CN112880699B - Vehicle cooperative positioning method based on brain selective attention mechanism - Google Patents

Abstract

The invention discloses a vehicle cooperative localization method based on a brain selective attention mechanism, which specifically aims at the influence of a geometric structure, RB position precision and a relative motion state formed between a cooperative reference node and a vehicle to be localized on cooperative localization; adopting a selective attention mechanism of human brain to information processing, taking the three influencing factors as characteristic points of information selection, and carrying out selective filtering processing on information; and then, comprehensively screening out the optimal RB through feature integration so as to further improve the reliability and precision of the cooperative positioning of the vehicle.

Description

A Cooperative Vehicle Localization Method Based on Brain Selective Attention Mechanism

technical field

The invention relates to the technical field of vehicle cooperative positioning, in particular to a vehicle cooperative positioning method based on a brain selective attention mechanism.

Background technique

Real-time precise positioning of driving vehicles is one of the necessary key technologies for the realization of Intelligent Transportation System (Intelligent Transportation System, ITS) and automatic driving technology (Automatic Driving Technology, ADT), such as: vehicle information management, traffic control, vehicle information in ITS Services, automatic obstacle avoidance, lane change and other technologies in ADT all require vehicles to have high reliability and high precision self-positioning capabilities; therefore, research on reliable vehicle positioning methods is of great significance for the realization of intelligent transportation systems;

At present, the main application of ground carrier positioning technology includes inertial navigation positioning, visual navigation positioning and Global Positioning System (GPS) technology. Due to its high cost, inertial navigation positioning technology is usually found in special ground carriers such as military and high Cost restricts its application in civilian vehicles; although visual navigation and positioning technology has been widely researched and applied, it usually only provides relative position information; and GPS is currently the most widely used due to its high cost performance and maturity. Vehicle positioning technology; GPS positioning accuracy varies from meter level to centimeter level according to different implementation technologies. For example, carrier phase difference GPS positioning accuracy can reach centimeter level; Satellite signal occlusion and multipath interference will seriously affect the accuracy and reliability of GPS positioning, even for carrier phase difference GPS, so it is difficult to meet the needs of urban ITS.

With the rapid development of wireless communication technology, Vehicular Ad hoc Network (VANET) is responsible for the interactive tasks of various traffic information and has become an important part of ITS; at the same time, the V2V, V2I and V2X communication methods in VANET are also important for vehicles. Navigation positioning and positioning enhancement have brought new solutions, that is, vehicle co-location technology, the essence of which is that vehicles use other vehicles or traffic infrastructure as reference points, and obtain information such as the position and relative position of the reference point through the wireless network. Positioning; in recent years, the strong demand for the application of artificial intelligence technology has prompted experts in many fields such as neuroscience, cognitive science, and computing science to work hard to reveal the mechanism of brain information perception and processing from different levels, and try to simulate its functions; research The personnel found that: the reason why the brain can quickly process and respond to the massive information of the surrounding environment is that the brain will consciously or unconsciously carry out targeted screening of information, which is a selective attention mechanism, and in the existing technology, and There is no application of the brain selective attention mechanism in vehicle cooperative positioning, and there are no known corresponding reports and studies.

Contents of the invention

In view of the above existing problems, the present invention aims to provide a vehicle cooperative positioning method based on the brain selective attention mechanism. This method studies the geometric structure, RB position accuracy and relative motion state formed between the collaborative reference node and the vehicle to be positioned. For the impact of collaborative positioning, it is proposed to use the human brain's selective attention mechanism for information processing, and use the above three factors as the feature points of information selection to selectively filter the information; and then comprehensively screen out the optimal RB through feature integration , in order to further improve the reliability and accuracy of vehicle cooperative positioning, which has the characteristics of high positioning accuracy and not limited by the environment.

