A self-vehicle speed estimation method and system, computer readable storage medium

Abstract

This invention relates to a method and system for estimating vehicle speed, and a computer-readable storage medium. The method includes: estimating a first vehicle speed using a pre-set Kalman filter model with wheel speed signals and acceleration signals as observation inputs; clustering the speed signals of multiple radar targets relative to the vehicle using a pre-set clustering algorithm; when clustering is successful, obtaining a second vehicle speed based on the clustering results; when clustering is successful, identifying the current driving condition, determining weighting factors Z1 and Z2 based on the current driving condition, and performing a fusion calculation based on the first vehicle speed, weighting factor Z1, second vehicle speed, and weighting factor Z2, outputting the fusion calculation result as the vehicle speed estimation result; when clustering fails, outputting the first vehicle speed as the vehicle speed estimation result. This invention ensures good estimation results under extreme nonlinear conditions, especially under rapid acceleration / deceleration conditions and low-friction road surfaces.

Description

Technical Field

[0001] This invention relates to the field of vehicle speed estimation technology, specifically to a method and system for estimating vehicle speed, and a computer-readable storage medium. Background Technology

[0002] The main algorithms for estimating vehicle speed include direct integration, Kalman filtering, neural networks, and nonlinear state observer methods. Direct integration uses integration to obtain estimated values ​​of vehicle driving state parameters. This method has low requirements for the vehicle model, includes fewer vehicle parameters, and has some robustness, but it is not suitable for long-term use because it integrates noise signals along with sensor measurement signals. Especially under poor road conditions, the signal-to-noise ratio of the measurement signal becomes very low over time, making it impossible to obtain reliable and accurate state estimates. Kalman filtering is a prediction-correction algorithm with numerical solutions. The prediction equation is combined with the vehicle dynamics model to calculate the state parameters and error covariance at the next moment, and the correction equation is combined with the newly measured output variables of the vehicle system to finally obtain the state estimate at the current moment. This method is suitable for state estimation of weakly nonlinear systems, but its estimation performance is extremely unstable for some highly nonlinear extreme conditions, limiting its application in real vehicles. Neural networks show good experimental results in the nonlinear region of vehicle operation, but this method is highly dependent on experimental data, has slow model parameter convergence speed, poor robustness to operating conditions, and low practical application value in engineering. The nonlinear state observer method for estimating longitudinal vehicle speed requires constructing a nonlinear vehicle model and a dynamic tire model, and completing nonlinear iterative calculations, which involves a large amount of computation and makes it difficult to guarantee real-time performance.

[0003] In summary, each single vehicle speed estimation algorithm has its own shortcomings and cannot effectively solve the technical problem of vehicle speed estimation, so further improvements are urgently needed. Summary of the Invention

[0004] The purpose of this invention is to propose a method and system for estimating vehicle speed, as well as a computer-readable storage medium, which can still ensure good estimation results under extreme nonlinear conditions, especially under conditions where wheel speed measurement is inaccurate, such as rapid acceleration and deceleration. The calculation workload is small, and the real-time requirements are met.

[0005] To achieve the above objectives, the first embodiment of the present invention proposes a method for estimating vehicle speed, the method comprising:

[0006] Acquire the wheel speed signal v and acceleration signal a;

[0007] Using the wheel speed signal v and the acceleration signal a as observation inputs, a pre-set Kalman filter model is used to estimate the vehicle speed to obtain the first vehicle speed V1.

[0008] Acquire the speed signals of multiple radar targets in the surrounding environment relative to the vehicle;

[0009] The speed signals of the multiple radar targets relative to the vehicle are input into a pre-set clustering model to estimate the vehicle speed. When the clustering is successful, the second vehicle speed V2 is obtained based on the clustering results.

[0010] When clustering is successful, the current driving condition is identified, and the weighting factor Z1 of the first vehicle speed V1 and the weighting factor Z2 of the second vehicle speed V2 are determined based on the current driving condition. The first vehicle speed V1, the weighting factor Z1, the second vehicle speed V2 and the weighting factor Z2 are fused and calculated, and the fusion calculation result is output as the vehicle speed estimation result.

[0011] When clustering fails, the first vehicle speed V1 is output as the vehicle speed estimation result.

[0012] Preferably, identifying the current driving condition includes:

[0013] The current accelerator pedal opening signal and the current brake pedal opening signal are acquired, and the current accelerator pedal opening signal and the current brake pedal opening signal are compared with a number of preset opening value ranges. The current driving condition is determined based on the comparison results.

[0014] Preferably, acquiring the wheel speed signal v includes:

[0015] The wheel speed signals of the two front wheels and the two rear wheels of the vehicle are acquired by the wheel speed sensor. The four wheel speed signals are filtered and the average value is taken as the wheel speed signal v.

