Target positioning method and device, electronic equipment and storage medium

Abstract

This application provides a target localization method, apparatus, electronic device, and storage medium. The method obtains the initial position of the target object by allocating distance through optimal sub-patterns; it pre-constructs a motion state model of the target object, predicts the state vector of the target object at the next moment based on the state transition equation of the motion state model, and determines the prior position of the target; it filters the current real observation set based on the prior position of the target to obtain a first observation set; it searches and locates the target based on the first observation set, using the previous position of the target object as the center, and determines the real-time target position; it narrows the search range to improve the real-time performance of localization; it fuses the real-time target position with the prior position of the target to obtain a corrected target position, and uses the corrected target position as the target localization result. By fusing the prediction and observation, the respective errors are offset, solving the problem of excessive localization error and improving localization accuracy and reliability.

Description

Technical Field

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

[0002] Currently, Global Navigation Satellite Systems (GNSS) are relatively mature, but in indoor environments or areas where signals are blocked, such as underground parking lots or shopping malls where vehicles or pedestrians are traveling, signals are often weak or even unusable. In such scenarios, positioning algorithms acquire wireless parameters from multiple base stations (BS), as well as information such as Time of Flight (TOF) and Angle of Arrival (AOA), to locate vehicles or pedestrians.

[0003] However, in large or complex environments, achieving a high-precision multi-base station positioning system will significantly increase hardware requirements and costs. In addition, base stations in a multi-base station system usually need to be precisely synchronized, and clock asynchrony between base stations may lead to large errors in positioning calculations. Meanwhile, for single-base station single-antenna positioning, it is necessary to rely on the angle parameters of the wireless signal for positioning, but since the corresponding terminal device has only a single antenna, it cannot meet the system hardware requirements. Therefore, there is an urgent need for a target positioning scheme that is highly accurate, low-cost, and computationally simple. Summary of the Invention

[0004] This application provides a target positioning method, apparatus, electronic device, and storage medium to solve the technical problems in related technologies where the target positioning scheme has low positioning accuracy, high cost, and relatively complex calculations when a vehicle travels into a GNSS signal failure area.

[0005] This application provides a target localization method, which includes: obtaining the initial position of a target object, such as a pedestrian or vehicle, by allocating distance based on the optimal sub-pattern; pre-constructing a motion state model of the target object, predicting the state vector of the target object at the next moment based on the state transition equation of the motion state model, and determining the prior position of the target; filtering the current real observation set based on the prior position of the target to obtain a first observation set, wherein the first observation set represents the arrival time difference formed only by a single reflection path; using the previous position of the target object as the center, performing search and localization based on the first observation set to determine the real-time target position; fusing the real-time target position with the prior position of the target to obtain a corrected target position, and using the corrected target position as the target localization result.

[0006] In some embodiments of this application, obtaining the initial position of a target object based on the optimal submode allocation distance includes: connecting virtual anchor points located in a ring around the perimeter to determine the motion area constituted by the target object; dividing the motion area into grids to determine the coordinates of each grid point; calculating the first flight time between each grid point and all virtual anchor points, and the second flight time from each grid point to the base station; determining the arrival time difference of each grid point based on the arrival time difference between the first and second flight times to obtain an estimated observation set; comparing the estimated observation set with the actual observation set based on the optimal submode allocation distance, and determining the grid point with the smallest comparison result as the initial position of the target object, wherein the actual observation set is determined by the multipath flight time and line-of-sight path flight time in the channel state information.

[0007] In some embodiments of this application, a motion state model of the target object is pre-constructed, and the state vector of the target object at the next moment is predicted based on the state transition equation of the motion state model to determine the prior position of the target. This includes: acquiring the three-dimensional coordinates, heading angle, linear velocity, and angular velocity of the target object to construct a motion state model of the target object; discretizing the motion state model to obtain the state transition equation; inputting the state vector of the previous moment into the state transition equation to obtain the state vector of the target object at the next moment, and using the state vector at the next moment as the prior position of the target.

[0008] In some embodiments of this application, filtering the current real observation set based on the target's prior location to obtain a first observation set includes: calculating the time between the target object and each virtual anchor point at the current time as a third flight time, and the time between the target object and the base station as a fourth flight time, based on the target's prior location; determining an estimated observation set based on the differences between multiple third and fourth flight times; subtracting each element of the real observation set from each element of the estimated observation set to obtain a first cost matrix; if the number of rows and columns of the first cost matrix are not the same, then matrix expansion is performed on the first cost matrix to generate a second cost matrix with the same number of rows and columns; preprocessing the number of rows and columns of the second cost matrix to obtain the minimum number of rows and columns; if the minimum number of rows and columns covers a preset element in the second cost matrix, and the sum of the covered rows and columns is equal to the order of the second cost matrix, then the matching is completed, and a matching pair between the estimated observation set and the real observation set is determined; and filtering is performed based on the difference value of each matching pair to determine the first observation set.

[0009] In some embodiments of this application, the real-time target location is determined by searching and locating the target object based on the previous location of the target object, using the previous location as the center, and based on the first observation set. This includes: determining the previous location of the target object at the previous time, where the previous location is the location result corresponding to the previous time; performing a local search in the first observation set with the previous location as the center, and selecting the optimal location based on the optimal sub-pattern allocation distance to determine the real-time target location.

[0010] In some embodiments of this application, fusing the real-time target position with the prior target position to obtain a corrected target position includes: using the real-time target position as an effective observation; differentiating the state transition equation based on each element of the effective observation to obtain a Jacobian matrix; calculating the covariance prediction value based on the Jacobian matrix; updating the covariance prediction value according to the Jacobian matrix and the Kalman gain; the Kalman gain being determined by the covariance prediction value, observation noise, and the Jacobian matrix; and obtaining the state update amount using the state transition equation, the Kalman gain, the state deviation between the effective observation and the prior target position.

[0011] The state update is superimposed on the state of the target's prior position to obtain the corrected target state, and the coordinates are extracted to obtain the corrected target position.