In order to achieve the above object, the technical scheme adopted in the present invention is as follows:

A collaborative vehicle localization method based on brain selective attention mechanism, including

Step 1: In the VANET environment, select N RB combinations from VTBP neighbor vehicles or ITS infrastructure nodes, calculate the geometric precision eigenvalues of the selected N RB combinations, and sort them in ascending order;

Step 2: The CSM model that introduces the variance adjustment factor is the state equation, and the position coordinates of the RB are used as observations to correct the position coordinates of the RB, and at the same time evaluate the accuracy of the position coordinates of the RB;

Step 3: Comprehensively evaluate the relative position characteristics, position accuracy characteristics and relative motion characteristics of RB; use the evaluated optimal RB position coordinates and the relative distance from VTBP to establish a cooperative positioning equation group, and solve the VTBP position coordinates;

Step 4: Establish a dead reckoning model with the solved VTBP position coordinates as observations, and correct the VTBP position coordinates to obtain the final VTBP position coordinates to complete vehicle collaborative positioning.

Preferably, the selection of N RB combinations described in step 1 and the calculation process of geometric precision eigenvalues include:

S1.1 Set the position coordinates of VTBP as unknown quantities (x, y), select n neighbor vehicles or ITS infrastructure nodes from the surrounding environment as positioning RB _i , where i=1,2,...,n; RB _i The position coordinates are known quantities (x _i , y _i );

S1.2 The relative distance d _i between VTBP and RB _i can be obtained as:

S1.3 Calculate the geometric precision eigenvalues of the selected N RB combinations, and sort them in ascending order.

Preferably, the calculation process of the geometric precision eigenvalue described in step S1.3 includes:

(1) Introduce the measurement error e _i , then the position error of the corresponding VTBP is (e _x , e _y ), then the formula (1) after adding the error is:

Transform (2) into a linear equation to get

a_i＝(x-x_i)/d_i，b_i＝(y-y_i)/d_i，可将式(3)转化为矩阵形式：(2) order

a _i ＝(xx _i )/d _i , b _i ＝(yy _i )/d _i , the formula (3) can be transformed into matrix form:

L=HX+e (4)

In formula (4): L=[l ₁ ,l ₂ ,…l _n ] ^T , X=[e _x ,e _y ] ^T , e=[e ₁ ,e ₂ ,…e _n ] ^T ,

(3) When H is full rank, H ^T H is reversible, then the error between the estimated value and the real value of position is:

The size of the error in formula (5) is measured by covariance, and we get:

In formula (6): σ ² is the variance when the noises in e are uncorrelated;

(4) It can be seen from formula (6) that (H ^T H) ^-1 represents the magnification of the distance measurement error. Therefore, the geometric position precision factor of vehicle cooperative positioning is defined as G, then

Wherein, in formula (7), tr[] represents the inverse operation of seeking matrix;

(5) Simplify the operation of formula (7), matrix inversion is equivalent to the sum of matrix eigenvalues, and the corresponding formula (7) can be written as

In formula (8): λ ₁ and λ ₂ are the eigenvalues of the matrix H ^T H;

Finally, the geometric position precision factor of vehicle cooperative positioning can be obtained:

In formula (9), det[] means to find the determinant of the matrix.

Preferably, the process of correcting the position coordinates of the RB in real time described in step 2 includes:

S2.1 uses CSM as the RB carrier motion state equation, expressed as:

为k时刻加速度均值，W_k为均值为零、方差为Q_k的高斯分布噪声向量；所述In formula (10), X _k+1 is the state vector, φ _k is the state transition matrix, U _k is the input control matrix,

Be the mean value of the acceleration at time k, W _k is a Gaussian distribution noise vector with a mean value of zero and a variance of Q _k ;

分别为RB的位置、速度和加速度；在式(12)和式(13)中，T为采样周期；在式(14)中，τ为机动频率，

为载体机动加速度的方差，q为噪声矩阵；In formula (11), x _k ,

are the position, velocity and acceleration of RB, respectively; in formula (12) and formula (13), T is the sampling period; in formula (14), τ is the maneuvering frequency,

is the variance of the vehicle’s maneuvering acceleration, and q is the noise matrix;