[0016] Preferably, the step of estimating the vehicle speed V1 using a pre-set Kalman filter model with the wheel speed signal v and the acceleration signal a as observation inputs includes:

[0017] The current state vector X1 is obtained based on the wheel speed signal v and the acceleration signal a, where X1 = [v, a]. T ;

[0018] Calculate the Kalman gain matrix K based on the current state vector X, where K = P1 * H T *(H*P1*H T ) -1H is the preset state measurement matrix, and P1 is the target covariance matrix. When the vehicle speed is estimated for the first time using the Kalman filter model after the vehicle is powered on again, P1 here is a preset fixed value. When the vehicle speed is estimated for the first time using the Kalman filter model in subsequent iterations, P1 here is the value updated in the previous vehicle speed estimation process.

[0019] The current state vector X1 and the target covariance matrix P1 are corrected based on the Kalman gain matrix K, where P2 = (IK*H)*P1, X2 = X1 + K*(ZH*X1), and Z = [v, 0]. T ;

[0020] The Kalman-filtered state vector X3 is obtained based on the corrected current state vector X2, and the first vehicle speed V1 is obtained based on the state vector X3; where X3 = F * X2, and F is a pre-set state transition matrix;

[0021] The target covariance matrix P1 is updated based on the corrected target covariance matrix P2. The updated P1 is then used for the next vehicle speed estimation in the Kalman filter model; where P1 = F * P2 * F T +Q, where Q is a preset measurement noise value.

[0022] Preferably, the step of inputting the speed signals of the multiple radar targets relative to the vehicle into a pre-set clustering model to perform clustering estimation of the vehicle speed includes:

[0023] The speed signals of the multiple radar targets relative to the vehicle are clustered based on a pre-set clustering algorithm. In each clustering iteration, when the distance between any two clusters is less than a pre-set threshold d1, the number of clusters K is increased and the next clustering iteration begins. When the clustering center of any speed signal and its corresponding cluster is greater than a pre-set threshold d2, the number of clusters K is decreased and the next clustering iteration begins. When the clusters meet the pre-set clustering termination condition, the clustering iteration ends.

[0024] After the clustering iteration ends, select the cluster containing the most speed signals and compare the number of speed signals contained in the cluster with a preset threshold d3. If the number of speed signals contained in the cluster is greater than the threshold d3, the clustering is considered successful and the second vehicle speed V2 is obtained based on the speed signals contained in the cluster. Otherwise, the clustering is considered to have failed.

[0025] Preferably, acquiring the speed signals of multiple radar targets in the vehicle's surrounding environment relative to the vehicle includes:

[0026] The longitudinal and lateral velocity signals of multiple radar targets in the vehicle's surrounding environment detected by the vehicle-mounted millimeter-wave radar relative to the vehicle are acquired, and abnormal signals are removed from the longitudinal and lateral velocity signals of the multiple radar targets to obtain velocity signals for clustering.

[0027] Preferably, obtaining the second vehicle speed V2 based on the speed signal contained in the cluster includes:

[0028] Select the longitudinal velocity signal and the lateral velocity signal that are closest to the cluster center of the cluster, and calculate the second vehicle speed V2 based on the longitudinal velocity signal and the lateral velocity signal.

[0029] As part of the same inventive concept, a second embodiment of the present invention proposes a vehicle speed estimation system, comprising:

[0030] The first signal acquisition unit is used to acquire the wheel speed signal v and the acceleration signal a;

[0031] The first vehicle speed estimation unit is used to estimate the vehicle speed of the vehicle by using a pre-set Kalman filter model with wheel speed signal v and acceleration signal a as observation inputs to obtain the first vehicle speed V1.

[0032] The second signal acquisition unit is used to acquire the speed signals of multiple radar targets in the vehicle's surrounding environment relative to the vehicle.

[0033] The second vehicle speed estimation unit is used to input the speed signals of the multiple radar targets relative to the vehicle into a pre-set clustering model to perform clustering estimation of the vehicle speed. When the clustering is successful, the second vehicle speed V2 is obtained based on the clustering results.

[0034] The vehicle speed output unit is used to identify the current driving condition when the second vehicle speed estimation unit successfully clusters the data, determine the weighting factor Z1 of the first vehicle speed V1 and the weighting factor Z2 of the second vehicle speed V2 based on the current driving condition, perform a fusion calculation based on the first vehicle speed V1, the weighting factor Z1, the second vehicle speed V2 and the weighting factor Z2, and output the fusion calculation result as the vehicle speed estimation result; when the second vehicle speed estimation unit fails to cluster the data, the first vehicle speed V1 is output as the vehicle speed estimation result.

[0035] Preferably, the vehicle speed output unit is specifically used for:

[0036] The current accelerator pedal opening signal and the current brake pedal opening signal are acquired, and the current accelerator pedal opening signal and the current brake pedal opening signal are compared with a number of preset opening value ranges. The current driving condition is determined based on the comparison results.