[0012] In some embodiments of this application, after obtaining the corrected target position, the method further includes: determining whether the tracking of the corrected target position at the current moment has ended based on preset rules; if the tracking has not ended, then continuing to locate the target object; if the tracking has ended, then using the corrected target position as the target location result; the preset rules include at least one of tracking duration, number of tracking times, tracking range, and positioning accuracy.

[0013] This application embodiment also provides a target positioning device, which includes: an initial positioning module, used to obtain the initial position of a target object based on the optimal sub-pattern distance allocation, wherein the target object is a pedestrian or a vehicle; a prediction module, used to pre-construct a motion state model of the target object, and predict the state vector of the target object at the next moment based on the state transition equation of the motion state model to determine the prior position of the target; a filtering module, used to filter the current real observation set based on the prior position of the target to obtain a first observation set, wherein the first observation set represents the arrival time difference formed only by a single reflection path; a positioning module, used to perform search positioning based on the first observation set with the previous position of the target object as the center to determine the real-time target position; and a positioning correction module, used to fuse the real-time target position with the prior position of the target to obtain a corrected target position, and use the corrected target position as the target positioning result.

[0014] This application also provides an electronic device, including: a memory storing a computer program thereon; and a processor for executing the computer program in the memory to implement the steps of any of the above embodiments.

[0015] This invention also provides a computer-readable storage medium having a computer program stored thereon, the computer program being used to cause a computer to perform the method provided in any of the above embodiments.

[0016] The beneficial effects of this application are as follows: The target localization method, device, electronic device, and storage medium proposed in this application obtain the initial position of the target object based on the optimal sub-pattern distance allocation; construct a motion state model of the target object, predict the state vector of the target object at the next moment based on the state transition equation of the motion state model, and determine the prior position of the target without the need for prior information of the initial position; filter the current real observation set based on the prior position of the target to obtain a first observation set, and eliminate invalid interference data through observation filtering, thereby improving the quality of observation data; use the previous position of the target object as the center, and perform search and localization based on the first observation set to determine the real-time target position, thereby improving the real-time performance of localization by narrowing the search range; fuse the real-time target position with the prior position of the target to obtain a corrected target position, and use the corrected target position as the target localization result. By fusing prediction and observation, the respective errors are offset, solving the problem of excessive localization error and improving localization accuracy and reliability. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0018] In the attached diagram: Figure 1 A schematic flowchart of a target localization method provided in an embodiment of this application; Figure 2 A specific schematic diagram illustrating the construction of a motion environment according to an embodiment of this application; Figure 3 A schematic flowchart of an extended Kalman filter loosely coupled localization method based on optimal sub-pattern allocation distance provided in an embodiment of this application; Figure 4 A schematic flowchart illustrating a specific process for constructing a range of motion according to an embodiment of this application; Figure 5A schematic diagram of a target positioning device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0019] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0020] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. The drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the shape, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0021] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the present application. However, it will be apparent to those skilled in the art that embodiments of the present application may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present application.

[0022] It should be noted that, in practical applications, the collection and processing of data such as production process data to be verified in this application must strictly comply with the requirements of relevant national laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of laws and regulations and the authorization of the personal information subject.

[0023] In related technologies, target tracking and positioning processes suffer from problems such as high observation noise interference, insufficient real-time positioning, large deviations between prediction and actual observation, and susceptibility of positioning accuracy to environmental influences. For example, inaccurate prediction of target motion state leads to significant deviations between the prior and actual positions at the next moment; the original observation set contains a large amount of invalid interference data, resulting in redundant positioning calculations and decreased accuracy; the full observation set search positioning efficiency is low and cannot meet real-time tracking requirements; and a single predicted or observed position cannot simultaneously ensure stability and accuracy, leading to insufficient reliability of the positioning results.

[0024] To address the aforementioned technical problems, this application provides a target localization method. This method obtains the initial position of the target object by allocating distance using the optimal sub-pattern; constructs a motion state model of the target object, and predicts the target object's state vector at the next moment based on the state transition equation of the motion state model to determine the target's prior position, eliminating the need for prior information about the initial position; filters the current real-time observation set based on the target's prior position to obtain a first observation set, eliminating invalid interference data and improving the quality of the observation data; uses the target object's previous position as the center and performs a search and localization based on the first observation set to determine the real-time target position, improving the real-time performance of localization by narrowing the search range; and fuses the real-time target position with the target's prior position to obtain a corrected target position, using the corrected target position as the target localization result. The fusion of prediction and observation cancels out their respective errors, solving the problem of excessive localization error and improving localization accuracy and reliability.

[0025] Please see Figure 1 , Figure 1 A schematic flowchart of a target localization method provided in an embodiment of this application is shown below. Figure 1 As shown, the method includes the following steps: Step S110: Obtain the initial position of the target object based on the distance allocated by the optimal sub-pattern. The target object is a pedestrian or a vehicle. For example, an estimated observation set is obtained by using observation equipment to acquire spatial correlation information between the target object and the observation sources (each observation device). The observation equipment includes, but is not limited to, UWB (Ultra Wide Band) base stations, lidar, and visual sensors. Since the estimated observation set includes a set of locations represented by multiple sensor sources, to acquire the most accurate initial position, the distances in the location set are calculated by allocating distances through optimal sub-patterns. Smaller distances indicate smaller errors. The point with the smallest calculated distance is used as the initial position of the target object. The target object can be a pedestrian indoors, a vehicle or pedestrian in an underground parking lot, or a tunnel scene. Alternatively, the observation equipment can also calculate the initial position of the target object using the time difference of arrival based on the grid partitioning method described below.

