Joint positioning method, system and central processing unit based on radar multipath utilization

Abstract

This invention relates to the field of radar sensing and target localization technology, specifically to a joint localization method, system, and central processing unit based on radar multipath utilization. The method includes: extracting multiple peak distances from received radar echo signals; obtaining a set of candidate path distance triples based on the peak distances; sorting and conflict filtering the triple set to obtain a filtered target set; performing master-slave judgment and filtering on any two targets in the filtered target set based on path distances to obtain multiple independent targets; determining the available measurement set for each independent target; determining the loss function and constraints based on the available measurement set, performing least-squares optimization, and performing line-of-sight type judgment and reverse verification based on the optimized values ​​to finally obtain the coordinate estimate of each independent target. This invention can improve the accuracy and robustness of joint line-of-sight and non-line-of-sight localization.

Description

Technical Field

[0001] This invention relates to the field of radar sensing and target localization technology, specifically to a joint localization method, system, and central processing unit based on radar multipath utilization. Background Technology

[0002] Multipath-based radar positioning technology, by modeling non-line-of-sight paths such as reflections and diffractions caused by obstacles like walls and corners, can achieve target perception and localization even when targets are obscured. This is applicable to scenarios such as indoor surveillance, urban sensing, and through-wall detection. Since line-of-sight paths are often blocked in real-world environments, radar signals typically rely on wall reflections or edge diffraction for propagation. Although multipath propagation can lead to path ambiguity and false targets, its latency is strictly constrained by environmental geometry, thus it can be treated as a virtual anchor point for localization.

[0003] Existing research has proposed various multipath utilization methods, such as multipath artifact association and inverse mapping methods based on point spread functions, quasi-analytical propagation models that consider multiple specular reflections of buildings, and methods that track arrival times by nearest neighbors and use least squares to complete diffraction and first- and second-order reflection paths in L-shaped corner scenes.

[0004] However, the following problems still exist in practical applications: First, the target coupling processing capability is insufficient. Existing models mostly assume that the echoes of each target are independent and ignore the mutual scattering between targets, resulting in coupling clutter, false peaks, and positioning offset in multi-target environments. Second, in mixed line-of-sight and non-line-of-sight scenarios, single-channel radars lack the ability to quickly and autonomously distinguish between LOS and NLOS, and usually rely on prior information or multi-channel hardware. Third, in complex environments, there are many high-order reflection and diffraction paths, and traditional matching algorithms have difficulty distinguishing between real targets and high-order derived false targets, which easily leads to high false alarm rates. Fourth, ranging data is often affected by system bias, environmental changes, and non-Gaussian noise, while traditional least squares methods are sensitive to outliers and lack robustness and convergence in low signal-to-noise ratio and multipath dense scenarios. Summary of the Invention

[0005] In view of the above problems, the present invention provides a joint positioning method, system and central processing unit based on radar multipath utilization, which solves the technical problems of low positioning accuracy, poor robustness and easy generation of false targets caused by ignoring the coupling between targets, difficulty in handling mixed propagation scenarios and lack of deviation correction in the prior art.

[0006] On the one hand, the present invention provides a joint positioning method based on radar multipath utilization, comprising the following steps: Step S1: Illuminate the object under test with radar, preprocess the received radar echo signal and extract multiple peak distances; The radar is located on one side of the L-shaped spatial region, and the object being measured is located in the line-of-sight area or the non-line-of-sight area of ​​the radar within the L-shaped spatial region. Step S2: Based on the midpoint of any two different peak distances among the multiple peak distances, a set of candidate path distance triples is obtained; Step S3: Establish an initial target set based on the short path support of the path distances in the candidate path distance triplet set. The targets in the initial target set include path distances and corresponding path distance triplet sets. Sort and filter the path distance triplet sets of each target to obtain the filtered target set. Step S4: Based on path distance, perform master-slave judgment and filtering on any two targets in the filtered target set to obtain multiple independent targets; determine the six-path group list for each independent target; A usable measurement set is established for each independent target based on the path distance triplet set and the path distance in the six-path group list; the usable measurement set includes multiple path distances. Step S5: Determine the loss function and constraints based on the available measurement set of independent targets, perform least squares optimization based on the loss function and constraints, and perform line-of-sight type judgment and reverse verification based on the optimized values ​​to finally obtain the coordinate estimate of each independent target.

[0007] Preferably, the step of preprocessing the received radar echo signal and extracting multiple peak distances specifically includes: (1) Receive single-channel radar echo signals and obtain a one-dimensional range profile through matched filtering or range FFT processing; (2) Discretely sample and perform Hilbert transform on the one-dimensional distance profile to obtain the amplitude envelope sequence, detect the local maxima of the amplitude envelope sequence, and obtain multiple peak positions; (3) For multiple peak positions, sort the peak distances of the peak positions whose peak amplitude is greater than the median of the amplitude envelope sequence from smallest to largest to obtain the multiple peak distances.