为时刻k的平均加速度，同时引入方差调节因子η_k＝μ(r_k)，即得到：S2.2 Selection

is the average acceleration at time k, and introduces the variance adjustment factor η _k =μ(r _k ), that is:

的更新公式为：then the acceleration variance

The update formula for is:

In formula (25): a _max and a _min represent the maximum value and minimum value of acceleration respectively;

Then the equation for filtering and correcting the position coordinates of the RB carrier is:

Preferably, the introduction process of the variance adjustment factor η _k described in step S2.2 includes:

(1) Let the innovation vector of the filter in KF be defined as:

The innovation vector d _k is ideally uncorrelated, and d _k is a Gaussian white noise with zero mean and variance S _k . When the RB carrier maneuvers, the maneuver changes the orthogonality of the innovation, making The mean value of d _k changes and is no longer zero, that is

(2) Normalize the sequence of innovation vectors to obtain statistics

(3) Establish window detection statistics, set the window size as m, and the window statistics wg k at time _k are defined as

The coefficient of determination r _k is defined as

r _k ＝wg _k /wg _k-1 (22)

When the RB carrier is not maneuvering, the r _k value is usually close to 1, and when the RB carrier is maneuvering, the r _k value increases rapidly and is proportional to the degree of maneuvering;

(4) In order to transform the mobility of the RB carrier into a variance adjustment factor, the raised half-normal distribution function μ(u) is introduced,

(5) Taking r _k as the input variable of the raised half-normal distribution function μ(u), the variance adjustment factor η _k can be obtained

_ηk = μ( _rk ) (24).

Preferably, the specific process of step 3 includes:

S3.1 Use the fuzzy evaluation method to comprehensively evaluate the relative position characteristics, position accuracy characteristics and relative motion characteristics of the RB combination, and select the optimal RB cooperative positioning combination;

S3.2 Use the position coordinates of the optimal RB and the relative distance to the VTBP to establish a cooperative positioning equation group, and use the Taylor series analysis method to solve the VTBP position coordinates to establish a cooperative positioning equation.

Preferably, the specific process of step 4 includes:

S4.1 Establish an observation equation with the solved VTBP position coordinates as the observation quantity, and establish a dead reckoning model with the velocity and heading angle measured by the VTBP itself;

S4.2 Take the relative distance and relative angle of the two workshops as observations, and establish a measurement model in conjunction with the reference vehicle position coordinates transmitted by the reference vehicle;

S4.3 Use the extended Kalman to correct the VTBP position coordinates, so as to obtain the final VTBP position coordinates, and complete the vehicle collaborative positioning.

The beneficial effects of the present invention are: the present invention discloses a vehicle cooperative positioning method based on the selective attention mechanism of the brain. Compared with the prior art, the improvement of the present invention lies in:

Aiming at the problems existing in the prior art, the present invention proposes a vehicle cooperative positioning method based on the selective attention mechanism of the brain. The influence of the geometric structure formed between VTBP), RB position accuracy and relative motion state on collaborative positioning; the selective attention mechanism of the human brain on information processing is adopted, and the above three influencing factors are used as the characteristic points of information selection, and the information is selected. Selective filtering processing; and then comprehensively screen out the optimal RB through feature integration to further improve the reliability and accuracy of vehicle collaborative positioning, which has the advantages of high positioning accuracy and no environmental restrictions.

Description of drawings

FIG. 1 is a flow chart of the vehicle cooperative positioning method based on the brain selective attention mechanism of the present invention.

FIG. 2 is a schematic diagram of the calculation process of the vehicle cooperative positioning method based on the brain selective attention mechanism of the present invention.

Fig. 3 is a schematic diagram of vehicle cooperative positioning in the VANET environment of the present invention.