[0026] Step S120: Pre-construct a motion state model of the target object, and predict the state vector of the target object at the next moment based on the state transition equation of the motion state model to determine the prior position of the target. For example, the motion state model is a mathematical model pre-constructed for the target object to describe its motion laws and the evolution of its state over time; it characterizes the changing relationships of motion parameters such as position, velocity, and acceleration of the target object between consecutive moments, and is an abstraction and quantification of the target's physical motion characteristics. The state transition equation is established based on the motion state model and is a quantitative calculation formula used to map the motion state of the previous moment to the predicted state of the current moment.

[0027] The state transition equation is derived based on the motion law, clarifying the mapping relationship between state vectors at adjacent time points. The initial position (parking lot, driving road) and initial motion parameters (time interval, any driving mode including uniform speed, uniform acceleration or turning, and at least one of the vehicle's directional speed, heading angle or angular velocity at the previous time point) are substituted into the state transition equation to calculate the predicted value of the target's state vector at the next time point. The position parameter is extracted from this predicted value, that is, the target's prior position is used to predict the target's position at the next time point through the motion state model, which reduces the search range of the observation data and reduces the impact of environmental interference on the prediction results.

[0028] Step S130: Filter the real observation set at the current moment based on the prior position of the target to obtain the first observation set. The first observation set represents the arrival time difference formed only by a single reflection path. For example, the observations output by all observation devices at the current moment are collected to form a real observation set; the estimated observation set corresponding to the current moment is calculated based on the prior position of the target object; each observation in the estimated observation set is filtered in combination with the elements in the real observation set, and abnormal observation data that exceeds the filtering range or does not meet the filtering threshold are removed; the remaining valid observations after filtering are combined to form a first observation set; interference data and noise data in the first observation set are effectively removed, and only the arrival time difference formed by a single reflection path is retained in the first observation set, which improves the effectiveness and reliability of the observation data; it not only reduces computational redundancy, but also improves the real-time positioning performance.

[0029] Step S140: Using the previous location of the target object as the center, perform a search and location based on the first observation set to determine the real-time target location; For example, the target object's position coordinates at the previous moment are extracted, and a reasonable search range is set with that position as the center. Within the first observation set, a localization matching algorithm, such as least squares, maximum likelihood estimation, or weighted matching of observation data, is used to achieve secondary localization by searching for the optimal combination of observation data. Based on this optimal combination, the real-time target position at the current moment is calculated. By leveraging the continuity of the target object's motion and the effectiveness of the filtered observation data, the localization efficiency and accuracy are improved by narrowing the search space. This significantly enhances the efficiency of the localization search, ensuring the real-time performance of target tracking. Furthermore, the precise search based on the high-quality first observation set also improves the localization accuracy of the real-time target position.

[0030] Step S150: The real-time target position is fused with the prior target position to obtain the corrected target position, and the corrected target position is used as the target localization result.

[0031] For example, the selected fusion algorithm includes, but is not limited to, weighted fusion algorithm, Kalman filter update algorithm, Bayesian estimation fusion algorithm, etc.; the fusion weights of the real-time target position and the prior target position are determined, such as the weight of the real-time position based on the observation accuracy of the first observation set, and the weight of the prior position based on the prediction accuracy of the motion state model; the real-time target position and the prior target position are substituted into the fusion algorithm, and the fused corrected target position is obtained by calculation; the corrected target position is used as the final positioning result at the current moment, and at the same time, it is used as the target object at the next moment to realize continuous iteration of tracking, combining the advantages of prediction and observation, thereby improving the accuracy and reliability of the target positioning result; through closed-loop iterative design, the continuity of the tracking process is ensured, and tracking interruption is avoided.

[0032] By employing the above methods, motion state prediction provides directional guidance for observation filtering and localization search, significantly reducing computational redundancy; observation filtering eliminates invalid interference data, improving the quality of observation data; narrowing the search range improves the real-time performance of localization; and the fusion of prediction and observation cancels out their respective errors, improving localization accuracy and reliability.

[0033] In some embodiments, the initial localization process for a target object suffers from several problems, including poor adaptability to target localization in complex scenarios, low matching degree between observation data and the actual multipath scenario, significant impact of multipath interference on the accuracy of initial position estimation, and large positioning deviations due to unreasonable observation set matching metrics. Initial localization struggles to accurately represent the target's actual location, leading to redundancy in the initial search area or omission of the target's actual location. Furthermore, in multipath propagation environments, single observation data fails to reflect the true propagation characteristics between the target and the base station, resulting in an unreliable observation basis for initial position estimation.

[0034] For example, obtaining the initial position of the target object includes: connecting the virtual anchor points located in a ring around the perimeter to determine the movement area of ​​the target object; dividing the movement area into grids to determine the coordinates of each grid point; calculating the time from each grid point to all virtual anchor points as the first flight time and the time from each grid point to the base station as the second flight time; determining the arrival time difference of each grid point based on the arrival time difference between the first flight time and the second flight time; combining the arrival time differences of all grid points into an estimated observation set; comparing the estimated observation set with the actual observation set based on the optimal submode allocation distance, and determining the grid point with the smallest comparison result as the initial position of the target object, wherein the actual observation set is determined by the multipath flight time and line-of-sight path flight time in the channel state information.

[0035] For example, the coordinates of virtual anchor points in the unknown environment are extracted based on the virtual anchor point mapping algorithm, and all virtual anchor points are connected end to end in sequence according to the geometric distribution of virtual anchor points to form a motion region that wraps the target object, such as an irregular polygon or an approximate ellipse, to ensure that the fit error between the motion region and the actual motion region of the target object is within a preset threshold. The irregular virtual anchor points are transformed into quantifiable motion regions through discrete virtual anchor points.