[0008] Preferably, step S2 specifically includes: Step S2-1: Calculate the midpoint value for two different peaks in the observation distance vector; search among the multiple peak distances to find the peak closest to the midpoint value; Step S2-2: Sort the two different peaks and the peak closest to the midpoint value to form a valid triplet; The conditions for forming a valid triple are: ,

[0009] in, The midpoint tolerance threshold. This indicates obtaining the absolute value. This represents the distance to the r-th peak. This represents the midpoint of the distance between the p-th and q-th peaks; Step S2-3: Return to step S2-1 until all two different peaks in the observation distance vector are traversed, and all valid triples are used to form a candidate triple set.

[0010] Preferably, step S3 specifically includes: Step S3-1: For each peak index in the candidate path distance triplet set, calculate the corresponding short path support; the expression for the short path support is:

[0011] in, This represents the support of the i-th peak path. This represents the distance of the i, j, k-th path. This represents the number of triples with i as the shortest path index; Step S3-2: Construct a short path occurrence index set by indexing all peak paths with a support greater than or equal to 1; For each index in the set of short path occurrence indexes, construct the initial target structure, expressed as:

[0012] in, Let i' represent the i'-th initial target structure. This represents the path distance to the i'th initial target structure. This represents the set of triplets representing the path distances to the i'-th initial target structure. This represents the set of peak path indices with a support greater than or equal to 1. Step S3-3: For each initial target structure , will the corresponding Sort by midpoint consistency error from smallest to largest. The expression for midpoint consistency error is:

[0013] in, This represents the midpoint consistency error of the path distances for the i, j, and k-th paths. Step S3-4: For each sorted triplet of the initial target structure, perform greedy filtering, including recording the used middle path index set and long path index set respectively, and traversing the sorted triplets in turn. If the index of the current triplet is... and index If it is not occupied, the triplet will be retained, and... and Mark as occupied; otherwise discard. Finally, the filtered target structures are obtained, and all the filtered target structures constitute the filtered target set.

[0014] Preferably, step S4 specifically includes: Step S4-1: For two targets in the filtered target set Check if two distinct triples exist. and and a triple This makes the following proximity condition true:

[0015] in, Indicates the merged tolerance. and The longer path distance between the two triples of the candidate primary target is given. The shortest path distance to the candidate false target. The long path distance in the candidate false target triple; Step S4-2: If the proximity condition is met, then determine... To be subordinate to the main objective False targets, for Establish a six-path group; the six-path group includes six different radar paths to the object being measured; Step S4-3: Return to step S4-1 until any two targets have been processed, and finally obtain a list of each independent target and its associated six-path groups; Step S4-4: For each independent target, the observed distance values ​​corresponding to all peak indices in the corresponding filtered triplet set are summarized with all distance measurements contained in each group in the six-path group list to obtain the available measurement set, which includes all path distances.

[0016] Preferably, in step S5, the loss function expression is:

[0017]

[0018]

[0019] in, This represents the minimum Huber loss corresponding to the propagation hypothesis H. This represents the Huber loss function. This represents the input residual of the Huber loss function. This is a transition parameter, with a value range of 0.01m to 0.1m; Indicates measurement residuals, Represents the theoretical distance function. Represents the state vector. For the target coordinates, For system common deviation, This represents the distance of the k-th path; The expression for the constraint condition is:

[0020]

[0021] in, This represents the constraint condition corresponding to the propagation hypothesis H, where LOS and NLOS represent line-of-sight and non-line-of-sight, respectively. and This indicates the corridor width parameter. Represents the ordinate of the corner point. This represents the minimum value of the ordinate of the detected object. These represent the minimum and maximum values ​​of the system's common deviation, respectively.

[0022] Preferably, in step S5, the step of determining the line-of-sight type based on the optimized value and performing reverse verification to finally obtain the coordinate estimate of each independent target specifically includes: Solving the constrained optimization problem under both LOS and NLOS assumptions yielded two sets of optimization results. Calculate the minimum Huber loss under each of the two assumptions; take the assumption with the smaller minimum Huber loss under the two assumptions as the initial selection mode, and obtain the corresponding state vector estimate as the initial selection result; For the initial selection results, calculate the maximum absolute residual of all elements in the available measurement set, expressed as:

[0023] in, This indicates that the maximum value is determined based on all elements in the available measurement set. Let represent the available measurement set for the i-th independent target. express The elements in This indicates the coordinates of the detected object and the common deviation of the system in the preliminary selection results. This indicates a preset tolerance threshold; If the maximum absolute residual is less than the preset tolerance threshold, the initial selection result will be used as the final location result of the current independent target. If the maximum absolute residual exceeds the tolerance threshold, then calculate whether the maximum absolute residual under another propagation hypothesis is less than the preset tolerance threshold. If the maximum absolute residual under another propagation hypothesis is less than the preset tolerance threshold, then use the state vector under the other propagation hypothesis as the final localization result of the current independent target. The final positioning result is output, which includes the coordinate estimate of each independent target and the estimate of the system common deviation.