[0037] Preferably, the second signal acquisition unit is specifically used for:

[0038] The longitudinal and lateral velocity signals of multiple radar targets in the vehicle's surrounding environment detected by the vehicle-mounted millimeter-wave radar relative to the vehicle are acquired, and abnormal signals are removed from the longitudinal and lateral velocity signals of the multiple radar targets to obtain velocity signals for clustering.

[0039] As part of the same inventive concept, a third embodiment of the present invention proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described vehicle speed estimation method.

[0040] The above-mentioned method and system for estimating vehicle speed, as well as the computer-readable storage medium, have at least the following beneficial effects:

[0041] A fusion model combining Kalman filtering and clustering is proposed. Kalman filtering and clustering are used to estimate the first vehicle speed V1 and the second vehicle speed V2, respectively. Based on the current driving conditions, weighting factors Z1 and Z2 for the first vehicle speed V1 and second vehicle speed V2 are determined. A fusion calculation is performed using the first vehicle speed V1, weighting factor Z1, second vehicle speed V2, and weighting factor Z2, and the fusion calculation result is output as the vehicle speed estimation result. Compared to the single Kalman filtering method, this model does not require linear driving conditions and exhibits better robustness, especially under conditions where wheel speed measurement is inaccurate, such as rapid acceleration and deceleration. The estimation effect is as follows: When clustering fails, the first vehicle speed V1 is output as the vehicle speed estimation result, so that even when clustering fails, the vehicle speed estimation value can still be output based on Kalman filtering, ensuring the normal use of backend positioning, navigation and other functions; moreover, compared with the nonlinear state observer method, the computational load is smaller and meets the real-time requirements; compared with the direct integration method, it will not fail to obtain a reliable and accurate state estimation value due to the low signal-to-noise ratio of the measurement signal over time, and the estimation accuracy is high, making it suitable for long-term use; in summary, this invention helps to ensure the reliability of advanced driver assistance functions such as high-precision positioning, high-precision navigation, and advanced cruise control.

[0042] Other features and advantages of the present invention will be set forth in the following description. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart of a vehicle speed estimation method according to an embodiment of the present invention.

[0045] Figure 2 This is a flowchart of the cluster estimation of vehicle speed in one embodiment of the present invention.

[0046] Figure 3 This is a schematic diagram of a vehicle speed estimation system according to an embodiment of the present invention. Detailed Implementation

[0047] The various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. Furthermore, numerous specific details are set forth in the following detailed embodiments to better illustrate the invention. Those skilled in the art will understand that the invention can be practiced without certain specific details. In some instances, means well known to those skilled in the art have not been described in detail in order to highlight the spirit of the invention.

[0048] One embodiment of the present invention proposes a method for estimating vehicle speed, see reference. Figure 1 The method in this embodiment includes the following steps:

[0049] Step S11: Obtain the wheel speed signal v and acceleration signal a;

[0050] Specifically, the wheel speed signal v can be obtained through the vehicle's wheel speed sensor, and the acceleration signal a can be obtained through the vehicle's acceleration sensor;

[0051] Step S12: Using the wheel speed signal v and the acceleration signal a as observation inputs, a pre-set Kalman filter model is used to estimate the vehicle speed to obtain the first vehicle speed V1;

[0052] Step S21: Acquire the speed signals of multiple radar targets in the surrounding environment of the vehicle relative to the vehicle.

[0053] Specifically, all vehicles are equipped with radar, which can be used to detect the speed signals of obstacles relative to the vehicle in the driving environment. Obstacles are radar targets, such as vehicles, pedestrians, traffic light equipment, trees on both sides of the road, etc. In this embodiment, multiple radars can be installed at the front and rear. One radar can generally output 32 radar targets, which depends on the detection capability of the radar. However, the number of radar targets does not affect the application of the method in this embodiment. All of them can be subjected to cluster analysis in the following steps.

[0054] Step S22: Input the speed signals of the multiple radar targets relative to the vehicle into a pre-set clustering model to estimate the vehicle speed by clustering. When the clustering is successful, obtain the second vehicle speed V2 based on the clustering results.

[0055] Specifically, since stationary obstacles are not moving, the speed signal of the stationary obstacle relative to the vehicle can be regarded as the speed signal of the vehicle during the vehicle's movement. However, determining the vehicle's speed signal solely based on the speed signal of a single stationary obstacle relative to the vehicle may easily lead to errors. Therefore, in this embodiment, the speed signals of multiple radar targets relative to the vehicle are used for cluster analysis to obtain a more accurate second vehicle speed V2.

