Doppler radar static fusion smooth variable structure filtering method and device

Abstract

The present application discloses a Doppler radar static fusion smooth variable structure filtering method and device, wherein the method comprises the following steps: using a first generalized smooth variable structure filter to update and estimate the target state, and obtain a posterior estimate of the target state; using a second generalized smooth variable structure filter to update and estimate the target pseudo state, and obtain a posterior estimate of the target pseudo state; using a static fuser to fuse the posterior estimate of the target state and the posterior estimate of the target pseudo state, and obtain the final posterior state estimate of the target. The present application solves the underdetermination problem caused by the nonlinear observation model and the non-full rank observation matrix while maintaining the robust estimation performance under the condition of model uncertainty, thereby improving the target tracking performance of the Doppler radar system. It can be used for the nonlinear observation model of the Doppler radar, and realizes robust radar target state estimation under the condition of target motion model uncertainty.

Description

Technical Field

[0001] The present application relates to the field of radar data processing technology, and in particular to a Doppler radar static fusion smooth variable structure filtering method and device. Background Art

[0002] The radar target tracking system uses the target distance, azimuth, Doppler frequency deviation and other information contained in the radar echo to continuously and effectively estimate the target quantity and motion state such as position, velocity, acceleration, etc. Radar tracking systems are widely used in the fields of autonomous driving, air traffic control, weather monitoring, security, national defense and military. Among them, Doppler radar can observe the radial velocity information of target motion, increase the effective information dimension in the radar target state estimation link, and thus greatly improve the target tracking performance.

[0003] Doppler radar target tracking algorithms and systems first need to establish radar observation models and target state transition models, which face two important challenges: 1) Highly nonlinear observation models, including distance, azimuth, pitch angle, and radial velocity; highly nonlinear models will cause common radar tracking methods such as Kalman filters (KF) to produce large nonlinear system errors, and even cause filter divergence. 2) Uncertainty of the target state transition model; traditional Bayesian filtering methods, including KF, rely on an accurate description of the target motion model; if the system state equation deviates greatly from the motion model of the real target, such as when the target undergoes strong maneuvering motion, it will cause the tracking error to increase sharply, or even cause the filter to diverge and lose the target.

[0004] Methods used to solve problem 1) of nonlinear observation models include a series of nonlinear filtering methods, such as extended Kalman filter (EKF), insensitive Kalman filter (UKF), particle filter (PF), etc. In particular, for the mathematical model of Doppler radar, the static fusion Kalman filter (SF-CMKF) method decouples the nonlinear position measurement and radial velocity measurement, establishes the target state equation and pseudo-state equation respectively, and uses the static fusion method to avoid the transmission and accumulation of nonlinear truncation errors in the iteration process, thereby significantly improving the estimation accuracy and is widely welcomed. However, these traditional Bayesian methods, including the static fusion Kalman filter and its numerous improved forms, cannot solve the uncertainty problem of the target state transfer model.

[0005] Methods used to solve problem 2) the uncertainty of the target state transition model include robust Kalman filtering, H-infinite filtering, and the smooth variable structure filter (SVSF) that has attracted much attention in recent years. In particular, the smooth variable structure filter (SVSF) is a robust filtering method under model uncertainty. Based on the sliding mode control theory and the variable structure control method, it can ensure the boundedness of the state estimation error and has low computational complexity. However, it is worth noting that the standard smooth variable structure filter method requires a linear observation model and a full-rank observation matrix to achieve a bijective mapping from the target state space to the observation space. In order to solve the problem of nonlinear observation models, related technologies attempt to use the Jacobian matrix instead of the nonlinear function to substitute into the standard smooth variable structure filter structure, but this will introduce a new problem of inverting the underdetermined matrix, resulting in a large error in the estimation of the target velocity. Therefore, the existing smooth variable structure filter method is still not applicable to Doppler radar systems. Summary of the invention

[0006] The present application provides a Doppler radar static fusion smooth variable structure filtering method and device, which solves the nonlinear underdetermined observation model problem of the Doppler radar and the robust tracking problem under the uncertainty of the motion model of the radar target. The method and device can be used for the nonlinear observation model of the Doppler radar and realize robust radar target state estimation under the uncertainty of the target motion model.

[0007] In a first aspect, an embodiment of the present application provides a Doppler radar static fusion smooth variable structure filtering method, comprising the following steps: using a first generalized smooth variable structure filter to update and estimate a target state to obtain a posterior estimate of the target state; using a second generalized smooth variable structure filter to update and estimate a target pseudo-state to obtain a posterior estimate of the target pseudo-state; using a static fuser to fuse the target state posterior estimate and the target pseudo-state posterior estimate to obtain a final posterior state estimate of the target.

