize and iterate the weighted values, effectively improving the positioning accuracy of the wide-area positioning platform.

[0191] This application also provides a positioning device. Figure 5This is a schematic diagram of the structural composition of the positioning device provided in the embodiments of this application, as shown below. Figure 5 As shown, the device includes:

[0192] The construction unit 501 is used to construct a wide-area coordinate system for multiple local coordinate systems and determine the transformation relationship between each local coordinate system and the wide-area coordinate system.

[0193] Processing unit 502 is used to obtain second position data of the target object in the wide-area coordinate system based on the first position data of the target object in one or more local coordinate systems among the plurality of local coordinate systems and the transformation relationship;

[0194] The first determining unit 503 is used to determine at least one position prediction data of the target object in the wide-area coordinate system;

[0195] The second determining unit 504 is used to determine the positioning position of the target object in the wide-area coordinate system based on the at least one position prediction data and the second position data.

[0196] In an optional embodiment of this application, the construction unit 501 is specifically used to: establish a multi-branch tree structure based on the hierarchical relationship between multiple local coordinate systems; wherein, each node in the multi-branch tree structure represents a positioning coordinate system, the root node of the multi-branch tree structure corresponds to the wide-area coordinate system; the child node of each node in the multi-branch tree structure is a sub-positioning coordinate system of the positioning coordinate system corresponding to that node, and the parent node of that node is the superior positioning coordinate system of the positioning coordinate system corresponding to that node.

[0197] In an optional embodiment of this application, the construction unit 501 is further specifically used for: for each of the plurality of local coordinate systems, determining the shortest path between the node corresponding to the local coordinate system in the multi-branch tree structure and the root node in the multi-branch tree structure; obtaining the path node vector corresponding to the shortest path based on the shortest path; and determining the transformation relationship between the local coordinate system and the wide-area coordinate system based on the coordinate system transformation relationship between each adjacent node in the path node vector.

[0198] In an optional embodiment of this application, the construction unit 501 is further specifically used to: for each of the multiple local coordinate systems, establish the transformation equation between the local coordinate system and the wide area coordinate system using the Mologinski seven-parameter model;

[0199] Based on the location data of the target object at multiple identical locations in the local coordinate system and the wide-area coordinate system, determine the coordinate values ​​of each location of the target object at the multiple identical locations in the local coordinate system and the wide-area coordinate system.

[0200] The seven transformation parameters in the Mologinsky seven-parameter model are determined based on the coordinate values ​​of the target object at each of the multiple identical locations in the local coordinate system and the wide-area coordinate system.

[0201] The coordinate transformation formula between the local coordinate system and the wide-area coordinate system is determined based on the seven transformation parameters.

[0202] In one optional embodiment of this application, the at least one position prediction data includes one or more of the following position prediction data: first position prediction data determined based on the target object's inertial navigation history data, second position prediction data determined based on the target object's historical position data, third position prediction data determined based on the target object's motion direction data, and fourth position prediction data determined based on the target object's historical trajectory.

[0203] In an optional embodiment of this application, the first determining unit 503 is specifically configured to: acquire a first acceleration vector from the inertial navigation data at a first moment; calculate each acceleration component of the first acceleration vector at the first moment in the coordinate system corresponding to the acceleration sensor of the target object; convert the local observation value observed by the acceleration sensor into a second acceleration vector in the local coordinate system of the target object using an attitude vector; acquire historical acceleration data from multiple moments prior to the first moment; construct a third acceleration vector from the historical acceleration data based on the chronological order of the historical acceleration data; fit an acceleration fitting formula corresponding to the third acceleration vector and a first acceleration difference fitting formula using a polynomial fitting method; and determine the... The method involves defining the first coefficient corresponding to the acceleration fitting formula and the second coefficient corresponding to the first acceleration difference fitting formula, obtaining the first functional relationship corresponding to the acceleration fitting formula based on the first coefficient, and obtaining the second functional relationship corresponding to the first acceleration difference fitting formula based on the second coefficient; calculating the second acceleration difference sequence using the first functional relationship; determining the first mean difference value corresponding to the first acceleration difference sequence and the second mean difference value corresponding to the second acceleration difference sequence; selecting the smaller mean difference value acceleration sequence between the first mean difference value and the second mean difference value to calculate the target acceleration fitting formula; and determining the first position prediction data corresponding to the first time point based on the target acceleration fitting formula.