[0024] On one hand, the present invention provides a joint positioning system based on radar multipath utilization, characterized in that it includes: The peak distance extraction module is used to illuminate the object under test with radar, preprocess the received radar echo signal and extract multiple peak distances; the radar is located on one side of the L-shaped spatial region, and the object under test is located in the line-of-sight area of ​​the radar or the non-line-of-sight area of ​​the radar in the L-shaped spatial region. The candidate triplet construction module determines the candidate path distance triplet set by judging the midpoint of any two different peak distances among the multiple peak distances. The target set construction module is used to establish an initial target set based on the short path support of the path distances in the candidate path distance triplet set. The targets in the initial target set include path distances and corresponding path distance triplet sets. The path distance triplet sets of each target are sorted and conflict filtered to obtain the filtered target set. A usable measurement set construction module is used to perform master-slave judgment and filtering on any two targets in the filtered target set based on path distance to obtain multiple independent targets; determine a six-path group list for each independent target; and establish a usable measurement set for each independent target based on the path distance triplet set and the path distance in the six-path group list; the usable measurement set includes multiple path distances; The coordinate estimation module is used to determine the loss function and constraints based on the available measurement set of independent targets, perform least squares optimization based on the loss function and constraints, and perform line-of-sight type judgment and back-verification based on the optimized value, so as to finally obtain the coordinate estimation value of each independent target.

[0025] On one hand, the present invention provides a joint positioning center processing unit based on radar multipath utilization, characterized in that it includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the operations performed by the joint positioning method based on radar multipath utilization.

[0026] Compared with the prior art, the present invention has at least the following beneficial effects: (1) This invention can adapt to the complex propagation environment in L-shaped spatial regions where line-of-sight (LOS) and non-line-of-sight (NLOS) coexist. By extracting multiple peak distances from radar echoes and combining them with the layer-by-layer screening of "midpoint judgment - candidate path distance triple - short path support", this invention does not rely on whether the target is in LOS or NLOS state known in advance, but directly recovers the set of path distances that conform to the physical propagation law from the echoes, thereby effectively breaking through the limitation of "all targets must be in the same propagation mode" in the prior art, and significantly improving the applicability and generalization ability in complex scenarios.

[0027] (2) This invention can effectively suppress misjudgments caused by multipath aliasing, false peaks, and path conflicts. By sorting and conflict filtering the path distance triples in the initial target set, and then further performing master-slave judgment, independent target screening, and six-path group construction, this invention achieves step-by-step purification from candidate paths to the usable measurement set, reducing the impact of abnormal echoes, overlapping paths, and interference peaks on the positioning results. This not only improves the accuracy of path identification but also enhances the target separation capability, and is especially suitable for mixed scenarios with large differences in the energy of direct waves and diffracted waves and heterogeneous path structures.

[0028] (3) This invention can improve the stability and accuracy of the final coordinate estimation and has the ability to automatically select the propagation mode. Based on the available measurement set, two theoretical distance functions, LOS and NLOS, are constructed. Combined with Huber weighted least squares to jointly estimate the target coordinates and system bias, the influence of outlier error can be effectively reduced during the optimization process. Then, by comparing the residual norms under the two assumptions and performing reverse verification, the optimal propagation mode can be automatically determined. This mechanism enables the positioning results to complete pattern recognition and parameter solving without manual intervention. Under noise-free conditions, the maximum positioning error can be strictly controlled within 0.20 meters, showing high robustness and accuracy upper limit. Attached Figure Description

[0029] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0030] Figure 1 A schematic diagram of a typical scenario for joint positioning provided by the present invention.

[0031] Figure 2 The diagram shows a joint positioning simulation scenario provided by this invention.

[0032] Figure 3 A schematic diagram of the path recognition results of the simulation data provided by this invention.

[0033] Figure 4 A schematic diagram of the positioning results of the simulation data provided by this invention.

[0034] Figure 5 This is a schematic diagram of a real-world scenario for joint positioning provided by the present invention.

[0035] Figure 6 This is a schematic diagram of the joint positioning results in a real-world scenario provided by the present invention.

[0036] Figure 7 The flowchart of the joint positioning method based on radar multipath utilization provided by the present invention is shown. Detailed Implementation

[0037] To better understand the above-described objectives, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other. Furthermore, the present invention can be implemented in other ways different from those described herein; therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0038] The line-of-sight and non-line-of-sight positioning methods targeted by this invention are applicable to L-shaped corner scenarios, such as... Figure 1 As shown, the geometric corner is formed by walls W1 and W2. The corridor width parameter is and The single-channel radar is located in It is deployed at a corner. The target's actual location is... It may be located in the line-of-sight (LOS) or non-line-of-sight (NLOS) region. Based on the Uniform Diffraction Theory (UTD), an equivalent virtual target point is defined. as follows:

[0039] in, These represent the first to fourth equivalent virtual target points, respectively.

[0040] To unify line-of-sight and non-line-of-sight conditions, a one-way path length is defined. ( The expression is:

[0041] in, Indicating the non-line-of-sight assumption, Indicating the line-of-sight assumption, This represents the vector between the i-th equivalent virtual target point and the radar position. This indicates that the vector length is being retrieved. This represents the position vector of the radar relative to the geometric corner point. Let represent the position vector of the i-th equivalent virtual target point and the geometric corner point.

[0042] signal via path From radar to target, and then along the path Returning the corresponding equivalent one-way distance measurement value:

[0043] in, Indicates the path From radar to target, and then along the path The one-way distance of the return trip.