[0008] In one embodiment of the present application, the use of the first generalized smoothly variable structure filter to update the estimate of the target state to obtain a posterior estimate of the target state includes: using an unbiased measurement conversion method to convert the nonlinear position measurement of the Doppler radar in polar coordinates or spherical coordinates to a Cartesian coordinate system to obtain a linear position measurement and its conversion deviation and conversion measurement error covariance; using the first generalized smoothly variable structure filter and the linear position measurement and its conversion deviation and conversion measurement error covariance to obtain the posterior estimate of the target state.

[0009] In one embodiment of the present application, the target state posterior estimate is obtained by utilizing the first generalized smooth variable structure filter and the linear position measurement and its conversion deviation and conversion measurement error covariance, including: calculating the target state prior estimate, radar measurement prior estimate and prior state estimate error covariance matrix at time k+1; calculating position measurement information, and calculating a first mixed error term and a first innovation gain term based on the position measurement information; calculating the target state posterior estimate and its covariance matrix based on the target position measurement.

[0010] In one embodiment of the present application, the target pseudo-state is updated and estimated using a second generalized smoothly variable structure filter to obtain a posterior estimate of the target pseudo-state, including: constructing a distance-radial velocity product pseudo-state vector and a pseudo-measurement variable; using an unbiased measurement conversion method to calculate the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable; and using the second generalized smoothly variable structure filter and the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable to obtain a posterior estimate of the target pseudo-state.

[0011] In one embodiment of the present application, the target pseudo-state posterior estimate is obtained by utilizing the conversion deviation, measurement error variance and covariance of the second generalized smooth variable structure filter and the pseudo-measurement variable, including: calculating the target pseudo-state a priori estimate, the radar pseudo-measurement a priori estimate and the pseudo-state a priori estimate covariance matrix at time k+1; calculating the pseudo-measurement innovation, and calculating the second mixed error term and the second innovation gain term based on the pseudo-measurement innovation; calculating the target pseudo-state posterior estimate and its covariance matrix.

[0012] In one embodiment of the present application, the target state posterior estimate and its covariance matrix output by the first generalized smooth variable structure filter and the target pseudo-state posterior estimate and its covariance matrix output by the second generalized smooth variable structure filter are utilized, and a static fuser based on the linear minimum mean square error criterion is adopted to calculate the target final posterior state estimate and its covariance.

[0013] In a second aspect, an embodiment of the present application provides a Doppler radar static fusion smooth variable structure filtering device, comprising: a first estimation module, used to use a first generalized smooth variable structure filter to update and estimate the target state, and obtain a posterior estimate of the target state; a second estimation module, used to use a second generalized smooth variable structure filter to update and estimate the target pseudo-state, and obtain a posterior estimate of the target pseudo-state; a fusion module, used to use a static fuser to fuse the target state posterior estimate and the target pseudo-state posterior estimate, and obtain a final posterior state estimate of the target.

[0014] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the Doppler radar static fusion smooth variable structure filtering method as described in the above embodiment.

[0015] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the Doppler radar static fusion smooth variable structure filtering method as described in the above embodiment.

[0016] The Doppler radar static fusion smooth variable structure filtering method and device of the embodiment of the present application can significantly improve the accuracy of Doppler radar target state estimation, ensure the filtering robustness under model uncertainty conditions, and have good application value, specifically:

[0017] 1) Good robustness of target tracking

[0018] The static fusion smooth variable structure filtering method of the present application is based on the sliding mode variable structure control theory, which can ensure the boundedness of the tracking error under the condition of unknown target motion model, and solves the problem that the tracking filtering error of traditional nonlinear Bayesian filters (such as Kalman filters and static fusion Kalman filters, etc.) increases sharply or even filters diverge when there is a modeling error in the target state transfer model. In the first generalized smooth variable structure filter (CPM-SVSF), the method of the present application uses an unbiased conversion method to convert the position measurement of the Doppler radar into a Cartesian coordinate system, so that a linear classical smooth variable structure filter can be constructed to achieve robust estimation of the target state; in the second generalized smooth variable structure filter (CDM-SVSF), the present application uses radial velocity and its derivatives to construct linear pseudo-state equations and pseudo-measurement equations, thereby using a linear classical smooth variable structure filter to achieve robust estimation of the target pseudo-state; in static fusion processing, the minimum mean square error criterion strategy is used to fuse the robust estimate of the target state output by the CPM-SVSF and the robust estimate of the target pseudo-state output by the CDM-SVSF to obtain the final state estimate of the target, thereby ensuring the robust state estimation performance under model uncertainty conditions.