Fig. 4 is a diagram of an optimal RB combination selective attention model in the present invention.

Fig. 5 is a schematic diagram of the movement track of the vehicle in the experimental verification of the first embodiment of the present invention.

Fig. 6 is a comparison diagram of experimental results of Example 1 of the method of the present invention.

Detailed ways

In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

With reference to a kind of vehicle cooperative localization method based on brain selective attention mechanism shown in accompanying drawing 1-4, comprise

Step 1: Assuming that in the VANET environment, the position coordinates of VTBP (vehicle to be positioned) are unknown quantities (x, y), select N RB (reference vehicle) combinations from neighboring vehicles or ITS infrastructure nodes according to the geometric center of gravity method, The selection rule is that the position of VTBP at the last moment should be near the geometric center of gravity of the polygon formed by the RB combination, calculate the geometric precision eigenvalues of the selected N RB combinations, and sort them in ascending order, and the smaller the geometric precision eigenvalue Indicates that the RB combination status is better; the specific process includes:

S1.1 In the VANET environment, information can be exchanged between vehicles and between vehicles and ITS infrastructure in real time (as shown in Figure 3). Suppose the position coordinates of the VTBP in the VANET environment are unknown quantities (x, y). Select n neighbor vehicles or ITS infrastructure nodes in the environment as positioning RB _i , where i=1,2,...,N; the position coordinates of RB _i are known quantities (xi _, y _i );

S1.2 The relative distance d _i between VTBP and RB _i can be obtained as:

S1.3 Calculate the geometric precision eigenvalues of the selected N RB combinations, and the calculation process is:

(1) Due to the measurement error of the ranging equipment itself, there will be an error e _i in the relative distance between the measured VTBP and the reference point RB _i in practice, and the error of the corresponding VTBP position estimation is ( _ex , e _y ), Then the formula (1) after adding the error is:

Transform (2) into a linear equation to get

a_i＝(x-x_i)/d_i，b_i＝(y-y_i)/d_i，(2) order

a _i =(xx _i )/d _i , b _i =(yy _i )/d _i ,

Formula (3) can be transformed into matrix form:

L=HX+e (4)

In formula (4): L=[l ₁ ,l ₂ ,…l _n ] ^T , X=[e _x ,e _y ] ^T , e=[e ₁ ,e ₂ ,…e _n ] ^T ,

(3) When H is full rank, H ^T H is reversible, then the error between the estimated value and the real value of position is:

The size of the error in formula (5) can be measured by covariance, and we get:

In formula (6): σ ² is the variance when the noises in e are uncorrelated;

(4) It can be seen from formula (6) that (H ^T H) ^-1 is the magnification of the distance measurement error. Therefore, the geometric position precision factor of vehicle cooperative positioning is defined as G, then

In formula (7), tr[] represents the inverse operation of seeking the matrix;

In the cooperative positioning of the vehicle to be positioned, the combination with the minimum value of G should be selected as the optimal RB combination. At this time, the amplification effect on the error is the smallest, but the calculation of G using formula (7) requires matrix inversion operation, and the calculation amount Larger will inevitably reduce the real-time performance, so the calculation of formula (7) must be simplified;

(5) Simplify the operation of formula (7), matrix inversion is equivalent to the sum of matrix eigenvalues, and the corresponding formula (7) can be written as

In formula (8): λ ₁ and λ ₂ are the eigenvalues of the matrix H ^T H;

Finally you can get

In formula (9), det[] represents the determinant of the matrix;

That is, the geometric precision eigenvalue G is obtained, and the N RB combinations are sorted according to the geometric precision eigenvalue G. When the geometric precision eigenvalue G is smaller, the RB combination status is better;

Step 2: It can be seen from the formula (1) that the position coordinates of the RB must be obtained during the collaborative positioning of the vehicle. Usually, there will be a certain error in the position coordinates of the RB, and this error will inevitably affect the accuracy of the collaborative positioning; Therefore, when selecting RBs, it is necessary to evaluate and correct the positional coordinate accuracy of RBs, and select RBs with higher positional accuracy as the optimal RB combination;