[0036] For example, an orthogonal grid partitioning algorithm is used to uniformly divide the region within the motion area into grids. The precise coordinates of each effective grid point are calculated using a preset coordinate system, forming a set containing the coordinates of all grid points. This set serves as a candidate set. Grid discretization transforms the continuous space containing the target object into discrete candidate location points, reducing the complexity of the initial location search. Virtual anchor points can be understood as virtual base stations. The coordinates of the virtual anchor points and the precise coordinates of the base stations are obtained. Based on the propagation speed of electromagnetic waves or ultrasound, the straight-line distance from each grid point to each virtual anchor point is calculated, and the first flight time is obtained by combining the propagation speed. Similarly, the straight-line distance from each grid point to the base station is calculated to obtain the second flight time. The arrival time difference between the first and second flight times corresponding to each grid point is calculated, and the set consisting of the arrival time differences of all grid points is determined as the estimated observation set. Each element in the set corresponds one-to-one with a grid point. This realizes the transformation of spatial information of grid points into observation feature information, constructing an estimated observation set that can be compared with real observation data.

[0037] For example, when vehicle-mounted sensors and base stations transmit and receive signals wirelessly, channel state information is used as a known condition. Multipath flight time and line-of-sight path flight time are extracted from the channel state information, and the time difference corresponding to each multipath is calculated. All time differences form the real observation set. Parameters for the optimal sub-pattern assignment distance are set. The OSPA distance between the estimated observation set and the real observation set corresponding to each grid point is calculated sequentially, and the comparison result for each grid point is recorded. The grid point with the smallest OSPA distance value is selected, and its coordinates are chosen as the initial position of the target object. Using the OSPA distance achieves accurate matching of the observation set, effectively solving the dual problems of mismatched observation set numbers and element errors, thus improving the positioning accuracy of the initial position. Based on the real observation set extracted from the channel state information, and fully considering the multipath propagation characteristics, the adaptability of the positioning results to the real scene is improved.

[0038] By employing the above methods, precise representation of complex scenes is achieved through virtual anchor points and motion areas, solving the problem of poor adaptability of traditional positioning to irregular targets. Grid discretization transforms continuous space into discrete candidate points, reducing the computational complexity of positioning. The estimated observation set is constructed using time-of-flight difference (TOF), enabling the correlation between spatial information and channel characteristics. Matching OSPA distance with the actual observation set takes into account both the quantity and elemental errors of the observation set in multipath environments, improving positioning accuracy. This effectively addresses the technical pain points of traditional initial positioning, such as significant multipath interference, poor adaptability to complex scenes, and low matching accuracy. It can widely adapt to the initial positioning needs of complex scene targets in indoor and outdoor multipath propagation environments, exhibiting high positioning accuracy and scene adaptability.

[0039] In some embodiments, a motion state model of the target object is pre-constructed, and the state vector of the target object at the next moment is predicted based on the state transition equation of the motion state model to determine the prior position of the target. This includes: acquiring the three-dimensional coordinates, heading angle, linear velocity, and angular velocity of the target object to construct a motion state model of the target object; discretizing the motion state model to obtain the state transition equation; inputting the state vector of the previous moment into the state transition equation to obtain the state vector of the target object at the next moment, and using the state vector at the next moment as the prior position of the target.

[0040] For example, the three-dimensional coordinates of the target object are obtained by 3D LiDAR or visual stereo matching algorithm, and the heading angle of the target object is extracted by IMU (Inertial Measurement Unit) or visual attitude estimation algorithm to characterize the angle between the target object's motion direction and the reference direction of the preset coordinate system. The linear velocity of the target object is calculated by the velocity detection module or the position difference between adjacent time moments. The angular velocity of the target object is calculated by the angular rate sensor or the heading angle change rate.

[0041] Determine the sampling time interval. If the motion state model is a continuous-time domain model, such as a differential equation, use discretization methods, such as the Euler method or the Runge-Kutta method, to discretize the motion state model, transforming continuous-time variables into discrete-time series. Based on the discretized motion laws, derive the state transition matrix, clarify the mapping relationship between the state vectors of adjacent current time k and the next time k+1, and construct the state transition equation. Calculate the predicted state vector value for the next time k+1 through matrix operations, and extract the three-dimensional coordinate parameters from the predicted state vector value. These coordinate parameters represent the target object's prior position at the next time step.

[0042] The above methods provide a comprehensive representation of the target object's motion state, effectively addressing the limitation of single model parameters. Furthermore, the state transition equation based on complete motion information significantly improves the prediction accuracy of the target's prior position at the next moment.

[0043] In some embodiments, filtering the current real observation set based on the target's prior location to obtain a first observation set includes: calculating the time from the target object to each virtual anchor point as a third flight time and the time from the target object to the base station as a fourth flight time based on the target's prior location; determining an estimated observation set based on the differences between multiple third and fourth flight times; subtracting each element of the real observation set from each element in the estimated observation set to obtain a first cost matrix; if the number of rows and columns of the first cost matrix are not the same, then expanding the first cost matrix to generate a second cost matrix with the same number of rows and columns; preprocessing the number of rows and columns of the second cost matrix to obtain the minimum number of rows and columns; if the minimum number of rows and columns covers a preset element in the second cost matrix, and the sum of the covered rows and columns is equal to the order of the second cost matrix, then the matching is completed, and a matching pair between the estimated observation set and the real observation set is determined; and filtering is performed based on the difference value of each matching pair to determine the first observation set.

[0044] For example, the coordinates of virtual anchor points of the target object within the motion area and the precise coordinates of the base station are obtained; signal transceiver equipment is deployed to collect signal propagation data between the target object and each virtual anchor point at the current moment, and multiple third flight times are calculated based on the signal propagation speed; similarly, signal propagation data between the target object and the base station is collected, and a fourth flight time is calculated; the difference between the third flight time and the fourth flight time corresponding to each virtual anchor point is calculated one by one, and the set of all differences is determined as the estimated observation set. The elements in the estimated observation set correspond one-to-one with the virtual anchor points. The estimated observation set and the estimated observation set are both calculated, but the calculation time and calculation method are different.