[0019] 2) High target tracking accuracy

[0020] The static fusion smooth variable structure filtering method proposed in this application can significantly improve the tracking accuracy, which comes from two advantages:

[0021] First, the static fusion smooth variable structure filtering method proposed in the present application solves the underdetermination problem of the Doppler radar nonlinear model and non-full rank observation matrix faced by the existing smooth variable structure filtering method, thereby making full use of the radial velocity measurement information of the Doppler radar to effectively improve the target state estimation accuracy of the smooth variable structure filter. Specifically, the static fusion smooth variable structure filter solves the problem that the standard smooth variable structure filter cannot utilize nonlinear Doppler measurement, and also solves the underdetermination problem of inverting the non-full rank observation matrix faced by the improved smooth variable structure filter in the existing literature when using the Jacobian matrix local linearization operation. Therefore, compared with the existing smooth variable structure filtering method, the method of the present application more effectively utilizes the Doppler velocity information, thereby significantly improving the accuracy of target state tracking.

[0022] Second, the static fusion smooth variable structure filtering method proposed in the present application adopts the method of static fusion after decoupling processing of position measurement and Doppler measurement and parallel robust filtering, which avoids the transmission and accumulation of truncation errors caused by nonlinear measurement in the iterative estimation process. Specifically, in order to reduce the influence of nonlinear errors, the first generalized smooth variable structure filter (CPM-SVSF) and the second generalized smooth variable structure filter (CDM-SVSF) are modeled as linear state / pseudo-state equations and measurement / pseudo-measurement equations, and the iterative estimation process is only completed inside the linear sub-filter, and there is no transmission and accumulation process of nonlinear system errors. Therefore, compared with the existing nonlinear smooth variable structure filtering method, the method of the present application effectively reduces the influence of nonlinear system errors, thereby significantly improving the tracking accuracy of the target state.

[0023] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0025] Figure 1 A schematic diagram of a flow chart of a Doppler radar static fusion smooth variable structure filtering method provided according to an embodiment of the present application;

[0026] Figure 2 A flow chart of a Doppler radar static fusion smooth variable structure filtering method provided according to an embodiment of the present application;

[0027] Figure 3 This is a diagram showing an example of a simulation scenario of a 2-D Doppler millimeter wave radar tracking a vehicle target in an autonomous driving scenario according to an embodiment of the present application;

[0028] Figure 4 A schematic diagram of the target state estimation error of the method of the present application changing over time in a simulation scenario provided according to an embodiment of the present application; Figure 4 (a) is the coordinate estimation error, Figure 4 (b) is the speed estimation error;

[0029] Figure 5 This is an example diagram of a Doppler radar static fusion smooth variable structure filtering device according to an embodiment of the present application;

[0030] Figure 6 A schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0032] Figure 1 The present invention is a flowchart of a Doppler radar static fusion smooth variable structure filtering method provided according to an embodiment of the present application.

[0033] In an embodiment of the present application, a static fusion smooth variable structure filter (SF-SVSF) includes two parallel sub-filters (a first generalized smooth variable structure filter and a second generalized smooth variable structure filter) and a static fuser. In the first generalized smooth variable structure filter (referred to as CPM-SVSF), the nonlinear position measurement (including target distance, azimuth, and pitch angle) of the Doppler radar is first used to unbiasedly transform to a Cartesian coordinate system, and then the linear position measurement in the Cartesian coordinate system is used to construct a linear first generalized smooth variable structure filter, and the target state estimation based on the position measurement is iteratively updated. In the second generalized smooth variable structure filter (referred to as CDM-SVSF), the radial velocity measurement of the Doppler radar is first used to construct a distance-radial velocity product pseudo-state equation and a pseudo-observation equation, and on this basis, the linear second generalized smooth variable structure filter is used to iteratively update the pseudo-state estimate. The static fuser fuses the target state posterior estimate and the target pseudo-state posterior estimate to obtain the final state estimate of the target.

[0034] like Figure 1 As shown, the Doppler radar static fusion smooth variable structure filtering method includes the following steps:

[0035] In step S101, a first generalized smooth variable structure filter is used to update and estimate the target state to obtain a posterior estimate of the target state.

[0036] In one embodiment of the present application, a first generalized smoothly variable structure filter is used to update an estimate of a target state to obtain a posterior estimate of the target state, including: using an unbiased measurement conversion method to convert the nonlinear position measurement of a Doppler radar in polar coordinates or a spherical coordinate system to a Cartesian coordinate system to obtain a linear position measurement and its conversion deviation and conversion measurement error covariance; using the first generalized smoothly variable structure filter and the linear position measurement and its conversion deviation and conversion measurement error covariance to obtain a posterior estimate of the target state.