[0204] In an optional embodiment of this application, the first determining unit 503 is specifically used to: acquire historical position data of multiple times prior to the first time; the historical position data is the historical position data of the target object in the wide-area coordinate system; and use the Kalman filter method to predict second position prediction data corresponding to the first time based on the multiple historical position data.

[0205] In an optional embodiment of this application, the first determining unit 503 is specifically used for: acquiring historical position data of multiple times prior to the first time; the historical position data being the historical position data of the target object in the wide-area coordinate system; determining the acceleration and velocity of a time prior to the first time based on the historical position data; and predicting third position prediction data corresponding to the first time based on the motion direction of the target object and the acceleration and velocity.

[0206] In an optional embodiment of this application, the first determining unit 503 is specifically used for:

[0207] Historical position data from multiple time points prior to the first time point are acquired; the historical position data refers to the historical position data of the target object in the wide-area coordinate system; each historical position data is multidimensional data; a polynomial regression function equation for the position of the target object over time is constructed based on the historical position data from the multiple time points; and a fourth position prediction data for the target object corresponding to the first time point is determined based on the polynomial regression function equation.

[0208] In an optional embodiment of this application, the second determining unit 504 is specifically used to: determine the weights corresponding to each location data in the at least one location data and the weights corresponding to the second location data using a neural network model; obtain the location fusion data of the target object based on each location prediction data in the at least one location prediction data, each weight corresponding to each location prediction data, the second location data and the weights corresponding to the second location data, and determine the location fusion data as the positioning position of the target object in the wide-area coordinate system.

[0209] In an optional embodiment of this application, the second determining unit 504 is specifically configured to: obtain N experimental samples, the N experimental samples including N third position data of the target object in the wide-area coordinate system and position prediction data corresponding to each of the N third position data; each of the N third position data and the position prediction data corresponding to that third position data corresponds to one experimental sample; construct an M-dimensional data set based on the N experimental samples; calculate the covariance matrix corresponding to the M-dimensional data set and determine the M-1 eigenvalues ​​corresponding to the covariance matrix; determine the M-dimensionalized eigenvectors corresponding to the M-1 eigenvalues, and obtain a mapping matrix based on the M-dimensionalized eigenvectors; reduce the dimensionality of the M-dimensional data set to an M-1 dimensional dataset based on the mapping matrix; normalize the M-1 dimensional dataset to obtain a target dataset; and train the neural network model to be trained using the target dataset to obtain the neural network model.

[0210] Those skilled in the art should understand that Figure 5 The functions of each unit in the positioning device shown can be understood by referring to the relevant description of the positioning method mentioned above. Figure 5 The functions of each unit in the positioning device shown can be implemented by a program running on a processor or by specific logic circuits.

[0211] This application also provides an electronic device. Figure 6 This is a schematic diagram of the hardware structure of the electronic device according to an embodiment of this application, such as... Figure 6 As shown, the electronic device includes: a communication component 603 for data transmission, at least one processor 601, and a memory 602 for storing computer programs capable of running on the processor 601. The various components in the terminal are coupled together via a bus system 604. It is understood that the bus system 604 is used to implement communication between these components. In addition to a data bus, the bus system 604 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 6 The general designated all buses as Bus System 604.

[0212] Wherein, when the processor 601 executes the computer program, it performs at least the following: Figure 3 The steps of the method in the text.

[0213] It is understood that memory 602 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); magnetic surface memory can be disk storage or magnetic tape storage. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memory 602 described in the embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0214] The methods disclosed in the embodiments of this application can be applied to or implemented by processor 601. Processor 601 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 601 or by instructions in software form. The processor 601 may be a general-purpose processor, DSP, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 601 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. A general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, which is located in memory 602. Processor 601 reads the information in memory 602 and, in conjunction with its hardware, completes the steps of the aforementioned method.

[0215] In an exemplary embodiment, the electronic device may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), FPGAs, general-purpose processors, controllers, MCUs, microprocessors, or other electronic components to perform the aforementioned travel method.

[0216] This application also provides a computer-readable storage medium storing a computer program thereon, characterized in that the program, when executed by a processor, is at least used to perform... Figure 3 The steps of the method are shown. The computer-readable storage medium may specifically be a memory. The memory may be, for example... Figure 6 The memory 602 shown.