[0056] After clustering, a cluster that best represents the vehicle's speed signal can be obtained. The corresponding second vehicle speed V2 can be obtained based on this cluster. When clustering fails, a clustering failure signal "fail" is output. It should be noted that in the actual radar detection process, because the speed signals of all stationary obstacles relative to the vehicle are very close to the actual speed of the vehicle, through clustering, the speed signals of all stationary obstacles relative to the vehicle will be clustered into the same cluster. The number of speed signals in this cluster must be the largest among all clusters. The speed signal represented by this cluster can be regarded as the speed signal of the vehicle. Therefore, the second vehicle speed V2 can be obtained based on the clustering results.

[0057] Step S31: When clustering is successful, identify the current driving condition, determine the weighting factor Z1 of the first vehicle speed V1 and the weighting factor Z2 of the second vehicle speed V2 based on the current driving condition, perform a fusion calculation based on the first vehicle speed V1, weighting factor Z1, second vehicle speed V2 and weighting factor Z2, and output the fusion calculation result as the vehicle speed estimation result.

[0058] Specifically, in this embodiment, Z1+Z2=1 is defined. The weighting factors Z1 and Z2 represent the influence weights V of the first vehicle speed V1 and the second vehicle speed V2 on the final vehicle speed, respectively. For example, the fusion calculation can be expressed as: V=Z1V1+2V2.

[0059] Step S32: When clustering fails, output the first vehicle speed V1 as the vehicle speed estimation result.

[0060] Specifically, when clustering fails, the second vehicle speed V2 cannot be obtained, but it is necessary to ensure the normal use of vehicle functions such as backend positioning and navigation. In this embodiment, the first vehicle speed V1 output in step S21 is used as the vehicle speed estimation result because the Kalman filter method does not have a situation where the first vehicle speed V1 cannot be obtained.

[0061] In some more specific embodiments, in step S31, identifying the current driving condition includes:

[0062] The current accelerator pedal opening signal and the current brake pedal opening signal are acquired, and the current accelerator pedal opening signal and the current brake pedal opening signal are compared with a number of preset opening value ranges. The current driving condition is determined based on the comparison results.

[0063] Specifically, driving conditions can be divided into multiple types, and different driving conditions correspond to different weighting factors Z1 and Z2. These can be pre-calibrated and directly called according to the current driving conditions during the execution of the steps. The weighting factors Z1 and Z2 mainly consider the accuracy of the first vehicle speed V1 and the second vehicle speed V2 in different driving conditions. The higher the accuracy, the larger the corresponding weighting factor, and always satisfy Z1+Z2=1.

[0064] The weighting factors Z1 and Z2 can be calibrated as follows: Take n sets of arbitrary vehicle speeds V1 and V2 under any working conditions, and denote them as x1 = [x 11 ,x 12 …x 1n ] and x2 = [x 21 ,x 22 …x 2n We set weighting factors Z1 and Z2, with Z1 + Z2 = 1, representing the weights of the Kalman filter-based vehicle speed estimate and the cluster-based vehicle speed estimate in the multi-model fusion vehicle speed estimation in this embodiment, respectively. Let the mean square errors of the two vehicle speed estimation models be... and The total mean square error is Based on the theory of finding extrema of multivariate functions, and according to the principle of minimizing the total mean square error of vehicle speed estimation, the corresponding weights are obtained as follows:

[0065]

[0066] Combining the above formulas and the constraint Z1+Z2=1, we can obtain the weighting factors Z1 and Z2 for any working condition.

[0067] In this embodiment, the current driving condition is determined based on the current accelerator pedal opening signal and the current brake pedal opening signal. The classification of driving conditions is shown in Table 1 below:

[0068] Table 1 - Driving Condition Classification

[0069]

[0070] Table 1 shows the scenarios where driving conditions are divided into three types. Of course, this invention is not limited to three driving conditions and can include more. Here, a1, a2, and a3 are pre-set opening thresholds, thus forming three opening value intervals [0, a1], (a1, a2], and (a2, a3]. These three pre-set opening value intervals correspond to different driving conditions, namely, gradual acceleration / deceleration, moderate acceleration / deceleration, and rapid acceleration / deceleration. Different driving conditions correspond to different weighting factors Z1 and Z2. With a weight factor of Z2, the current driving condition can be determined by comparing the current accelerator pedal opening signal and the current brake pedal opening signal with multiple opening value intervals. For the accelerator pedal opening signal and the brake pedal opening signal, when both are 0, corresponding to [0, a1], the current driving condition is a gradual acceleration / deceleration driving condition. When one of them is not 0, the other must be 0. The driving condition corresponding to the opening value interval into which the non-zero pedal opening signal falls is determined as the current driving condition.

[0071] It should be noted that the driving conditions can be further divided by combining the accelerator / brake pedal opening and the road surface adhesion coefficient. More accurate vehicle speed estimates based on fusion can be obtained through more detailed weighting coefficients.