[0037] Specifically, the first generalized smooth variable structure filter (CPM-SVSF) is used to update the target state. First, the target position measurement of the Doppler radar in polar coordinates or spherical coordinates is converted to the Cartesian coordinate system using the unbiased measurement conversion method, and then the generalized smooth variable structure filter and the unbiased converted position measurement are used to obtain the target state estimation. The specific steps are:

[0038] Step 1-1: Use the unbiased measurement conversion (UCM) method to convert the nonlinear position measurement in Doppler radar spherical coordinates or polar coordinates (representing distance, azimuth and elevation respectively) converted into linear position measurement in Cartesian coordinate system and its conversion deviation μ p =[μ x ,μ y ,μ z ] T and the transformed measurement error covariance R p (k) The UCM method may refer to the method in the relevant technical literature and is not specifically limited.

[0039] In one embodiment of the present application, a posterior estimate of the target state is obtained using a first generalized smooth variable structure filter and a linear position measurement and its conversion deviation and conversion measurement error covariance, including: calculating a priori estimate of the target state at time k+1, a radar measurement priori estimate and a priori state estimate error covariance matrix; calculating position measurement information, and calculating a first mixed error term and a first information gain term based on the position measurement information; calculating a posterior estimate of the target state based on the target position measurement and its covariance matrix.

[0040] Specifically, the target position after the unbiased transformation is measured As input, the following generalized smooth variable structure filter is used to obtain the target motion state estimation, that is, the target motion state estimation vector and the estimated error covariance matrix P p (k|k). The specific steps are as follows:

[0041] Step 1-2: Calculate the prior estimate of the target state at time k+1 Radar Measurement Prior Estimation and the prior state estimation error covariance matrix P p (k|k-1):

[0042]

[0043]

[0044]

[0045] in is the state estimation variable containing three-dimensional position and velocity, is the preset state transfer matrix, is the observation matrix, Q and Γ p is the process noise covariance and its coefficient matrix. and Γ p They are defined as:

[0046]

[0047] It should be noted that, unlike traditional Bayesian filters (such as SF-CMKF), which require an accurately known target state transition model F p , here Compared with the true value, the model error with bounded norm is allowed, that is, the method of this application has the advantage of robustness under model uncertainty; this application introduces the step of unbiased measurement conversion, so It is linear and time-invariant, which is superior to the traditional smooth variable structure filter that directly uses nonlinear position measurement.

[0048] Step 1-3: Calculate position measurement information:

[0049]

[0050] Compute the first mixed error term:

[0051]

[0052]

[0053] where e p (k-1|k-1) represents the a posteriori measurement error of subfilter 1 at the previous moment, ABS It means taking the absolute value of each element of the vector, γ p,z ∈i 3×1 and γ p,y ∈i 3×1 is the attenuation factor with a value in the interval (0,1); Represents a diagonal matrix whose main diagonal element is vector a; Where T pis the transformation matrix, which can be chosen as the identity matrix in this method.

[0054] Calculate the first innovation gain term K based on position measurement p (k):

[0055]

[0056]

[0057]

[0058] Where sat(.) represents the saturation function; Ψ p,z ∈i 3×1 and p,y ∈i 3×1 are the preset smoothing layer parameter vectors; H p,1 =I 3×3 In step 1-2 The full-rank block submatrix of is linearly time-invariant:

[0059] Steps: 1-4: Calculate the posterior estimate of the target state and the state covariance matrix P p (k|k):

[0060]

[0061]

[0062] in, is the unit matrix, R p (k) is the position measurement error covariance obtained in step 1-1.

[0063] The first generalized smooth variable structure filter (CPM-SVSF) outputs the target state estimate and its covariance matrix P p (k|k).

[0064] Step S102: using a second generalized smooth variable structure filter to update and estimate the target pseudo state to obtain a posterior estimation of the target pseudo state.

[0065] In an embodiment of the present application, a second generalized smoothly variable structure filter is used to update and estimate the target pseudo-state to obtain a posterior estimate of the target pseudo-state, including: constructing a distance-radial velocity product pseudo-state vector and a pseudo-measurement variable; using an unbiased measurement conversion method to calculate the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable; using the second generalized smoothly variable structure filter and the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable to obtain a posterior estimate of the target pseudo-state.