[0045] Determine if the number of rows n and columns m of the first cost matrix are equal, where m and n are both integers greater than 1. If n = m, no expansion is needed, proceed to the next step. If n ≠ m, calculate the difference between the number of rows and columns Δ = |nm|. If n > m, add a Δ column to the right of the first cost matrix, setting the element value of the new column to the preset maximum difference threshold. If m > n, add a Δ row below the first cost matrix, also setting the element value of the new row to the preset maximum difference threshold. Determine the expanded matrix as a square matrix with the same number of rows and columns as the second cost matrix, and specify the order of the second cost matrix as K. Use a row and column coverage optimization algorithm to filter the rows and columns of the second cost matrix to find the minimum number of rows and columns that can cover all elements in the second cost matrix that are less than the preset difference threshold. Record the filtered row set R and column set C, count the number of rows |R| and the number of columns |C|, and determine this set of rows and columns as the minimum row and column that covers all valid elements. Determine if two conditions are met simultaneously: Condition 1, the minimum number of rows and columns can cover all preset elements in the second cost matrix; Condition 2, the sum of the number of covered rows and the number of covered columns equals the order of the second cost matrix |R| + |C| = K; If both conditions are met simultaneously, the matching is considered complete; In the second cost matrix, extract the elements at the intersection of the minimum row and column. The predicted observation set elements corresponding to these elements are combined with the actual observations to form matching pairs, and the index relationship of all matching pairs is recorded; Determine the difference value corresponding to all matching pairs, set a filtering threshold, and determine whether the difference value of each matching pair is less than the filtering threshold: If it is less than the filtering threshold, retain the predicted observation set elements corresponding to the matching pair; if it is greater than the filtering threshold, remove the element; The set composed of all retained observation elements is determined as the first observation set.

[0046] Extract the target object's location result from the previous moment and determine the coordinates corresponding to that location result as the previous position to ensure the integrity and timeliness of the previous position data. Based on the previous position and the target object's motion characteristics, set a local search range, usually a spherical region with the previous position as the center and a preset radius r. Select observation elements within this local search range from the first observation set to form a local observation subset. Perform data fitting and positioning calculation on the local observation subset to obtain the real-time target position at the current moment.

[0047] The above methods are used to construct an estimated observation set by using the time difference of flight, thereby transforming target location information into matching observation features; the problem of dimensional mismatch between the observation set and the benchmark set is solved by constructing a cost matrix and expanding its dimensions; the integrity and effectiveness of the matching are ensured by using minimum row and column coverage and dual condition verification; a high-quality first observation set is obtained by filtering by difference values; and local search based on the continuous motion of the target object improves positioning accuracy while ensuring real-time positioning.

[0048] In some embodiments, the real-time target location is determined by searching and locating the target object based on the previous location of the target object using the first observation set as the center, including: determining the previous location of the target object at the previous time, where the previous location is the location result corresponding to the previous time; performing a local search in the first observation set with the previous location as the center, and selecting the optimal location based on the optimal sub-pattern allocation distance to determine the real-time target location.

[0049] For example, leveraging the temporal continuity of target tracking, historical positioning results are transformed into spatial constraint benchmarks for current positioning. During target tracking, the motion state of the target object exhibits spatiotemporal continuity; the actual position at the current moment has a clear correlation with the positioning result at the previous moment, preventing abrupt changes beyond the reasonable range of motion. The previous position, as the precise positioning result at the previous moment, provides a reliable spatial reference benchmark for current positioning, effectively avoiding blind searches without a benchmark. Furthermore, a local search range is set based on the coordinates of the previous position and the historical motion parameters of the target object: a spherical region with the previous position as the center and a search radius; all observation elements in the first observation set are traversed, and the distance between the spatial coordinates of each element and the previous position is calculated based on the optimal sub-mode distance allocation; the position with the minimum distance is selected as the optimal position and determined as the real-time target position.

[0050] By using the above methods, a spatial constraint benchmark is constructed using the positioning results of the previous moment, which effectively solves the problems of no benchmark and low efficiency in global search; the local search based on motion continuity eliminates residual interference and improves the relevance and reliability of the observation data.

[0051] In some embodiments, fusing the real-time target position with the prior target position to obtain the corrected target position includes: using the real-time target position as an effective observation, differentiating the state transition equation based on each element of the effective observation to obtain the Jacobian matrix; calculating the covariance prediction value based on the Jacobian matrix, updating the covariance prediction value according to the Jacobian matrix and the Kalman gain, wherein the Kalman gain is determined by the covariance prediction value, observation noise, and the Jacobian matrix; obtaining the state update amount using the state transition equation, the Kalman gain, the state deviation between the effective observation and the prior target position; and superimposing the state update amount onto the state of the prior target position to obtain the corrected target state and extracting its coordinates to obtain the corrected target position.

[0052] For example, the dimensions of the state vector are defined, including three-dimensional coordinates, heading angle, linear velocity, angular velocity, etc. For each element in the effective observations, the first-order partial derivatives of the state vector in the state transition equation are calculated. All partial derivatives are arranged in rows (observation dimensions) and columns (state component dimensions) to construct the Jacobian matrix. The covariance matrix of the previous time step is retrieved, and combined with the state transition matrix and process noise matrix, the predicted covariance value is calculated. An observation noise matrix is ​​defined and calibrated based on the noise statistical characteristics of effective observations; for example, the noise variance of position observations is calculated using historical observation data. The Kalman gain is calculated based on the predicted covariance value, the observation noise matrix, and the Jacobian matrix. The updated covariance matrix is ​​obtained by updating the predicted covariance value. The state vector corresponding to the target's prior position is retrieved, and the deviation between the effective observations and the predicted observations corresponding to the target's prior position is calculated. The state deviation is derived from the observation model. The state update amount is calculated based on the Kalman gain and the state deviation. This update amount represents the magnitude and direction of the correction to the prior state vector. The obtained state update amount is superimposed with the state vector corresponding to the target's prior position to obtain the corrected target state vector. The corrected target state vector is validated to ensure that each state component is within a reasonable range and there are no abnormal jumps. Three-dimensional coordinate components are extracted from the corrected target state vector, and these extracted three-dimensional coordinate components are used as the corrected target position, completing the correction process.