[0072] In some more specific embodiments, in step S31, acquiring the wheel speed signal v specifically includes:

[0073] The wheel speed signals of the two front wheels and two rear wheels of the vehicle are acquired from the wheel speed sensors, resulting in four wheel speed signals V. s1 V s2 V s3 and V s4 The wheel speed signals of the four wheels are filtered respectively, and the filtered V s1 V s2 V s3 and V s4 The average value is taken as the wheel speed signal v.

[0074] In some more specific embodiments, step S12 includes:

[0075] Step S121: Obtain the current state vector X1 based on the wheel speed signal v and the acceleration signal a, where X1 = [v, a] T ;

[0076] Step S122: Calculate the Kalman gain matrix K based on the current state vector X, where K = P1 * H T *(H*P1*H T ) -1 H is the preset state measurement matrix, and P1 is the target covariance matrix. When the vehicle speed is estimated for the first time using the Kalman filter model after the vehicle is powered on again, P1 here is a preset fixed value, which is determined based on engineering experience. When the vehicle speed is estimated in subsequent Kalman filter models, P1 here is the value updated in the previous vehicle speed estimation process.

[0077] Specifically, in this embodiment, the state measurement matrix is ​​H = [1,0], and T represents the transpose of the matrix; the Kalman gain matrix K is used to weigh the weights between the measured value and the estimated value;

[0078] Step S123: Correct the current state vector X1 and the target covariance matrix P1 according to the Kalman gain matrix K, where P2 = (IK*H)*P1, X2 = X1 + K*(ZH*X1), Z = [v, 0] T ;

[0079] Step S124: Obtain the Kalman-filtered state vector X3 based on the corrected current state vector X2, and obtain the first vehicle speed V1 based on the state vector X3; where X3 = F * X2, and F is a pre-set state transition matrix;

[0080] Specifically, in this embodiment, the state transition matrix Where Δt is the time interval between the two Kalman filter model speed estimations;

[0081] Step S125: Update the target covariance matrix P1 based on the corrected target covariance matrix P2. The updated P1 is used for the next vehicle speed estimation in the Kalman filter model; where P1 = F * P2 * F T +Q, where Q is a preset measurement noise value.

[0082] In some more specific embodiments, step S21 includes:

[0083] The longitudinal and lateral velocity signals of multiple radar targets in the vehicle's surrounding environment detected by the vehicle-mounted millimeter-wave radar relative to the vehicle are acquired, and abnormal signals are removed from the longitudinal and lateral velocity signals of the multiple radar targets to obtain velocity signals for clustering.

[0084] Specifically, the abnormal signal removal can be based on the 3σ principle of Gaussian distribution, but is not limited to this method;

[0085] Step S22 includes:

[0086] Step S221: Cluster the speed signals of the multiple radar targets relative to the vehicle based on a pre-set clustering algorithm; wherein, in each clustering iteration, when the distance between any two clusters is less than a pre-set threshold d1, the number of clusters K is increased and the next clustering iteration begins; when the clustering of any speed signal and the cluster center of its corresponding cluster is greater than a pre-set threshold d2, the number of clusters K is decreased and the next clustering iteration begins; when the clusters meet the pre-set clustering termination condition, the clustering iteration ends.

[0087] Specifically, this embodiment preferably, but is not limited to, using the KMeans++ algorithm for cluster analysis. Regarding the selection of the number of clusters, considering the small time interval between consecutive frames of millimeter-wave radar, the non-abrupt nature of target information on roads, and the non-abrupt nature of speed information in the millimeter-wave output channel, the cluster distribution will not change abruptly. Therefore, an adaptive adjustment method for the number of clusters K is used to select the most suitable value for K. During cluster initialization, the number of clusters K is assigned an initial value. After each clustering cycle, K is adjusted based on the distance between clusters and the distance between clusters and sample points, and the adjusted K is output to the next clustering cycle. To accelerate the convergence speed of the clustering algorithm and improve the quality of the clustering results, this embodiment initializes the clusters based on the KMeans++ algorithm, ensuring that the distance between the initially initialized clusters is as large as possible. This is an iterative clustering analysis algorithm. After determining K signals as cluster centers, it calculates the distance between each signal and each cluster center, and assigns each signal to the nearest cluster center. The cluster centers and the signals assigned to them represent a cluster. For each assigned sample, the cluster centers are recalculated based on the existing signals within the cluster. This process is repeated until the pre-set clustering termination criteria are met.

[0088] Furthermore, the clustering termination condition can be: the error between the results of two consecutive clustering sessions is less than a certain range; specifically, it can be expressed as: no (or minimum number) objects are reassigned to different clusters; or, no (or minimum number) cluster centers change again; or, the sum of squared errors is locally minimized.