By analyzing the motion characteristics of the vehicle, it is found that the motion of the vehicle has a certain motion law in the relative time period, and this motion regularity can be described by the current statistical model of the motor vehicle;

Therefore, the present invention uses the improved current statistical model as the state equation of the vehicle, and uses the RB position coordinates sent by the RB as observations to perform error correction and accuracy assessment on the RB position coordinates, so as to eliminate the error interference to the collaborative positioning results to a certain extent ;

That is, the CSM model that introduces the variance adjustment factor is the state equation, and the position coordinates of the RB are used as observations. Formula (17) is used to correct the position coordinates of the RB in real time. Evaluate the accuracy of RB position coordinates, and evaluate the stability of RB relative motion state stability according to the RB position coordinate value;

The derivation process of using the CSM model introducing the variance adjustment factor as the state equation and taking the position coordinates of the RB as the observation quantity to correct the position coordinates of the RB in real time includes:

S2.1 Current Statistical Motion Model

During the driving process, the vehicle will have a part of the process of constant speed movement. Due to the influence of the outside world, there will be variable speed movement and frequent steering movement. The vehicle has variable acceleration characteristics. The current statistical model adopts the time-varying acceleration probability density function and the acceleration non-zero mean time correlation model. , so it can better describe the motion characteristics of the moving vehicle;

Using CSM as the motion state equation of the RB carrier, it is expressed as:

为k时刻加速度均值，W_k为均值为零、方差为Q_k的高斯分布噪声向量；所述In formula (10), X _k+1 is the state vector, φ _k is the state transition matrix, U _k is the input control matrix,

Be the mean value of the acceleration at time k, W _k is a Gaussian distribution noise vector with a mean value of zero and a variance of Q _k ;

分别为RB的位置、速度和加速度；式(12)和式(13)中，T为采样周期；式(14)中，τ为机动频率；

为载体机动加速度的方差；q为噪声矩阵；In formula (11), x _k ,

are the position, velocity and acceleration of RB respectively; in formula (12) and formula (13), T is the sampling period; in formula (14), τ is the maneuvering frequency;

is the variance of the carrier maneuvering acceleration; q is the noise matrix;

的取值为当前加速预测值，即S2.2 Select the mean value of acceleration

The value of is the current acceleration prediction value, namely

的取值为then the acceleration variance

The value is

In formula (16): a _max and a _min represent the maximum value and minimum value of acceleration respectively;

Then the equation for filtering and correcting the position coordinates is

Use formula (17) to correct the position coordinates of RB in real time;

成正比；a_max和a_min值确定后，当载体以较小的加速度机动时，系统方差较大，滤波精度较低；当载体以较大的加速度机动时，系统方差较小，滤波精度较高；From the analysis of formula (14) and formula (16), it can be concluded that the system variance Q _k of the motion model and the acceleration variance

It is directly proportional; after the values of a _max and a _min are determined, when the carrier maneuvers with a small acceleration, the system variance is large and the filtering precision is low; when the carrier maneuvers with a large acceleration, the system variance is small and the filtering precision is relatively low. high;

Therefore, the traditional CSM has higher filtering accuracy for high-mobility carriers, but lower filtering accuracy for weak-mobility carriers; CSM cannot describe the acceleration value in the interval [(4-π)a _-max /4,(4- π)a _max /4] Weak motorized motion, resulting in low filtering accuracy.