[0044] Considering the attenuation of higher-order reflections, eight observable path types are defined, forming the observable path set. :

[0045] The physical meaning of each path type is as follows: : The direct path from radar to target to radar (equivalent to corner diffraction under NLOS). : Radar—Wall W2—Target—Wall W2—Radar first-order reflection path; The second-order reflection path of radar—wall W1—wall W2—target—wall W2—wall W1—radar; , , : A combination of the above basic paths (such as ), corresponding to asymmetric multipath , Paths involving higher-order reflections or diffractions experience significant attenuation in actual measurements and can be selectively used based on the signal-to-noise ratio.

[0046] In subsequent implementations of this invention, the observed distance set is used. The first 6 strong path types Construct a six-path group to describe the observation distance of the observed object.

[0047] To illustrate the effectiveness of the method proposed in this invention, the following detailed description of the above technical solution is provided through a specific embodiment, such as... Figure 7 As shown, a joint positioning method based on radar multipath utilization is disclosed, and the specific implementation steps are as follows: Step S1: Illuminate the object under test with radar, preprocess the received radar echo signal and extract multiple peak distances; The radar is located on one side of the L-shaped spatial region, and the object being measured is located in the line-of-sight area or the non-line-of-sight area of ​​the radar within the L-shaped spatial region. In this step, a single-channel radar is used to transmit a detection signal to the target object and receive the echo signal containing the target's multipath components. After signal preprocessing, the effective peak distance is extracted to form the observation range vector. The specific processing procedure is as follows.

[0048] A one-dimensional range profile is obtained by receiving a single-channel radar echo signal and processing it through matched filtering or range FFT. The mathematical model of the range profile is the superposition of multipath components, and its expression is:

[0049] in, This represents the echo distance at time t. Indicates the first The delay of the path, express Echo distance at time, The total number of transmission paths, and The first The complex gain coefficient and equivalent round-trip delay of each path, Let be the additive white Gaussian noise at time t.

[0050] Discrete sampling is performed on the obtained one-dimensional distance profile to obtain a discrete one-dimensional distance profile sequence, expressed as:

[0051] in, For discrete distance cell indexes, This represents the discrete distance value of index n. Indicates the sampling interval.

[0052] Perform a Hilbert transform on the discrete one-dimensional distance profile sequence to obtain the amplitude envelope sequence, expressed as:

[0053] in, Represents the amplitude envelope sequence. This indicates obtaining the absolute value; It is the discrete Hilbert transform operator.

[0054] Set detection threshold The median of the amplitude envelope sequence; in Local maxima are detected in the middle, and multiple peak positions are obtained. If the peak amplitude at the peak position satisfies: Furthermore, the difference between the peak distance and the retained peak distance is greater than the preset spacing constraint. ( If the distance is (cm), then the peak distance at the current peak position is retained.

[0055] The retained peak distances are sorted from smallest to largest to obtain the final number of peak distances. Where K is the total number of peak distances, and each peak distance can be expressed as the following expression:

[0056] in, This represents the distance to the k-th peak. At the speed of light, This is the discrete index corresponding to the peak value.

[0057] Through the above steps, the multiple peak distances obtained by this invention include all paths from the target direct path, the first-order reflection path of the wall, the second-order reflection path and their mixed paths, providing basic data for subsequent multipath triplet association and target localization.

[0058] Step S2: Based on the midpoint of any two different peak distances among the multiple peak distances, a set of candidate path distance triples is obtained; In this step, the inherent midpoint consistency geometric constraint between the distances of multiple paths in the L-shaped corner scene is used to perform pairwise matching and association of the peak values ​​in the observed distance vector, and extract valid candidate triples that satisfy the constraint.

[0059] For any two different peaks in the observed distance vector ( ), calculate the arithmetic midpoint value of the two:

[0060] in, This represents the midpoint of the distance between the p-th and q-th peaks. These represent the peak distances of the p-th and q-th peaks, respectively.

[0061] The midpoint value is derived from the algebraic relationship of multipath distance: based on scene geometry modeling, if If we consider the equivalent one-way distance measurements for two different path types, then there exists a third path whose distance measurement is numerically equal to the average of the two. This property is strictly constrained by the mirror-symmetric geometry of the L-shaped corner.

[0062] At multiple peak distances Searching for the midpoint The closest peak value is expressed as:

[0063] Where r represents the index of the nearest peak. This indicates that the minimum value is searched for the distance between the multiple peaks.

[0064] like If the following proximity conditions are met, then the peak values ​​indexed as p, q, r will form a valid triplet. .

[0065]

[0066] in, The midpoint tolerance threshold is 0.01m to 0.1m.

[0067] Sort the triples in ascending order of distance to obtain ordered index triples. ,satisfy Collect all valid triples to form a candidate triple set:

[0068] in, Let i, j, k represent the set of candidate triples, where i, j, k are peak indices. Let represent the peak distances of the i-th, j-th, and k-th peaks, respectively.

[0069] The candidate path distance triplet set represents the geometric constraint relationship between multipath signals in multiple peak distances, providing structured association data for subsequent target clustering and path topology analysis.