[0089] Step S222: After the clustering iteration ends, select the cluster containing the most speed signals and compare the number of speed signals contained in the cluster with the preset threshold d3. If the number of speed signals contained in the cluster is greater than the threshold d3, the clustering is considered successful, and the second vehicle speed V2 is obtained based on the speed signals contained in the cluster. Otherwise, the clustering is considered to have failed.

[0090] The process for clustering and estimating vehicle speed can be found in [reference needed]. Figure 2 .

[0091] In some more specific embodiments, in step S222, obtaining the second vehicle speed V2 based on the speed signals contained in the cluster includes:

[0092] Select the longitudinal velocity signal and the lateral velocity signal that are closest to the cluster center of the cluster, and calculate the second vehicle speed V2 based on the longitudinal velocity signal and the lateral velocity signal.

[0093] Another embodiment of the present invention proposes a vehicle speed estimation system. The functional units of this system can be used to implement the corresponding steps of the method in the above embodiments. See reference. Figure 2 The system in this embodiment includes the following functional units:

[0094] The first signal acquisition unit 1 is used to acquire the wheel speed signal v and the acceleration signal a;

[0095] The first vehicle speed estimation unit 2 is used to estimate the vehicle speed of the vehicle by using a pre-set Kalman filter model with wheel speed signal v and acceleration signal a as observation inputs to obtain the first vehicle speed V1.

[0096] The second signal acquisition unit 3 is used to acquire the speed signals of multiple radar targets in the surrounding environment of the vehicle relative to the vehicle.

[0097] The second vehicle speed estimation unit 4 is used to input the speed signals of the multiple radar targets relative to the vehicle into a pre-set clustering model to perform clustering estimation of the vehicle speed. When the clustering is successful, the second vehicle speed V2 is obtained according to the clustering result. When the clustering fails, a clustering failure signal fail is output.

[0098] Additionally, the vehicle speed output unit 5 is used to identify the current driving condition when the second vehicle speed estimation unit successfully clusters, determine the weighting factor Z1 of the first vehicle speed V1 and the weighting factor Z2 of the second vehicle speed V2 based on the current driving condition, perform a fusion calculation based on the first vehicle speed V1, the weighting factor Z1, the second vehicle speed V2 and the weighting factor Z2, and output the fusion calculation result as the vehicle speed estimation result; when the second vehicle speed estimation unit fails to cluster, the first vehicle speed V1 is output as the vehicle speed estimation result.

[0099] In some more specific embodiments, the vehicle speed output unit 5 is specifically used for:

[0100] The current accelerator pedal opening signal and the current brake pedal opening signal are acquired, and the current accelerator pedal opening signal and the current brake pedal opening signal are compared with a number of preset opening value ranges. The current driving condition is determined based on the comparison results.

[0101] In some more specific embodiments, the second signal acquisition unit 3 is specifically used for:

[0102] The longitudinal and lateral velocity signals of multiple radar targets in the vehicle's surrounding environment detected by the vehicle-mounted millimeter-wave radar relative to the vehicle are acquired, and abnormal signals are removed from the longitudinal and lateral velocity signals of the multiple radar targets to obtain velocity signals for clustering.

[0103] In some more specific embodiments, the first signal acquisition unit 1 is specifically used to acquire the wheel speed signals of the two front wheels and the two rear wheels of the vehicle collected by the wheel speed sensor, and the four wheel speed signals are filtered and the average value is taken as the wheel speed signal v.

[0104] In some more specific embodiments, the first vehicle speed estimation unit 2 is specifically used for:

[0105] The current state vector X1 is obtained based on the wheel speed signal v and the acceleration signal a, where X1 = [v, a]. T ;

[0106] Calculate the Kalman gain matrix K based on the current state vector X, where K = P1 * H T *(H*P1*H T ) -1H is the preset state measurement matrix, and P1 is the target covariance matrix. When the vehicle speed is estimated for the first time using the Kalman filter model after the vehicle is powered on again, P1 here is a preset fixed value. When the vehicle speed is estimated for the first time using the Kalman filter model in subsequent iterations, P1 here is the value updated in the previous vehicle speed estimation process.

[0107] The current state vector X1 and the target covariance matrix P1 are corrected based on the Kalman gain matrix K, where P2 = (IK*H)*P1, X2 = X1 + K*(ZH*X1), and Z = [v, 0]. T ;

[0108] The Kalman-filtered state vector X3 is obtained based on the corrected current state vector X2, and the first vehicle speed V1 is obtained based on the state vector X3; where X3 = F * X2, and F is a pre-set state transition matrix;

[0109] The target covariance matrix P1 is updated based on the corrected target covariance matrix P2. The updated P1 is then used for the next vehicle speed estimation in the Kalman filter model; where P1 = F * P2 * F T +Q, where Q is a preset measurement noise value.