成正比，a_max和a_min值确定后，当载体以较小的加速度机动时，系统方差较大，滤波精度较低；当载体以较大的加速度机动时，系统方差较小，滤波精度较高；From the analysis of formula (14) and formula (16), it can be concluded that the system variance Q _k of the motion model and the acceleration variance

In direct proportion, after the values of a _max and a _min are determined, when the carrier maneuvers with a small acceleration, the system variance is large and the filtering precision is low; when the carrier maneuvers with a large acceleration, the system variance is small and the filtering precision is relatively low. high;

与载体机动性的不协调是导致滤波效果差的主要原因，因此，本发明针对此问题设计了方差调节因子η_k，将方差调节因子η_k引入步骤S2.2，对加速度方差

进行自适应调整，以提高CSM的滤波精度，所述方差调节因子η_k的设计过程包括：Among them, the analysis found that the acceleration variance

The inconsistency with the mobility of the carrier is the main cause of the poor filtering effect. Therefore, the present invention designs a variance adjustment factor η _k for this problem, and introduces the variance adjustment factor η _k into step S2.2, and the acceleration variance

Carry out adaptive adjustment, to improve the filter accuracy of CSM, the design process of described variance adjustment factor η _k comprises:

(1) Let the innovation vector of the filter in KF be defined as:

The innovation vector d _k is ideally uncorrelated, and d _k is a Gaussian white noise with zero mean and variance S _k . When the RB carrier maneuvers, the maneuver changes the orthogonality of the innovation, making The mean value of d _k changes and is no longer zero;

(2) The innovation vector d _k obeys the multidimensional Gaussian distribution, and its variance S _k obeys the χ ² distribution with m degrees of freedom (m is the vector dimension). Generally, the maneuverability of the carrier can be detected according to the multidimensional distribution property of the innovation and filter divergence, but this direct method is more complicated, so the statistic g _k can be obtained by normalizing the innovation sequence

(3) Establish window detection statistics, set the window size as m, and the window statistics wg k at time _k are defined as

The coefficient of determination r _k is defined as

r _k ＝wg _k /wg _k-1 (22)

When the RB carrier is not maneuvering, the r _k value is usually close to 1, and when the RB carrier is maneuvering, the r _k value increases rapidly and is proportional to the degree of maneuvering;

(4) In order to transform the mobility of the RB carrier into a variance adjustment factor, the raised half-normal distribution function μ(u) is introduced,

(5) Taking r _k as the input variable of the raised half-normal distribution function μ(u), the variance adjustment factor η _k can be obtained

η _k = μ(r _k ) (24);

That is, the variance adjustment factor η _k is obtained, and the variance of the maneuvering acceleration is adjusted in real time according to the variance adjustment factor η _k , and the acceleration variance update formula is obtained:

为时刻k的平均加速度；；In the formula:

is the average acceleration at time k;

与系统方差Q_k会自适应的调整，当载体机动性较弱时，方差调节因子η_k较小，使得

与Q_k变小以提高滤波性能，方差调节因子的引入使得滤波对于模型的不确定保持了较好的鲁棒性。According to formula (25), the acceleration variance

and the system variance Q _k will be adjusted adaptively, when the mobility of the carrier is weak, the variance adjustment factor η _k is small, so that

The Q _k becomes smaller to improve the filtering performance, and the introduction of the variance adjustment factor makes the filtering more robust to the uncertainty of the model.

Step 3: Use the fuzzy evaluation method to comprehensively evaluate the relative position characteristics, position accuracy characteristics and relative motion characteristics of RB, and select the optimal RB cooperative positioning combination; use the position coordinates of the optimal RB and the relative distance to VTBP to establish cooperation Positioning equations, and using the Taylor series analysis method to solve the VTBP position coordinates, the specific process includes:

S3.1 Use the fuzzy evaluation method to comprehensively evaluate the relative position characteristics, position accuracy characteristics and relative motion characteristics of the RB combination, and select the optimal RB cooperative positioning combination;

First, normalize the relative position feature, position accuracy feature and relative motion feature of the RB combination; the relative position feature is the geometric position accuracy value in step 1; the position accuracy feature is the original vehicle position of each RB in step 2 The average value of the difference between the value and the correction value; the relative motion feature is the average value of the difference between the RB vehicle speed and the VTBP vehicle speed;