[0070] Step S3: Establish an initial target set based on the short path support of the path distances in the candidate path distance triplet set. The targets in the initial target set include path distances and the corresponding path distance triplet sets. Sort and filter the path distance triples for each target to obtain the filtered target set; In this step, the first step is to process the set of candidate path distance triples. For each peak index i in the graph, this invention determines the corresponding support as the shortest path, expressed as:

[0071] in, This represents the support of the i-th peak path. This represents the number of triples with i as the shortest path index.

[0072] The physical meaning of the support is that, in an L-shaped corner scenario, the direct path or shortest equivalent path to a target will be the minimum distance component among all multipath triples associated with that target. Therefore, the high support of short paths can serve as an important basis for identifying real targets.

[0073] All peak paths with a support greater than or equal to 1 are indexed to form a short path occurrence index set.

[0074] For each index in the set of short path occurrence indexes, construct the initial target structure, expressed as:

[0075] in, Let i' represent the i'-th initial target structure. This represents the path distance to the i'th initial target structure. This represents the set of triplets representing the path distances to the i'-th initial target structure. This represents the set of peak path indices with a support greater than or equal to 1.

[0076] For each initial target structure , will the corresponding Sort by midpoint consistency error from smallest to largest. The expression for midpoint consistency error is:

[0077] in, This represents the midpoint consistency error of the path distances for the i, j, and k-th paths.

[0078] For each sorted triplet of the initial target structure, a greedy filtering strategy is applied, including: maintaining two sets of used indices, specifically recording the used mid-path index set and long-path index set respectively; iterating through the sorted triplets sequentially; and if the index of the current triplet is... and index If none of them are occupied, then the triplet will be retained, and... and Mark as occupied; otherwise discard.

[0079] The final filtered target structure is expressed as follows:

[0080] in, Let i' represent the target structure after filtering. Let represent the set of path distance triples corresponding to the i'th filtered target structure.

[0081] All filtered target structures are combined to form the filtered target set, which is used for subsequent target discrimination.

[0082] Step S4: Based on path distance, perform master-slave judgment and filtering on any two targets in the filtered target set to obtain multiple independent targets; determine the six-path group list for each independent target; A usable measurement set is established for each independent target based on the path distance triplet set and the path distance in the six-path group list; the usable measurement set includes multiple path distances. In this step, any two targets in the filtered target set are judged, false targets introduced by inter-target coupling scattering or higher-order reflection are removed, real independent targets are extracted, and a six-path group list containing six physical path distances is constructed for each independent target. Finally, they are integrated to form a usable measurement set, as described in detail below.

[0083] For any two targets in the filtered target set ,Will As a candidate primary target As a candidate false target.

[0084] Check if two distinct triples exist. and and a triple This makes the following proximity condition true:

[0085] in, Indicates the merged tolerance. and The longer path distance between the two triples of the candidate primary target is given. The shortest path distance to the candidate false target. The long path distance in the candidate false target triplet.

[0086] The physical basis of the judgment condition is that in an L-shaped corner multi-target scenario, when the scattered signal of target b is reflected back to the radar by target a, there is a strict algebraic constraint relationship between the resulting coupling path distance and the distance of two multipath paths of the main target a. Therefore, it can be identified and distinguished by the above-mentioned distance proximity condition.

[0087] If the proximity condition is met, then determine To be subordinate to the main objective False targets, for A six-path group is established, which has been described in detail above. The expression for the six-path group is:

[0088] This group corresponds to the set of theoretical paths. . Figure 3 The path identification results for the simulation data correspond to .

[0089] false target Marked as occupied, it will no longer be treated as a separate target and will be grouped. Add to main target Group list Repeat the above process until all target pairs have been processed, ultimately obtaining a list of each independent target and its associated six-path groups.

[0090] For each independent target, the corresponding set of filtered triples will be... The observed distance values ​​corresponding to all peak indices, and the six-path group list. The available measurement set is obtained by summing all distance measurements contained in each group. The available measurement set includes all path distances and can be labeled with their theoretical path types. .

[0091] Step S5: Determine the loss function and constraints based on the available measurement set of independent targets, perform least squares optimization based on the loss function and constraints, and perform line-of-sight type judgment and reverse verification based on the optimized values ​​to finally obtain the coordinate estimate of each independent target.

[0092] In this step, the present invention constructs theoretical distance functions for each independent target under both LOS and NLOS propagation assumptions, establishes a robust weighted least squares problem with the Huber loss function as the optimization objective, performs joint optimization under geometric constraints, and then automatically determines the propagation mode and outputs the target coordinate estimate through residual comparison and back-verification. The specific process is as follows.

[0093] For each independent objective, define a state vector. ,in For the target coordinates, To define the common bias of the system, two propagation hypotheses are set. LOS and NLOS represent line-of-sight and non-line-of-sight, respectively, corresponding to the physical scenarios where the measured object is in the radar's direct line of sight area and the corner obstruction area.

[0094] Based on the propagation hypothesis H, a theoretical distance function is defined for each path type p in the available measurement set. The expression is:

[0095] in, Indicates radar coordinates, Indicates the coordinates of the corner point. This represents the coordinates of the 0th virtual target point. This represents the coordinates of the i-th virtual target point. Indicates the path type. This indicates the direct path from radar to target and back to radar. This represents the path numbered ii. , , This represents a combination of the direct path from radar to target to radar, the first-order reflection path from radar to wall 2 to target to wall 2 to radar, and the second-order reflection path from radar to wall W1 to wall W2 to target to wall W2 to wall W1 to radar. This represents the theoretical distance of the path numbered ii. This indicates that the path with the number type ii is different from The theoretical distance.