[0066] To update the state of the second generalized smooth variable structure filter, first, define the pseudo state vector:

[0067]

[0068] where the scalar η(k) is the range-radial velocity product, defined as The corresponding pseudo-measurement is defined as The unbiased measurement conversion (UCM) method is used to calculate the conversion deviation μ of the pseudo measurement variable. η , measurement error variance R η and covariance R pη ; The calculation method can use the method in the relevant technology without specific limitation.

[0069] In an embodiment of the present application, a second generalized smoothly variable structure filter and a conversion deviation of a pseudo-measurement variable, a measurement error variance and a covariance are used to obtain a posterior estimate of a target pseudo-state, including: calculating a priori estimate of the target pseudo-state at time k+1, a radar pseudo-measurement prior estimate and a covariance matrix of a priori estimate of the pseudo-state; calculating pseudo-measurement innovation, and calculating a second mixed error term and a second innovation gain term based on the pseudo-measurement innovation; calculating a posterior estimate of the target pseudo-state and its covariance matrix.

[0070] Specifically, the unbiased transformed pseudo-measurement z η As input, the target pseudo-state estimate is obtained using the following second generalized smooth variable structure filter and the estimated error covariance matrix P η (k|k). The specific steps are as follows:

[0071] Step 2-1: Calculate the target pseudo-state prior estimate at time k+1 A priori estimation of radar pseudo-measurements and the pseudo-state prior estimate covariance matrix P η (k|k-1):

[0072]

[0073]

[0074]

[0075] in is a pseudo-state estimation variable, is the preset pseudo-state transfer matrix, G η and u η is the input coefficient matrix and input term introduced by the high-order process noise, H η is the pseudo-measurement matrix, Q x and Γ xis the process noise covariance and its coefficient matrix introduced related to the target state, Q s and Γ s is the process noise covariance and its coefficient matrix related to the high-order process noise. They are defined as follows:

[0076]

[0077]

[0078] Where q is the variance of the process noise (Gaussian white noise) in the Cartesian coordinate system, and T is the radar scanning period.

[0079] It should be noted that, unlike traditional Bayesian filters (such as SF-CMKF), which require an accurate known target pseudo-state transition model F η , here Compared with the true value, the model error with bounded norm is allowed, that is, the method of the present application has the advantage of robustness under model uncertainty; the method of the present application introduces the step of unbiased measurement conversion, so H η (k) = [1, 0] is linear time-invariant and is superior to the conventional smooth variable structure filter that directly uses the nonlinear Doppler measurement.

[0080] Step 2-2: Calculate pseudo-measurement information:

[0081]

[0082] Compute the second mixed error term:

[0083] E η,z =|e η (k|k-1)|+γ η,z |e η (k-1|k-1)|

[0084]

[0085] where e η (k-1|k-1) represents the a posteriori measurement error of the second generalized smooth variable structure filter at the previous moment, γ η,z and γ η,y is the attenuation factor with a value in the interval (0,1); Where T η is a transformation matrix, which can be selected as a unit matrix in the method of this application.

[0086] Calculate the second innovation gain term K based on pseudo-measurement η (k):

[0087]

[0088] K η,u (k) = H η,1 -1 ·sat(e η (k|k-1),Ψ η,z )·E η,z ·e p (k|k-1) -1

[0089]

[0090] Where sat(.) represents the saturation function; Ψ η,z and η,y are the preset smoothing layer parameter vectors; H η,1 =1 is H in step 2-1 η The full-rank block submatrix of is linear time invariant.

[0091] Step 2-3: Calculate the posterior estimate of the target state and the state covariance matrix P η (k|k):

[0092]

[0093] P η (k|k)=[IK η (k)H η (k)]P η (k|k-1)[IK η (k)H η (k)] T +K η (k)R η (k)K η (k) T

[0094] in, is the unit matrix, R η (k) is the pseudo measurement variance obtained in the above steps.

[0095] Output pseudo-state estimation of the second generalized smooth variable structure filter (CDM-SVSF) and its covariance P η (k|k).

[0096] In step S103, a static fuser is used to fuse the target state posterior estimate and the target pseudo-state posterior estimate to obtain a final target posterior state estimate.

[0097] In an embodiment of the present application, a posterior estimate of the target state and its covariance matrix output by the first generalized smooth variable structure filter and a posterior estimate of the target pseudo-state and its covariance matrix output by the second generalized smooth variable structure filter are utilized, and a static fuser based on the linear minimum mean square error criterion is adopted to calculate the final posterior state estimate of the target and its covariance.

[0098] The static fusion estimator is a local approximate linear minimum mean square error estimator. It uses the output fusion processing of the two generalized smooth variable structure filters to obtain the final estimate of the target state in the current frame. The above process is iteratively updated to achieve continuous tracking of the target state.