Secondly, the normalized relative position features, position accuracy features and relative motion features are weighted and summed, and the summed value is the comprehensive evaluation value. The smaller the value, the better the corresponding RB vehicle combination;

S3.2 Use the position coordinates of the optimal RB and the relative distance to VTBP to establish a cooperative positioning equation group, and use the Taylor series analysis method to solve the VTBP position coordinates, and establish a cooperative positioning equation

The positioning coordinates are calculated by the Taylor series method, and the calculation process is as follows:

Step1: Set the estimated position of the initial coordinates (x ₀ , y ₀ ) and the error threshold. The error between the estimated position and the actual value of the coordinates is (e _x , e _y );

Step2: Expand the positioning equations at (x ₀ , y ₀ ) with Taylor series, ignoring the quadratic and above items, and get

Transform the above formula into matrix form: A _b d = b

When A _b is full rank, A _b is reversible;

It can be concluded that:

Step3: The position of the updated coordinates is x=x ₀ +e _x , y=y ₀ +e _y ; judge whether (e _x ,e _y ) is less than the set threshold, if it is less than the threshold, stop iterating; if the error is greater than the set threshold , then go to step Step2 and continue to iterate until the error is less than the set error threshold.

Step 4: Establish the observation equation with the solved VTBP position coordinates as the observation quantity, establish the dead reckoning model with the velocity and heading angle measured by the VTBP itself, and use the extended Kalman to correct the VTBP position coordinates, so as to obtain the final VTBP position coordinates, complete Vehicle collaborative positioning, the specific process includes:

S4.1 Establish the observation equation with the solved VTBP position coordinates as the observation quantity, and establish the dead reckoning model with the velocity and heading angle measured by the VTBP itself:

In formula (30): x _k is the eastward position coordinate of the vehicle; y _k is the northward position coordinate of the vehicle; v _k is the driving speed of the vehicle; θ _k is the driving steering angle.

S4.2 Establish measurement model

Taking the relative distance and relative angle of the two workshops as the observation quantity, and combining the reference vehicle position coordinates transmitted by the reference vehicle, the system measurement equation generated is

为参考车辆在k时刻的位置坐标；R_k为k时刻车辆间的相对距离；

为两车间的相对角度；

为参考车辆与X轴方向所成角度；In formula (31):

is the position coordinate of the reference vehicle at time k; R _k is the relative distance between vehicles at time k;

is the relative angle of the two workshops;

is the angle formed by the reference vehicle and the X-axis direction;

S4.3 Use extended Kalman to correct the VTBP position coordinates

In the filtering iteration process, R _k is the variance matrix of the measurement error, and the value of R _k is usually determined according to the error of the measurement vector, where

Through the above process, the final VTBP position coordinate value is obtained, and the vehicle cooperative positioning is completed; the vehicle cooperative positioning method based on the brain selective attention mechanism described in the present invention is used to specifically target the collaborative reference node (Reference Beacon, RB) and the vehicle to be positioned The influence of geometric structure formed between vehicles (Vehicle to be Positioned, VTBP), RB position accuracy and relative motion state on collaborative positioning, using the human brain's selective attention mechanism for information processing, and taking the above three factors as information selection The feature points are used to selectively filter the information; and then the optimal RB is screened out through feature integration, which further improves the reliability and accuracy of vehicle collaborative positioning.