[0096] The theoretical distance determined by the above analysis unifies the distance calculation framework for multipaths under both LOS and NLOS propagation modes, making the same optimization solution process applicable to both assumptions.

[0097] Based on the aforementioned theoretical distance function For each path distance in the available measurement set, the measurement residual under assumption H is calculated as follows:

[0098] in, This represents the measurement residual.

[0099] To improve the robustness of the localization algorithm to outliers and non-Gaussian noise, this invention uses the Huber loss function as the optimization objective. The expression for the Huber loss function is:

[0100] in, This represents the Huber loss function. This represents the input residual of the Huber loss function. This is a transition parameter, with a value range of 0.01m to 0.1m.

[0101] The geometric constraints determined in this invention are set differently depending on the different propagation assumptions H, and are expressed as follows:

[0102]

[0103] in, This represents the constraint condition corresponding to the propagation hypothesis H. Indicates the coordinates of the object being detected. and This indicates the corridor width parameter. Represents the ordinate of the corner point. This represents the minimum value of the ordinate of the detected object. represents the minimum and maximum values ​​of the system's common deviation, typically taking values ​​of [0.01 m, 0.10 m]. These geometric constraints limit the optimization search space to a region physically consistent with the propagation assumption, effectively avoiding potential physical inconsistencies in unconstrained optimization.

[0104] Solve the constrained optimization problem under both the LOS and NLOS assumptions. The expression for the optimization problem is as follows:

[0105] in, This represents the estimated state vector corresponding to the propagation hypothesis H. This indicates that the optimization is performed to minimize the estimated state vector. Let represent the available measurement set for the i-th independent target. express The elements in.

[0106] Solving the constrained optimization problem under both LOS and NLOS assumptions yielded two sets of optimization results. The minimum Huber loss is calculated under both assumptions, and the expression is as follows:

[0107] in, This represents the minimum Huber loss corresponding to the propagation hypothesis H.

[0108] The assumption that the minimum Huber loss is smaller under the two assumptions is taken as the initial selection mode, and the corresponding state vector estimate is taken as the initial selection result. .

[0109] For the initial selection results, calculate the maximum absolute residual of all elements in the available measurement set, expressed as:

[0110] in, This indicates that the maximum value is determined based on all elements in the available measurement set. This indicates the coordinates of the detected object and the common deviation of the system in the preliminary selection results. This indicates the preset tolerance threshold, typically 0.1m.

[0111] If the maximum absolute residual is less than the preset tolerance threshold, the positioning result and propagation mode are accepted, and the initial selection result is taken as the final positioning result of the current independent target. If the maximum absolute residual exceeds the tolerance threshold, the maximum absolute residual under another propagation assumption is calculated to see if it is less than the preset tolerance threshold. If the maximum absolute residual under both propagation assumptions exceeds the tolerance threshold, the target is determined to be an invalid target and discarded, and no positioning result is output.

[0112] The reverse verification mechanism effectively suppresses positioning deviations caused by incorrect path associations or false target residues by applying strict residual consistency checks to the final positioning results, significantly improving the robustness and reliability of the algorithm.

[0113] Finally, the validated coordinate estimates for each individual target are output. Propagation mode identifier H*∈{LOS,NLOS} and system common deviation estimate .

[0114] like Figure 2 As shown, this invention establishes a simulation scenario and performs... Figure 3 The path identification shown results in the final localization. Figure 4 As shown.

[0115] This invention provides a specific embodiment of a joint positioning method based on radar multipath utilization. This embodiment uses a commercial 77 GHz MIMO radar (Texas Instruments AWR1843) operating in single-input single-output (SISO) mode to verify the actual performance of this invention under minimal hardware requirements.

[0116] Specifically, the experimental equipment and general parameters are set as follows.

[0117] Radar system: TI AWR1843, operating center frequency 77 GHz, sweep bandwidth 1.5 GHz, chirp duration 60 μs. Each frame contains 256 chirs, with 512 samples per chirp. The radar is deployed at the origin (0,0)(0,0) of an L-shaped corner.

[0118] Environment: L-shaped corridor, such as Figure 5 As shown, wall boundary parameters =0.69m, =2.73m.

[0119] Preprocessing: Mean subtraction is performed along the slow time dimension to suppress static clutter from walls and reflections from stationary objects. A one-dimensional range profile is extracted after range FFT. The sensing area is divided into a LOS corridor and an NLOS region. A single-channel radar is deployed at the origin (0,0) and simultaneously tracks two targets with different propagation characteristics: Target 1 (NLOS) is located at (1.3,1.8)m and is obscured by a corner; Target 2 (LOS) is located at (1.8,0.5)m and is within the radar's line of sight.

[0120] Based on the above experimental equipment and general parameters, this invention implements a joint positioning method based on radar multipath utilization. The specific steps are the same as described above and will not be repeated here.

[0121] This invention establishes as follows Figure 5 The real-world scenario shown, the final localization result is as follows: Figure 6 As shown.