[0099] Specifically, the outputs of the first generalized smooth variable structure filter (CPM-SVSF) and the second generalized smooth variable structure filter (CDM-SVSF) are used as input, and the static fuser is used to obtain the final estimate of the target motion state at the current moment. and P(k|k), which are calculated as follows:

[0100]

[0101]

[0102] in, and P p (k|k) is the state estimate and its covariance of the output of the first generalized smooth variable structure filter (CPM-SVSF), is the pseudo-state estimate of the output of the first generalized smooth variable structure filter (CDM-SVSF). The pseudo-state mean E(η(k)), the cross-correlation matrix P of the static fuser xz And the autocorrelation matrix P zz The calculation method of can utilize the calculation method in the related technology, and is not specifically limited.

[0103] For the k=1, 2, ... radar scanning echoes, the iterative estimation process of steps S101 to S103 is repeated until the Doppler radar system determines that the track ends, thereby realizing nonlinear robust estimation of the Doppler radar target state under model uncertainty conditions.

[0104] The Doppler radar static fusion smooth variable structure filtering method according to the embodiment of the present application is described below with reference to the accompanying drawings and specific embodiments.

[0105] like Figure 2As shown in the figure, the execution steps of the Doppler radar static fusion smooth variable structure filtering method are shown, where subfilter 1 is the first generalized smooth variable structure filter and subfilter 2 is the second generalized smooth variable structure filter. Maneuvering target tracking is a typical scenario of model uncertainty problems. Consider the simulation scenario of 2-D Doppler millimeter wave radar tracking filtering of a single maneuvering target in an autonomous driving scenario. The radar is established at the origin of the coordinate system, and the target maneuvering trajectory is as follows: Figure 3 As shown. The embodiment of the present application intends to compare the static fusion smooth variable structure filtering method (denoted as SF-SVSF) proposed in the present application, and the extended Kalman filter method (EKF), the smooth variable structure filtering method using only position measurement (PO-SVSF), and the smooth variable structure filtering method (denoted as L-SVSF) using Jacobian matrix local linearization in the existing literature. The average result of 2000 Monte Carlo experiments is used as the comparison data.

[0106] The simulated radar target data is generated using the simulation parameters in Table 1. The target state variables are defined as the position and velocity in the XY coordinate axis, that is, x = [xyv x v y ] T ; The target pseudo-state variable is defined as Radar Measurement Variable Definition In sub-filter 1 (CPM-SVSF), a two-dimensional uniform motion model (CV) is used to estimate the target state, namely:

[0107]

[0108] The relevant settings of sub-filter 2 (CDM-SVSF) are set by the method described in the above steps.

[0109] Table 1 Radar target tracking scenario parameters

[0110]

[0111] The target state is iteratively estimated according to the steps described in the above embodiment.

[0112] Figure 4 The variation of the coordinate error and velocity error of the target state estimation with the scanning time is given, where: Figure 4 The L-SVSF method in (b) faces the numerical stability problem of underdetermined matrix inversion, which causes the velocity estimation error to exceed the display range. Table 2 gives the root mean square error (RMSE) of 2000 Monte Carlo experiments. Figure 4It can be seen that the static fusion smooth variable structure filter (SF-SVSF) method proposed in this application has achieved the best tracking accuracy. Compared with the extended Kalman filter (EKF), the SF-SVSF method has more robust tracking performance when the target is maneuvering (10-15s and 23-28s, the peak coordinate error is reduced by more than half). Compared with the PO-SVSF that only uses position measurement, the proposed SF-SVSF can effectively utilize the Doppler velocity information and greatly improve the target state estimation accuracy. The local linearization L-SVSF in the existing literature faces the underdetermination problem caused by the non-full rank observation matrix, and the target state estimation performance (especially the velocity estimation performance) is the worst. In summary, the method of this application solves the problem of nonlinear underdetermined observation model of Doppler radar. Under the condition of uncertainty of the target motion model, it can make full use of the Doppler velocity measurement information to achieve more robust and accurate radar target state estimation.

[0113] Table 2 Root mean square error of state estimation for uncertain targets in motion model

[0114]

[0115] According to the Doppler radar static fusion smooth variable structure filtering method proposed in the embodiment of the present application, the state space static fusion strategy is combined with the smooth variable structure filtering method, which solves the problem that the standard smooth variable structure filter cannot be applied to the Doppler radar system with nonlinear and non-full rank observation matrix, and solves the problem that the traditional nonlinear Bayesian filter is easy to filter diverge under the condition of uncertainty of the target motion model; in addition, the structure of target position measurement and Doppler measurement decoupling processing, parallel filtering and static fusion can avoid the transmission and accumulation of nonlinear truncation error in the iterative process, thereby further reducing the estimation error. Therefore, the present application can effectively use Doppler observation information to improve the tracking accuracy of the target state, and maintain robust estimation performance under the condition of uncertainty of the target motion model, and has a high engineering application value.