[0053] The above methods solve the problem of fusion adaptation of nonlinear systems by using the Jacobian matrix, thus expanding the applicable scenarios of the scheme; the optimal balance between prediction error and observation error is achieved by dynamic covariance update and adaptive Kalman gain calculation, improving fusion accuracy; the cumulative error of prior state is significantly reduced by accurate feedback and superposition correction of state update, improving positioning accuracy and stability; and the adaptability to dynamic noise, state changes and complex motions is enhanced by various optimization schemes, improving robustness.

[0054] In some embodiments, after obtaining the corrected target position, the method further includes: determining whether the tracking of the corrected target position at the current moment has ended based on a preset rule; if the tracking has not ended, then continuing to locate the target object; if the tracking has ended, then using the corrected target position as the target location result; the preset rule includes at least one of tracking duration, number of tracking attempts, tracking range, and location accuracy meeting the standard.

[0055] For example, configuring preset rule parameters includes: if a tracking duration rule is included, setting a minimum tracking duration threshold T (in seconds); if a tracking count rule is included, setting a minimum tracking count threshold N (in frames); if a tracking range rule is included, setting the spatial range boundary of the target object that is allowed to be tracked; if a positioning accuracy compliance rule is included, setting a positioning accuracy threshold (in meters). By correcting the deviation between the target position and the reference position, the corrected target position data and cumulative tracking data at the current moment are obtained, and the preset rules are verified one by one: if it is a single rule, the tracking is determined to be over if the rule is met; if it is a combination of multiple rules, the tracking is determined to be over if the preset rule combination conditions are met; otherwise, it is determined to be not over.

[0056] If tracking is determined to be incomplete, the tracking process loop mechanism is triggered, using the current corrected target position as the previous position for the next positioning step, and the positioning calculation for the next step continues. If tracking is determined to be complete, the positioning result output process is initiated; the current corrected target position undergoes a final validity check, and the corrected target position that passes the check is used to generate the positioning result, thus completing the tracking process.

[0057] By constructing a quantifiable tracking termination criterion through the above methods, the problems of strong subjectivity and timing deviation in traditional judgments are effectively solved; the closed-loop logic ensures the continuity of the tracking process and improves resource utilization efficiency; and the end-of-track verification and fusion mechanism ensures the accuracy and reliability of the output positioning results.

[0058] Please see Figure 3 This is a schematic flowchart illustrating a loosely coupled localization method based on optimal sub-pattern allocation distance provided in an embodiment of this application, including: Step 1: During indoor positioning, the terminal performs parameter estimation on the extracted channel state information to obtain the Time-of-Flight (TOF) and angles of the LOS path and multipath. When both the base station and the terminal have single antennas, the information obtained through parameter estimation is only the TOF. Subtracting the TOF of the LOS path from the TOF of the multipath path yields the true observation set. .like Figure 2As shown, the virtual aiming points VA at the outermost edge of the environment are connected end to end to construct a closed geometric shape. Among them, VA1 to VA7 are connected end to end to form a closed shape, which forms the motion area of ​​the target object.

[0059] The motion area is divided into grids according to a preset precision, thereby obtaining the specific coordinates of grid points within the target motion area. That is, each grid point has at least one point and corresponding to at least one coordinate. First, the Time-of-Flight (TOF1) between each point in the grid and all VAs is calculated. Then, the TOF2 between each point and the base station (BS) is subtracted to obtain the Time Difference of Arrival (TDOA) for each point. The TDOAs of all points are combined to form an estimated observation set. Where l represents the l-th point within the grid; then, the estimated observation set is compared using the OSPA distance. With the actual observation set The difference between them can be expressed as: (1) (2) In formulas (1) and (2), M is m, which is... The number of elements in the middle; To estimate the observation set, N is The number of elements in the observation set, where m is the number of elements in the actual observation set; The cutoff point is a constant greater than 0, reflecting the degree of penalty for the two vectors having different dimensions; The order reflects the sensitivity of the OSPA distance to outliers; for The set of all permutations and combinations for The Middle The first permutation and combination One element, The minimum distance value, It is the minimum value.

[0060] If each grid point has only one point, then calculate the difference between the estimated observation set and the actual observation set for the same grid point in turn, and select the coordinate position corresponding to the grid point with the smallest difference as the initial position of the target object.

[0061] Step 2, the motion state model of the target object can be modeled as follows: (3) In formula (3) , , These are the three-dimensional coordinates of the target object; The heading angle of the target object; The linear velocity of the target object; For the angular velocity of the target object, The sampling period is denoted as . Therefore, the state transition equation for the target object is: (4) In formula (4) Let be the prior state of the target object at time k. This represents the prior state of the target object at time k-1. The sampling period; The noise is a process noise that follows a Gaussian distribution. , This is the predicted covariance value at time k. Therefore, the target's prior position... It can be obtained through the state transition equation.

[0062] Step 3: After obtaining the target's prior position, determine the estimated observation set based on the target's prior position, and filter the current real observation set. Since the observations received at each time step include line-of-sight paths, single-reflection paths, multiple-reflection paths, scattering, and clutter, only the single-reflection Time-of-Flight (TOF) is helpful for positioning. Too much clutter increases the computational load and affects positioning accuracy. Therefore, it is necessary to match the real observations. Here, the Hungarian algorithm is used for matching to filter out clutter. The steps are as follows: 3.1 Calculate the Time-of-Flight (TOF) 3 between the target object and each VA at the current time, and then subtract the TOF 4 between the target and BS to obtain the predicted observation set; 3.2 Subtracting each element of the actual observation set from each element of the predicted observation set yields the first cost matrix. If the number of rows and columns of the first cost matrix are inconsistent, the matrix with the larger of the two numbers is used. Expand the matrix by padding with zeros to The matrix, i.e., the second cost matrix; 3.3 in Subtract the minimum value from each row of the second cost matrix; 3.4 Subtract the minimum value from each column of the second cost matrix completed in the previous step; 3.5 Use the smallest row and column to cover the elements in the second cost matrix that are already 0 (i.e., preset elements), if the sum of the row and column is... If the match is successful, the match is considered complete; otherwise, proceed to the next step. 3.6 For all uncovered elements, find the minimum value among them, subtract the minimum value from all uncovered elements, add the minimum value to all elements that were covered twice (both rows and columns were covered), and return to step 3.5.