[0122] The present invention also provides a joint positioning system based on radar multipath utilization, comprising: The peak distance extraction module is used to illuminate the object under test with radar, preprocess the received radar echo signal and extract multiple peak distances; the radar is located on one side of the L-shaped spatial region, and the object under test is located in the line-of-sight area of ​​the radar or the non-line-of-sight area of ​​the radar in the L-shaped spatial region. The candidate triplet construction module determines the candidate path distance triplet set by judging the midpoint of any two different peak distances among the multiple peak distances. The target set construction module is used to establish an initial target set based on the short path support of the path distances in the candidate path distance triplet set. The targets in the initial target set include path distances and corresponding path distance triplet sets. The path distance triplet sets of each target are sorted and conflict filtered to obtain the filtered target set. A usable measurement set construction module is used to perform master-slave judgment and filtering on any two targets in the filtered target set based on path distance to obtain multiple independent targets; determine a six-path group list for each independent target; and establish a usable measurement set for each independent target based on the path distance triplet set and the path distance in the six-path group list; the usable measurement set includes multiple path distances; The coordinate estimation module is used to determine the loss function and constraints based on the available measurement set of independent targets, perform least squares optimization based on the loss function and constraints, and perform line-of-sight type judgment and back-verification based on the optimized value, so as to finally obtain the coordinate estimation value of each independent target.

[0123] The present invention also provides a joint positioning center processing unit based on radar multipath utilization, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the operations performed by the joint positioning method based on radar multipath utilization.

[0124] While the specific embodiments of the present invention depict actions or steps in a particular order, this should be understood as requiring such actions or steps to be performed in the specific order shown or in sequential order, or requiring all illustrated actions or steps to be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.

[0125] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention 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 the present invention should be included within the scope of protection of the present invention.

Claims

1. A joint positioning method based on radar multipath utilization, characterized in that, Includes the following steps: Step S1: Illuminate the object under test with radar, preprocess the received radar echo signal and extract multiple peak distances; The radar is located on one side of the L-shaped spatial region, and the object being measured is located in the line-of-sight area or the non-line-of-sight area of ​​the radar within the L-shaped spatial region. Step S2: Based on the midpoint of any two different peak distances among the multiple peak distances, a set of candidate path distance triples is obtained; Step S3: Establish an initial target set based on the short path support of the path distances in the candidate path distance triplet set. The targets in the initial target set include path distances and corresponding path distance triplet sets. Sort and filter the path distance triplet sets of each target to obtain the filtered target set. Step S4: Based on path distance, perform master-slave judgment and filtering on any two targets in the filtered target set to obtain multiple independent targets; determine the six-path group list for each independent target; A usable measurement set is established for each independent target based on the path distance triplet set and the path distance in the six-path group list; the usable measurement set includes multiple path distances. Step S5: Determine the loss function and constraints based on the available measurement set of independent targets, perform least squares optimization based on the loss function and constraints, and perform line-of-sight type judgment and reverse verification based on the optimized values ​​to finally obtain the coordinate estimate of each independent target.

2. The joint positioning method based on radar multipath utilization according to claim 1, characterized in that, In step S1, the step of preprocessing the received radar echo signal and extracting multiple peak distances specifically includes: (1) Receive single-channel radar echo signals and obtain a one-dimensional range profile through matched filtering or range FFT processing; (2) Discretely sample and perform Hilbert transform on the one-dimensional distance profile to obtain the amplitude envelope sequence, detect the local maxima of the amplitude envelope sequence, and obtain multiple peak positions; (3) For multiple peak positions, sort the peak distances of the peak positions whose peak amplitude is greater than the median of the amplitude envelope sequence from smallest to largest to obtain the multiple peak distances.

3. The joint positioning method based on radar multipath utilization according to claim 2, characterized in that, Step S2 specifically includes: Step S2-1: Calculate the midpoint value for two different peaks in the observation distance vector; search among the multiple peak distances to find the peak closest to the midpoint value; Step S2-2: Sort the two different peaks and the peak closest to the midpoint value to form a valid triplet; The conditions for forming a valid triple are: ， in, The midpoint tolerance threshold, This indicates obtaining the absolute value. This represents the distance to the r-th peak. This represents the midpoint of the distance between the p-th and q-th peaks; Step S2-3: Return to step S2-1 until all two different peaks in the observation distance vector are traversed, and all valid triples are used to form a candidate triple set.

4. The joint positioning method based on radar multipath utilization according to claim 3, characterized in that, Step S3 specifically includes: Step S3-1: For each peak index in the candidate path distance triplet set, calculate the corresponding short path support; the expression for the short path support is: in, This represents the support of the i-th peak path. This represents the distance of the i, j, k-th path. This represents the number of triples with i as the shortest path index; Step S3-2: Construct a short path occurrence index set by indexing all peak paths with a support greater than or equal to 1; For each index in the set of short path occurrence indexes, construct the initial target structure, expressed as: in, Let i' represent the i'-th initial target structure. This represents the path distance to the i'th initial target structure. This represents the set of triplets representing the path distances to the i'-th initial target structure. This represents the set of peak path indices with a support greater than or equal to 1. Step S3-3: For each initial target structure , will the corresponding Sort by midpoint consistency error from smallest to largest. The expression for midpoint consistency error is: in, This represents the midpoint consistency error of the path distances for the i, j, and k-th paths. Step S3-4: For each sorted triplet of the initial target structure, perform greedy filtering, including recording the used middle path index set and long path index set respectively, and traversing the sorted triplets in turn. If the index of the current triplet is... and index If it is not occupied, the triplet will be retained, and... and Mark as occupied; otherwise discard. Finally, the filtered target structures are obtained, and all the filtered target structures constitute the filtered target set.