[0110] In some more specific embodiments, the second vehicle speed estimation unit 4 is specifically used for:

[0111] The speed signals of the multiple radar targets relative to the vehicle are clustered based on a pre-set clustering algorithm. In each clustering iteration, when the distance between any two clusters is less than a pre-set threshold d1, the number of clusters K is increased and the next clustering iteration begins. When the clustering center of any speed signal and its corresponding cluster is greater than a pre-set threshold d2, the number of clusters K is decreased and the next clustering iteration begins. When the clusters meet the pre-set clustering termination condition, the clustering iteration ends.

[0112] After the clustering iteration ends, the cluster containing the most speed signals is selected, and the number of speed signals contained in this cluster is compared with a pre-set threshold d3. If the number of speed signals contained in this cluster is greater than the threshold d3, the clustering is considered successful, and the second vehicle speed V2 is obtained based on the speed signals contained in this cluster. That is, the longitudinal speed signal and the lateral speed signal that are closest to the cluster center of this cluster are selected, and the second vehicle speed V2 is calculated based on the longitudinal speed signal and the lateral speed signal. Otherwise, the clustering is considered to have failed.

[0113] The systems described in the embodiments above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the system solution in the embodiments, depending on actual needs.

[0114] It should be noted that the system in the above embodiments corresponds to the method in the above embodiments. Therefore, the parts of the system in the above embodiments that are not described in detail can be obtained by referring to the content of the method in the above embodiments. That is, the specific steps recorded in the method in the above embodiments can be understood as the functions that the system in the above embodiments can achieve, and will not be described again here.

[0115] Furthermore, if the system described in the above embodiments is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0116] Another embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the vehicle speed estimation method described in the above embodiments.

[0117] Specifically, the computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM, Ry), random access memory (RAM, Ry), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0118] As described in the above embodiments, the embodiments of the present invention propose a fusion model combining a Kalman filter model and a clustering model. The first vehicle speed V1 and the second vehicle speed V2 are estimated using Kalman filtering and clustering methods respectively. Based on the current driving conditions, a weighting factor Z1 for the first vehicle speed V1 and a weighting factor Z2 for the second vehicle speed V2 are determined. A fusion calculation is then performed based on the first vehicle speed V1, weighting factor Z1, second vehicle speed V2, and weighting factor Z2, and the fusion calculation result is output as the vehicle speed estimation result. Compared to a single Kalman filter method, it does not require the vehicle driving conditions to be linear, and has better robustness to driving conditions, especially under conditions of rapid acceleration and deceleration where wheel speed measurement is not... Even under precise operating conditions, it can still guarantee good estimation results; when clustering fails, the first vehicle speed V1 is output as the vehicle speed estimation result, so that even when clustering fails, the vehicle speed estimation value can still be output based on Kalman filtering, ensuring the normal use of backend functions; moreover, compared with the nonlinear state observer method, the computational load is smaller and meets the real-time requirements; compared with the direct integration method, it will not fail to obtain reliable and accurate state estimation values ​​due to the low signal-to-noise ratio of the measurement signal over time, and the estimation accuracy is high, making it suitable for long-term use; in summary, the various embodiments of the present invention help to ensure the reliability of advanced driver assistance functions such as high-precision positioning, high-precision navigation, and advanced cruise control.

[0119] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for estimating the speed of a vehicle, characterized in that, The method includes: Obtain wheel speed signal and acceleration signal ; The wheel speed signal and the acceleration signal The vehicle speed is estimated by using a pre-set Kalman filter model as the observation input to obtain the first vehicle speed. ; Acquire the speed signals of multiple radar targets in the surrounding environment relative to the vehicle; The speed signals of the multiple radar targets relative to the vehicle are input into a pre-set clustering model to estimate the vehicle speed. When clustering is successful, the second vehicle speed is obtained based on the clustering results. ; When clustering is successful, the current driving condition is identified, and the first vehicle speed is determined based on the current driving condition. The weighting factor Z1 and the second vehicle speed The weighting factor Z2 is based on the first vehicle speed. Weighting factor Z1, second vehicle speed The vehicle speed is estimated by combining the weighting factor Z2 with the weighting factor Z2. When clustering fails, the speed of the first vehicle will be... The output is the vehicle speed estimation result.

2. The method according to claim 1, characterized in that, The identification of the current driving condition includes: The current accelerator pedal opening signal and the current brake pedal opening signal are acquired, and the current accelerator pedal opening signal and the current brake pedal opening signal are compared with a number of preset opening value ranges. The current driving condition is determined based on the comparison results.

3. The method according to claim 1, characterized in that, The acquisition of wheel speed signal ,include: The wheel speed signals of the two front wheels and the two rear wheels of the vehicle are acquired by the wheel speed sensor. The four wheel speed signals are filtered and the average value is taken as the wheel speed signal.