[0116] Next, a Doppler radar static fusion smooth variable structure filtering device proposed according to an embodiment of the present application is described with reference to the accompanying drawings.

[0117] Figure 5 It is a block diagram of a Doppler radar static fusion smooth variable structure filtering device according to an embodiment of the present application.

[0118] like Figure 5 As shown, the Doppler radar static fusion smooth variable structure filter device 10 includes: a first estimation module 100, a second estimation module 200 and a fusion module 300.

[0119] The first estimation module 100 is used to update the target state using the first generalized smooth variable structure filter to obtain the target state posterior estimate. The second estimation module 200 is used to update the target pseudo state using the second generalized smooth variable structure filter to obtain the target pseudo state posterior estimate. The fusion module 300 is used to fuse the target state posterior estimate and the target pseudo state posterior estimate using a static fuser to obtain the target final posterior state estimate.

[0120] In one embodiment of the present application, the first estimation module 100 is further used to convert the nonlinear position measurement of the Doppler radar in polar coordinates or spherical coordinates into a Cartesian coordinate system using an unbiased measurement conversion method to obtain a linear position measurement and its conversion deviation and conversion measurement error covariance; and obtain a posterior estimate of the target state using a first generalized smooth variable structure filter and the linear position measurement and its conversion deviation and conversion measurement error covariance.

[0121] In one embodiment of the present application, the first estimation module 100 is specifically used to calculate the target state prior estimate, the radar measurement prior estimate and the prior state estimation error covariance matrix at time k+1; calculate the position measurement information, and calculate the first mixed error term and the first information gain term based on the position measurement information; calculate the target state posterior estimate and its covariance matrix based on the target position measurement.

[0122] In one embodiment of the present application, the second estimation module 200 is further used to construct a range-radial velocity product pseudo-state vector and a pseudo-measurement variable; use an unbiased measurement conversion method to calculate the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable; and use a second generalized smooth variable structure filter and the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable to obtain a posterior estimate of the target pseudo-state.

[0123] In one embodiment of the present application, the second estimation module 200 is specifically used to calculate the target pseudo-state prior estimate, the radar pseudo-measurement prior estimate and the pseudo-state prior estimate covariance matrix at time k+1; calculate the pseudo-measurement innovation, and calculate the second mixed error term and the second innovation gain term based on the pseudo-measurement innovation; calculate the target pseudo-state posterior estimate and its covariance matrix.

[0124] In one embodiment of the present application, the fusion module 300 is further used to use the target state posterior estimate and its covariance matrix output by the first generalized smooth variable structure filter and the target pseudo-state posterior estimate and its covariance matrix output by the second generalized smooth variable structure filter, and use a static fuser based on the linear minimum mean square error criterion to calculate the target final posterior state estimate and its covariance.

[0125] It should be noted that the above explanation of the Doppler radar static fusion smooth variable structure filtering method embodiment is also applicable to the Doppler radar static fusion smooth variable structure filtering device of this embodiment, which will not be repeated here.

[0126] According to the Doppler radar static fusion smooth variable structure filter device proposed in the embodiment of the present application, the state space static fusion strategy is combined with the smooth variable structure filter method, which solves the problem that the standard smooth variable structure filter cannot be applied to the Doppler radar system with nonlinear and non-full rank observation matrix, and solves the problem that the traditional nonlinear Bayesian filter is easy to filter divergence under the condition of uncertainty of the target motion model; in addition, the structure of target position measurement and Doppler measurement decoupling processing, parallel filtering and static fusion can avoid the transmission and accumulation of nonlinear truncation error in the iterative process, thereby further reducing the estimation error. Therefore, the present application can effectively use Doppler observation information to improve the tracking accuracy of the target state, and maintain robust estimation performance under the condition of uncertainty of the target motion model, and has a high engineering application value.

[0127] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0128] A memory 601 , a processor 602 , and a computer program stored in the memory 601 and executable on the processor 602 .

[0129] When the processor 602 executes the program, the Doppler radar static fusion smooth variable structure filtering method provided in the above embodiment is implemented.

[0130] Furthermore, the electronic device further comprises:

[0131] The communication interface 603 is used for communication between the memory 601 and the processor 602 .

[0132] The memory 601 is used to store computer programs that can be executed on the processor 602 .