[0217] The technical solutions described in the embodiments of this application can be combined arbitrarily without conflict.

[0218] In the several embodiments provided in this application, it should be understood that the disclosed methods and smart devices can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0219] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0220] In addition, each functional unit in the various embodiments of this application can be integrated into a second processing unit, or each unit can be a separate unit, or two or more units can be integrated into a unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0221] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A positioning method, characterized in that, The method includes: A multi-branch tree structure is established based on the hierarchical relationship between multiple local coordinate systems, and the transformation relationship between each local coordinate system and the wide-area coordinate system is determined; wherein, the root node of the multi-branch tree structure corresponds to the wide-area coordinate system of the local coordinate system. The second position data of the target object in the wide-area coordinate system is obtained based on the first position data of the target object in one or more local coordinate systems among the multiple local coordinate systems and the transformation relationship; Determine at least one predicted position of the target object in the wide-area coordinate system; The location of the target object in the wide-area coordinate system is determined based on the at least one location prediction data and the second location data; Wherein, determining the location of the target object in the wide-area coordinate system based on the at least one location prediction data and the second location data includes: The weights corresponding to each location data in the at least one location data and the weights corresponding to the second location data are determined using a neural network model. Based on each location prediction data in the at least one location prediction data, each weight corresponding to each location prediction data, the second location data, and the weight corresponding to the second location data, the location fusion data of the target object is obtained, and the location fusion data is determined as the positioning position of the target object in the wide-area coordinate system.

2. The method according to claim 1, characterized in that, Each node in the multi-branch tree structure represents a positioning coordinate system. The child nodes of each node in the multi-branch tree structure are the child positioning coordinate systems of the positioning coordinate system corresponding to that node, and the parent node of that node is the parent positioning coordinate system of the positioning coordinate system corresponding to that node.

3. The method according to claim 2, characterized in that, Determining the transformation relationship between each local coordinate system in the plurality of local coordinate systems and the wide-area coordinate system includes: For each of the multiple local coordinate systems, determine the shortest path between the node corresponding to that local coordinate system in the multi-branch tree structure and the root node in the multi-branch tree structure. Based on the shortest path, obtain the path node vector corresponding to the shortest path. Based on the coordinate system transformation relationship between each adjacent node in the path node vector, determine the transformation relationship between the local coordinate system and the wide-area coordinate system.

4. The method according to claim 2, characterized in that, Determining the transformation relationship between each local coordinate system in the plurality of local coordinate systems and the wide-area coordinate system includes: For each of the multiple local coordinate systems, the transformation equation between the local coordinate system and the wide area coordinate system is established using the Mologinsky seven-parameter model. Based on the location data of the target object at multiple identical locations in the local coordinate system and the wide-area coordinate system, determine the coordinate values ​​of each location of the target object at the multiple identical locations in the local coordinate system and the wide-area coordinate system. The seven transformation parameters in the Mologinsky seven-parameter model are determined based on the coordinate values ​​of the target object at each of the multiple identical locations in the local coordinate system and the wide-area coordinate system. The coordinate transformation formula between the local coordinate system and the wide-area coordinate system is determined based on the seven transformation parameters.

5. The method according to claim 1, characterized in that, The at least one position prediction data includes one or more of the following position prediction data: first position prediction data determined based on the target object's inertial navigation history data, second position prediction data determined based on the target object's historical position data, third position prediction data determined based on the target object's motion direction data, and fourth position prediction data determined based on the target object's historical trajectory.