4. The method according to claim 1, characterized in that, The wheel speed signal and the acceleration signal The vehicle speed is estimated by using a pre-set Kalman filter model as the observation input to obtain the first vehicle speed. ,include: According to the wheel speed signal and the acceleration signal Obtain the current state vector , ; According to the current state vector Calculate the Kalman gain matrix ,in, , The preset state measurement matrix, Given the target covariance matrix, when the vehicle speed is estimated for the first time using the Kalman filter model after the vehicle is powered back on, this is where... This is a pre-set fixed value; when estimating the vehicle speed using the Kalman filter model, this value will be used here. The value updated from the previous vehicle speed estimation process; According to the Kalman gain matrix For the current state vector and target covariance matrix The following corrections were made: , , ; Based on the corrected current state vector Obtain the state vector after Kalman filtering According to the state vector Obtain the first vehicle speed ;in, , A pre-set state transition matrix; Based on the corrected target covariance matrix For the target covariance matrix Updated, after the update Used for vehicle speed estimation in the next Kalman filter model; where, Q is a preset measurement noise value.

5. The method according to claim 1, characterized in that, The step of inputting the speed signals of the multiple radar targets relative to the vehicle into a pre-set clustering model to perform clustering estimation of the vehicle speed includes: The speed signals of the multiple radar targets relative to the vehicle are clustered based on a pre-set clustering algorithm. In each clustering iteration, when the distance between any two clusters is less than a pre-set threshold d1, the number of clusters K is increased and the next clustering iteration begins. When the clustering center of any speed signal and its corresponding cluster is greater than a pre-set threshold d2, the number of clusters K is decreased and the next clustering iteration begins. When the clusters meet the pre-set clustering termination condition, the clustering iteration ends. After the clustering iteration ends, the cluster containing the most speed signals is selected, and the number of speed signals contained in this cluster is compared with a pre-set threshold d3. If the number of speed signals contained in this cluster is greater than the threshold d3, the clustering is considered successful, and the second vehicle speed is obtained based on the speed signals contained in this cluster. Otherwise, the clustering is deemed a failure.

6. The method according to claim 5, characterized in that, The acquisition of speed signals of multiple radar targets in the vehicle's surrounding environment relative to the vehicle includes: The longitudinal and lateral velocity signals of multiple radar targets in the vehicle's surrounding environment detected by the vehicle-mounted millimeter-wave radar relative to the vehicle are acquired, and abnormal signals are removed from the longitudinal and lateral velocity signals of the multiple radar targets to obtain velocity signals for clustering.

7. The method according to claim 6, characterized in that, The second vehicle speed is obtained based on the speed signal contained in the cluster. ,include: Select the longitudinal velocity signal and the lateral velocity signal that are closest to the cluster center of the given cluster, and calculate the second vehicle speed based on the longitudinal velocity signal and the lateral velocity signal. .

8. A vehicle speed estimation system, characterized in that, include: The first signal acquisition unit is used to acquire wheel speed signals. and acceleration signal ; The first vehicle speed estimation unit is used to estimate the wheel speed signal. and the acceleration signal The vehicle speed is estimated by using a pre-set Kalman filter model as the observation input to obtain the first vehicle speed. ; The second signal acquisition unit is used to acquire the speed signals of multiple radar targets in the vehicle's surrounding environment relative to the vehicle. The second vehicle speed estimation unit is used to input the speed signals of the multiple radar targets relative to the vehicle into a pre-set clustering model to perform clustering estimation of the vehicle speed. When clustering is successful, the second vehicle speed is obtained based on the clustering results. ; The vehicle speed output unit is used to identify the current driving condition when the second speed estimation unit successfully clusters the data, and to determine the first vehicle speed based on the current driving condition. The weighting factor Z1 and the second vehicle speed The weighting factor Z2 is based on the first vehicle speed. Weighting factor Z1, second vehicle speed The speed is fused with the weighting factor Z2, and the fusion calculation result is output as the vehicle speed estimation result; when the second vehicle speed estimation unit fails to cluster, the first vehicle speed is used instead. The output is the vehicle speed estimation result.

9. The system according to claim 8, characterized in that, The vehicle speed output unit is specifically used for: The current accelerator pedal opening signal and the current brake pedal opening signal are acquired, and the current accelerator pedal opening signal and the current brake pedal opening signal are compared with a number of preset opening value ranges. The current driving condition is determined based on the comparison results.

10. The system according to claim 8, characterized in that, The second signal acquisition unit is specifically used for: The longitudinal and lateral velocity signals of multiple radar targets in the vehicle's surrounding environment detected by the vehicle-mounted millimeter-wave radar relative to the vehicle are acquired, and abnormal signals are removed from the longitudinal and lateral velocity signals of the multiple radar targets to obtain velocity signals for clustering.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the vehicle speed estimation method according to any one of claims 1-7.

Images