5. The joint positioning method based on radar multipath utilization according to claim 4, characterized in that, Step S4 specifically includes: Step S4-1: For two targets in the filtered target set Check if two distinct triples exist. and and a triple This makes the following proximity condition true: in, Indicates the merged tolerance. and The longer path distance between the two triples of the candidate primary target is given. The shortest path distance to the candidate false target. The long path distance in the candidate false target triple; Step S4-2: If the proximity condition is met, then determine... To be subordinate to the main objective False targets, for Establish a six-path group; the six-path group includes six different radar paths to the object being measured; Step S4-3: Return to step S4-1 until any two targets have been processed, and finally obtain a list of each independent target and its associated six-path groups; Step S4-4: For each independent target, the observed distance values ​​corresponding to all peak indices in the corresponding filtered triplet set are summarized with all distance measurements contained in each group in the six-path group list to obtain the available measurement set, which includes multiple path distances.

6. The joint positioning method based on radar multipath utilization according to claim 5, characterized in that, In step S5, the loss function expression is: in, This represents the minimum Huber loss corresponding to the propagation hypothesis H. This represents the Huber loss function. This represents the input residual of the Huber loss function. This is a transition parameter, with a value range of 0.01m to 0.1m; Indicates measurement residuals, Represents the theoretical distance function. Represents the state vector. For the target coordinates, For system common deviation, This represents the distance of the k-th path; The expression for the constraint condition is: in, This represents the constraint condition corresponding to the propagation hypothesis H, where LOS and NLOS represent line-of-sight and non-line-of-sight, respectively. and This indicates the corridor width parameter. Represents the ordinate of the corner point. This represents the minimum value of the ordinate of the detected object. These represent the minimum and maximum values ​​of the system's common deviation, respectively.

7. The joint positioning method based on radar multipath utilization according to claim 6, characterized in that, In step S5, the step of determining the line-of-sight type based on the optimized value and performing reverse verification to finally obtain the coordinate estimate of each independent target specifically includes: Solving the constrained optimization problem under both LOS and NLOS assumptions yielded two sets of optimization results. Calculate the minimum Huber loss under each of the two assumptions; take the assumption with the smaller minimum Huber loss under the two assumptions as the initial selection mode, and obtain the corresponding state vector estimate as the initial selection result; For the initial selection results, calculate the maximum absolute residual of all elements in the available measurement set, expressed as: in, This indicates that the maximum value is determined based on all elements in the available measurement set. Let represent the available measurement set for the i-th independent target. express The elements in This indicates the coordinates of the detected object and the common deviation of the system in the preliminary selection results. This indicates a preset tolerance threshold; If the maximum absolute residual is less than the preset tolerance threshold, the initial selection result will be used as the final location result of the current independent target. If the maximum absolute residual exceeds the tolerance threshold, then calculate whether the maximum absolute residual under another propagation hypothesis is less than the preset tolerance threshold. If the maximum absolute residual under another propagation hypothesis is less than the preset tolerance threshold, then use the state vector under the other propagation hypothesis as the final localization result of the current independent target. The final positioning result is output, which includes the coordinate estimate of each independent target and the estimate of the system common deviation.

8. A joint positioning system based on radar multipath utilization, characterized in that, include: The peak distance extraction module is used to illuminate the object under test with radar, preprocess the received radar echo signal and extract multiple peak distances; the radar is located on one side of the L-shaped spatial region, and the object under test is located in the line-of-sight area of ​​the radar or the non-line-of-sight area of ​​the radar in the L-shaped spatial region. The candidate triplet construction module determines the candidate path distance triplet set by judging the midpoint of any two different peak distances among the multiple peak distances. The target set construction module is used to establish an initial target set based on the short path support of the path distances in the candidate path distance triplet set. The targets in the initial target set include path distances and corresponding path distance triplet sets. The path distance triplet sets of each target are sorted and conflict filtered to obtain the filtered target set. A usable measurement set construction module is used to perform master-slave judgment and filtering on any two targets in the filtered target set based on path distance to obtain multiple independent targets; determine a six-path group list for each independent target; and establish a usable measurement set for each independent target based on the path distance triplet set and the path distance in the six-path group list; the usable measurement set includes multiple path distances; The coordinate estimation module is used to determine the loss function and constraints based on the available measurement set of independent targets, perform least squares optimization based on the loss function and constraints, and perform line-of-sight type judgment and back-verification based on the optimized value, so as to finally obtain the coordinate estimation value of each independent target.

9. A joint positioning center processing unit based on radar multipath utilization, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the operations performed by the joint positioning method based on radar multipath utilization as described in any one of claims 1 to 7.

Images