[0063] 3.7 After the matching algorithm is completed, matching pairs between the estimated observation set and the actual observation set are determined. The first observation set is determined by filtering based on the difference value of each matching pair. For example, the corresponding observation is obtained based on the position of the 0 element in the second cost matrix. At this time, the difference value between the matched observation and the calculated value is determined, i.e., by judging whether the difference value of the matching pair is greater than the upper threshold. If so, the matching pair is discarded; otherwise, the observations with a difference value less than the upper threshold are combined to determine the first observation set. This new observation set may not necessarily contain all observations of the VA, because due to interference from environmental noise, VAs far from the current position may not generate corresponding observations.

[0064] Step 4: After obtaining the new observation set, the target object is located again. To reduce computational load during localization, a global search is not performed on the target object; instead, a search is conducted within a certain range centered on the previous time step's position. For example... Figure 4 As shown, the real-time target location can be obtained quickly at this time.

[0065] Step 5: Obtain the real-time target location result. Then, this is used as an effective observation of the EKF (Extended Kalman Filter) and fused with the target prior position obtained through the state transition equation. This can smooth the positioning result and reduce the positioning error. The partial derivative of the above state transition equation with respect to each state variable is obtained, and its calculation formula is as follows:

[0066] In formula (5) Let Jacobian matrix be the state transition matrix. The heading angle of the target object; The linear velocity of the target object; The angular velocity of the target object; The sampling period.

[0067] Then, the update step of EFK is used to correct it. First, the partial derivative of each element in the effective observation with respect to each state is obtained, and the calculation formula is as follows: (6) In the above formula, in formula (6) The Jacobian matrix is ​​the effective observation matrix. The covariance prediction is first calculated using this Jacobian matrix, and the formula is as follows: (7) In formula (7) State noise, Let be the predicted covariance value at time k, used to verify the accuracy of the target's prior location. Let Jacobian matrix be the state transition matrix. Let be the Jacobian transpose of the state transition matrix. The covariance prediction value at time K-1 is then used to calculate the Kalman gain, which is calculated using the following formula: (8) In formula (8) To observe the noise, The Jacobian transpose of the effective observations, The Jacobian matrix of the effective observations, Given the predicted covariance value at time K, the state update is then calculated using the following formula: (9) In formula (9) The coordinates are derived from the prior location of the target. From The coordinates extracted from the state transition equation are then used to obtain the corrected target state, based on the state update quantity. The corrected coordinates are extracted from the matrix to obtain the updated covariance matrix, which is calculated using the following formula: (10) In formula (10), I is the identity matrix. For Kalman gain, The Jacobian matrix of the effective observations, The predicted covariance value at time K is updated by updating the covariance matrix.

[0068] Step 6: After obtaining the updated target position, determine whether the tracking has ended at the current time. If it has ended, then end the tracking. If the tracking has not ended, then continue to locate and track the target.

[0069] The above method calculates the target's observation set and compares it with the actual observation set using OSPA distance, employing a grid search approach to locate the target. By fusing the OSPA-derived location with the target's prior location and using EKF to correct the location result, the stability and accuracy of the positioning are improved. Target positioning using OSPA distance is achieved in a single-site, single-transmitter antenna configuration. An observation set filtering algorithm is used to filter out non-mirror reflection paths from the actual observation set to eliminate their interference with the positioning result. Target positioning utilizes TDOA between multipath paths, eliminating the need for signal synchronization between the base station and the terminal, significantly simplifying the signal processing flow.

[0070] In some embodiments, a target positioning device is provided for performing the target positioning method provided in any of the above embodiments. Please refer to... Figure 5 , Figure 5 A schematic diagram of a target positioning device provided in an embodiment of this application is shown below. Figure 5 As shown, the target positioning device 500 includes: The initial positioning module 510 obtains the initial position of the target object (a pedestrian or vehicle) based on the optimal sub-pattern distance allocation; the prediction module 520 constructs a motion state model of the target object and predicts the state vector of the target object at the next moment based on the state transition equation of the motion state model to determine the prior position of the target; the filtering module 530 filters the current real observation set based on the prior position of the target to obtain a first observation set, which represents the arrival time difference formed only by a single reflection path; the positioning module 540 searches and positions the target object based on the first observation set, using the previous position of the target object as the center, to determine the real-time target position; and the positioning correction module 550 fuses the real-time target position with the prior position of the target to obtain a corrected target position, and uses the corrected target position as the target positioning result.

[0071] Specific limitations regarding the target positioning device can be found in the limitations of the target positioning method described above, and will not be repeated here. Each module in the aforementioned target positioning device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware within or independently of the processor in the electronic device, or stored in software within the memory of the electronic device, so that the processor can call and execute the corresponding operations of each module.

[0072] In this embodiment, the target positioning device is essentially equipped with multiple modules to execute the target positioning method in any of the above embodiments. The specific functions and technical effects can be referred to in the above embodiments, and will not be repeated here.

[0073] See Figure 6 , Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown below. Figure 6 As shown, this embodiment of the invention also provides an electronic device 600, including a processor 601, a memory 602, and a communication bus 603; the communication bus 603 is used to connect the processor 601 and the memory 602; the processor 601 is used to execute a computer program stored in the memory 602 to implement the method described in any of the above embodiments.

[0074] This invention also provides a computer-readable storage medium having a computer program stored thereon, the computer program being used to cause a computer to perform the method provided in any of the above embodiments.

[0075] This application also provides a non-volatile readable storage medium storing one or more modules (programs) that, when applied to a device, enable the device to execute the instructions included in the steps provided in this application.