6. The method according to claim 5, characterized in that, The step of determining at least one predicted position of the target object in the wide-area coordinate system includes: The first acceleration vector in the inertial navigation data at the first moment is obtained, and the acceleration components of the first acceleration vector at the first moment are calculated in the coordinate system corresponding to the acceleration sensor of the target object. The local observation value observed by the acceleration sensor is transformed into the second acceleration vector in the local coordinate system of the target using the attitude vector. Acquire historical acceleration data from multiple moments prior to the first moment, and construct a third acceleration vector from the historical acceleration data based on the chronological order of the historical acceleration data. The acceleration fitting formula corresponding to the third acceleration vector and the first acceleration difference fitting formula are obtained by using the polynomial fitting method. Determine the first coefficient corresponding to the acceleration fitting formula and the second coefficient corresponding to the first acceleration difference fitting formula, and obtain the first functional relationship corresponding to the acceleration fitting formula based on the first coefficient, and obtain the second functional relationship corresponding to the first acceleration difference fitting formula based on the second coefficient; The acceleration difference between adjacent moments is calculated based on the historical acceleration data to obtain the first acceleration difference sequence; The second acceleration difference sequence is calculated using the first functional relationship; Determine the first mean difference value corresponding to the first acceleration difference value sequence and the second mean difference value corresponding to the second acceleration difference value sequence; The target acceleration fitting formula is calculated by selecting the smaller of the first and second mean difference acceleration sequences. Based on the target acceleration fitting formula, the predicted first position data corresponding to the first moment is determined.

7. The method according to claim 5, characterized in that, Determining the predicted location data of the target object in the wide-area coordinate system includes: Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system. The Kalman filter method is used to predict the second location prediction data corresponding to the first time point based on the historical location data at the multiple time points.

8. The method according to claim 5, characterized in that, Determining the predicted location data of the target object in the wide-area coordinate system includes: Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system. Based on the historical location data, determine the acceleration and velocity at a time preceding the first time. Based on the target object's motion direction, as well as the acceleration and velocity predictions, the third position prediction data corresponding to the first moment is obtained.

9. The method according to claim 5, characterized in that, The step of determining at least one predicted position of the target object in the wide-area coordinate system includes: Acquire historical location data from multiple time points prior to the first time point; the historical location data refers to the historical location data of the target object in the wide-area coordinate system; each historical location data point is multi-dimensional data. Based on the historical location data at the multiple times, a polynomial regression function equation is constructed to represent the change of the target object's position over time. Based on the polynomial regression function equation, the fourth position prediction data corresponding to the first time point of the target object is determined.

10. The method according to claim 1, characterized in that, The step of determining the weights corresponding to each location data in the at least one location data and the weights corresponding to the second location data using a neural network model includes: N experimental samples are obtained, the N experimental samples include N third position data of the target object in the wide-area coordinate system and position prediction data corresponding to each of the N third position data; each of the N third position data and the position prediction data corresponding to that third position data corresponds to one experimental sample; Construct an M-dimensional dataset based on the N experimental samples; Calculate the covariance matrix corresponding to the M-dimensional data set, and determine the M corresponding to the covariance matrix. One eigenvalue; Determine the M An M-dimensional eigenvector corresponding to one eigenvalue is used to obtain a mapping matrix; Based on the mapping matrix, the M-dimensional data set is reduced in dimensionality and mapped to an M-dimensional map. 1D dataset; For the M The 1D dataset is normalized to obtain the target dataset; The neural network model is trained using the target dataset to obtain the neural network model.

11. A positioning device, characterized in that, The device includes: A construction unit is used to establish a multi-branch tree structure based on the hierarchical relationship between multiple local coordinate systems, and to determine the transformation relationship between each local coordinate system and the wide-area coordinate system; wherein the root node of the multi-branch tree structure corresponds to the wide-area coordinate system of the local coordinate system; The processing unit is configured to obtain the second position data of the target object in the wide-area coordinate system based on the first position data of the target object in one or more local coordinate systems among the plurality of local coordinate systems and the transformation relationship; The first determining unit is used to determine at least one position prediction data of the target object in the wide-area coordinate system; The second determining unit is used to determine the positioning position of the target object in the wide-area coordinate system based on the at least one position prediction data and the second position data; The second determining unit is further configured to use a neural network model to determine the weights corresponding to each location data in the at least one location data and the weights corresponding to the second location data; obtain the location fusion data of the target object based on each location prediction data in the at least one location prediction data, each weight corresponding to each location prediction data, the second location data and the weights corresponding to the second location data, and determine the location fusion data as the positioning position of the target object in the wide-area coordinate system.

12. An electronic device, characterized in that, The electronic device includes a memory and a processor, wherein the memory stores computer-executable instructions, and the processor, when executing the computer-executable instructions in the memory, can implement the method of any one of claims 1 to 10.

13. A computer storage medium, characterized in that, The storage medium stores executable instructions that, when executed by a processor, implement the method of any one of claims 1 to 10.

Images