[0133] The memory 601 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0134] If the memory 601, the processor 602 and the communication interface 603 are implemented independently, the communication interface 603, the memory 601 and the processor 602 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0135] In a specific implementation, if the memory 601, the processor 602 and the communication interface 603 are integrated on a chip, the memory 601, the processor 602 and the communication interface 603 can communicate with each other through an internal interface.

[0136] The processor 602 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0137] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above Doppler radar static fusion smooth variable structure filtering method is implemented.

[0138] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0139] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0140] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0141] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0142] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0143] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0144] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0145] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A Doppler radar static fusion smooth variable structure filtering method, characterized in that: The following steps are involved: Using the first generalized smooth variable structure filter to update the target state and obtain a posterior estimate of the target state; Using a second generalized smooth variable structure filter to update and estimate the target pseudo state, and obtain a posterior estimate of the target pseudo state; Using a static fuser to fuse the target state posterior estimate and the target pseudo-state posterior estimate to obtain a final posterior state estimate of the target; The method of using the first generalized smooth variable structure filter to update the target state and obtain a posterior estimate of the target state includes: using an unbiased measurement conversion method to convert the nonlinear position measurement of the Doppler radar in the polar coordinate system or the spherical coordinate system to the Cartesian coordinate system to obtain the linear position measurement and its conversion deviation and conversion measurement error covariance; Obtaining a posteriori estimate of the target state using the first generalized smooth variable structure filter and the linear position measurement and its conversion deviation and conversion measurement error covariance; The target pseudo-state is updated and estimated by using the second generalized smoothly variable structure filter to obtain a posterior estimate of the target pseudo-state, including: constructing a distance-radial velocity product pseudo-state vector and a pseudo-measurement variable; using an unbiased measurement conversion method to calculate the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable; and using the second generalized smoothly variable structure filter and the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable to obtain a posterior estimate of the target pseudo-state.

2. The method according to claim 1, characterized in that: Obtaining the target state posterior estimate using the first generalized smooth variable structure filter and the linear position measurement and its conversion deviation and conversion measurement error covariance, including: Calculate the target state prior estimate, radar measurement prior estimate and prior state estimate error covariance matrix at time k+1; Calculating position measurement innovation, and calculating a first mixed error term and a first innovation gain term according to the position measurement innovation; The target state posterior estimate and its covariance matrix based on the target position measurement are calculated.

3. The method according to claim 1, characterized in that Obtaining the target pseudo-state posterior estimate using the second generalized smooth variable structure filter and the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable, including: Calculate the target pseudo-state prior estimate, radar pseudo-measurement prior estimate and pseudo-state prior estimate covariance matrix at time k+1; Calculating pseudo measurement innovation, and calculating a second mixed error term and a second innovation gain term according to the pseudo measurement innovation; Calculate the target pseudo-state posterior estimate and its covariance matrix.

4. The method according to any one of claims 1 to 3, characterized in that: Using the target state posterior estimate and its covariance matrix output by the first generalized smooth variable structure filter and the target pseudo-state posterior estimate and its covariance matrix output by the second generalized smooth variable structure filter, a static fuser based on the linear minimum mean square error criterion is used to calculate the target's final posterior state estimate and its covariance.

5. A Doppler radar static fusion smooth variable structure filtering device, characterized in that: include: A first estimation module is used to update and estimate the target state using a first generalized smooth variable structure filter to obtain a posterior estimate of the target state; A second estimation module is used to update and estimate the target pseudo state using a second generalized smooth variable structure filter to obtain a posterior estimation of the target pseudo state; as well as A fusion module, used for fusing the target state posterior estimate and the target pseudo-state posterior estimate using a static fuser to obtain a final posterior state estimate of the target; The first estimation module is further used to convert the nonlinear position measurement of the Doppler radar in the polar coordinate system or the spherical coordinate system into the Cartesian coordinate system by using an unbiased measurement conversion method to obtain the linear position measurement and its conversion deviation and conversion measurement error covariance; Obtaining a posteriori estimate of the target state using the first generalized smooth variable structure filter and the linear position measurement and its conversion deviation and conversion measurement error covariance; The second estimation module is further used to construct a range-radial velocity product pseudo-state vector and a pseudo-measurement variable; use an unbiased measurement conversion method to calculate the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable; and use the second generalized smooth variable structure filter and the conversion deviation, measurement error variance and covariance of the pseudo-measurement variable to obtain a posterior estimate of the target pseudo-state.

6. An electronic device, 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 program to implement the Doppler radar static fusion smooth variable structure filtering method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the Doppler radar static fusion smooth variable structure filtering method according to any one of claims 1 to 4.

Images