[0076] This application also provides a computer program product, including a computer program that, when executed by a processor, can implement the steps and corresponding content of the aforementioned method embodiments.

[0077] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0078] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0079] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0080] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0081] It should be understood that the terms "first," "second," etc., used in this application are used to distinguish similar objects and do not necessarily indicate a specific order or sequence. The technical features to which these terms are used can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in a sequence other than that shown in the figures or text.

[0082] It should be understood that although the flowcharts provided in the embodiments of this application indicate the various steps with arrows, the order indicated by the arrows does not necessarily limit the implementation order of these steps. Those skilled in the art can perform these steps in other orders according to different implementation scenarios and requirements.

[0083] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A target localization method, characterized in that, The method includes: The initial position of the target object, which may be a pedestrian or a vehicle, is obtained by allocating distance based on the optimal sub-pattern. A motion state model of the target object is pre-constructed, and the state vector of the target object at the next moment is predicted based on the state transition equation of the motion state model to determine the prior position of the target. Based on the prior position of the target, the current real observation set is filtered to obtain a first observation set, which represents the arrival time difference formed only by a single reflection path; Using the previous location of the target object as the center, a search and positioning process is performed based on the first observation set to determine the real-time target location; The real-time target location is fused with the prior target location to obtain the corrected target location, and the corrected target location is used as the target localization result.

2. The target localization method as described in claim 1, characterized in that, The method of obtaining the initial position of the target object based on the distance allocation of the optimal sub-pattern includes: Connect the virtual anchor points located around the perimeter end to determine the movement area of ​​the target object; The motion area is divided into grids, and the coordinates of each grid point are determined; Calculate the first flight time of each grid point to all the virtual anchor points, and the second flight time of each grid point to the base station. Based on the arrival time difference between the first flight time and the second flight time, determine the arrival time difference of each grid point to obtain an estimated observation set. The estimated observation set is compared with the real observation set based on the optimal sub-mode allocation distance. The grid point with the smallest comparison result is determined as the initial position of the target object. The real observation set is determined by the multipath flight time and line-of-sight path flight time in the channel state information.

3. The target localization method as described in claim 1, characterized in that, A motion state model of the target object is pre-constructed, and the state vector of the target object at the next moment is predicted based on the state transition equation of the motion state model to determine the prior position of the target, including: The motion state model of the target object is constructed by obtaining its three-dimensional coordinates, heading angle, linear velocity, and angular velocity. The motion state model is discretized to obtain the state transition equation; The state vector of the target object at the previous moment is input into the state transition equation to obtain the state vector of the target object at the next moment, and the state vector at the next moment is used as the prior position of the target.

4. The target localization method as described in claim 1, characterized in that, Based on the prior location of the target, the current real observation set is filtered to obtain a first observation set, including: Based on the prior location of the target, the time when the target object arrives at each virtual anchor point at the current time is calculated as the third flight time, and the time when the target object arrives at the base station is calculated as the fourth flight time. The estimated observation set is determined based on the differences between the multiple third flight times and the fourth flight time. Subtract each element of the actual observation set from each element of the estimated observation set to obtain the first cost matrix; If the number of rows and columns of the first cost matrix are not the same, then the first cost matrix is ​​expanded to generate a second cost matrix with the same number of rows and columns. The number of rows and columns of the second cost matrix is ​​preprocessed to obtain the minimum number of rows and columns; If the smallest row and column cover a preset element in the second cost matrix, and the sum of the covered rows and columns is equal to the order of the second cost matrix, then the matching is completed, and a matching pair between the estimated observation set and the actual observation set is determined. The first observation set is determined by filtering based on the difference value of each matching pair.

5. The target localization method as described in claim 1, characterized in that, Using the previous location of the target object as the center, and based on the first observation set, a search and positioning process is performed to determine the real-time target location, including: Determine the previous position of the target object at the previous moment, where the previous position is the positioning result corresponding to the previous moment; A local search is performed in the first observation set with the previous position as the center, and the optimal position is selected based on the distance allocated by the optimal sub-mode and determined as the real-time target position.

6. The target localization method as described in claim 1, characterized in that, The real-time target position is fused with the prior target position to obtain the corrected target position, including: Using the real-time target position as an effective observation, the state transition equation is differentiated based on each element of the effective observation to obtain the Jacobian matrix; The covariance prediction is calculated based on the Jacobian matrix. The covariance prediction is updated according to the Jacobian matrix and the Kalman gain. The Kalman gain is determined by the covariance prediction, the observation noise, and the Jacobian matrix. The state update quantity is obtained by using the state transition equation, the Kalman gain, the state deviation between the effective observation and the prior position of the target; The state update is superimposed on the state of the prior position of the target to obtain the corrected target state, and the coordinates are extracted to obtain the corrected target position.

7. The target localization method according to any one of claims 1-6, characterized in that, After obtaining the corrected target location, the following is also included: Based on preset rules, determine whether the tracking of the corrected target position at the current moment has ended; If the tracking is not completed, continue to locate the target object; If the tracking ends, the corrected target position will be used as the target localization result; The preset rules include at least one of the following: tracking duration, number of tracking attempts, tracking range, and positioning accuracy.

8. A target positioning device, characterized in that, The device includes: The initial positioning module is used to obtain the initial position of the target object, which is a pedestrian or a vehicle, based on the distance allocated by the optimal sub-pattern. The prediction module is used to pre-build a motion state model of the target object, predict the state vector of the target object at the next moment based on the state transition equation of the motion state model, and determine the prior position of the target. The filtering module filters the current real observation set based on the prior position of the target to obtain a first observation set, which only contains the arrival time difference formed by a single reflection path. The positioning module is used to perform a search and positioning based on the first observation set, with the previous location of the target object as the center, to determine the real-time target location; The positioning correction module is used to fuse the real-time target position with the prior target position to obtain a corrected target position, and use the corrected target position as the target positioning result.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images