Example 1: Data Simulation

Assume that the coverage of the Internet of Vehicles is a road area with a length of 500 meters and a width of 500 meters; there are 5 RB vehicles in this road area, and each vehicle moves along a different elliptical trajectory, and the starting coordinates are (60.0, 15.4) m, (230.0, 400.1) m, (425, 127.9) m, (320.4, 480.0) m, (480.0, 332.5) m; the movement speed is 20m/s, 40m/s, 30m/s, 45m/s, 25m/s; the trajectory is shown in Figure 5;

The initial position of the VTBP vehicle is (250.0, 90.0), and the moving speed is 50m/s; the method of the present invention and the traditional Taylor series positioning method are used in the experiment, and the number of RB vehicles selected by the two methods is 3, and the traditional Taylor series positioning method is three. The series positioning method randomly selects 3 vehicles among 5 RB vehicles; Figure 6 is a comparison chart of experimental results. In the 150s experimental verification, the average error of the method of the present invention is 1.9m, while the traditional Taylor series positioning method is 4.9m , can find out the validity of the inventive method from comparative data.

The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the industry should understand that the present invention is not limited by the above-mentioned embodiments, and what described in the above-mentioned embodiments and the description only illustrates the principles of the present invention, and the present invention will also have other functions without departing from the spirit and scope of the present invention. Variations and improvements all fall within the scope of the claimed invention. The protection scope of the present invention is defined by the appended claims and their equivalents.

Claims (4)

1. A vehicle cooperative positioning method based on brain selective attention mechanism, characterized in that: comprising steps Step 1: In the VANET environment, select N RB combinations from VTBP neighbor vehicles or ITS infrastructure nodes, calculate the geometric precision eigenvalues of the selected N RB combinations, and sort them in ascending order; The selection of N RB combinations described in step 1 and the calculation process of geometric precision eigenvalues include: S1.1 Set the position coordinates of VTBP as unknown quantities (x, y), select n neighbor vehicles or ITS infrastructure nodes from the surrounding environment as positioning RB _i , where i=1, 2,...,n; RB _i The position coordinates are known quantities (x _i , y _i ); S1.2 Get the relative distance d _i between VTBP and RB _i as:

S1.3 Calculate the geometric precision eigenvalues of the selected N RB combinations, and sort them in ascending order; Step 2: The CSM model that introduces the variance adjustment factor is the state equation, and the position coordinates of the RB are used as observations to correct the position coordinates of the RB, and at the same time evaluate the accuracy of the position coordinates of the RB; The process of real-time correction of the position coordinates of the RB described in step 2 includes: S2.1 uses CSM as the RB carrier motion state equation, expressed as:

In formula (10), X _k+1 is the state vector, φ _k is the state transition matrix, U _k is the input control matrix,

Be the mean value of the acceleration at time k, W _k is a Gaussian distribution noise vector with a mean value of zero and a variance of Q _k ;

In formula (11), x _k ,

are the position, velocity and acceleration of RB, respectively; in formula (12) and formula (13), T is the sampling period; in formula (14), τ is the maneuvering frequency,

is the variance of the vehicle’s maneuvering acceleration, and q is the noise matrix; S2.2 Selection

is the average acceleration at time k, and introduces the variance adjustment factor η _k =μ(r _k ), and obtains:

acceleration variance

The update formula for is:

In formula (25): a _max and a _min represent the maximum value and minimum value of acceleration respectively; The equation for filtering and correcting the position coordinates of the RB carrier is:

The introduction process of the variance adjustment factor η _k described in step S2.2 includes: (1) Let the innovation vector of the filter in KF be defined as:

The innovation vector D _k is ideally uncorrelated, and D _k is Gaussian white noise with zero mean and variance S _k . When the RB carrier maneuvers, the maneuver changes the orthogonality of the innovation, making The mean value of D _k changes and is no longer zero, that is

(2) Normalize the sequence of innovation vectors to obtain statistics

(3) Establish window detection statistics, set the window size as m, and the window statistics wg k at time _k are defined as

The coefficient of determination r _k is defined as r _k ＝wg _k /wg _k-1 (22) When the RB carrier is not maneuvering, the r _k value is close to 1, and when the RB carrier is maneuvering, the r _k value increases rapidly and is proportional to the degree of maneuvering; (4) In order to transform the mobility of the RB carrier into a variance adjustment factor, the raised half-normal distribution function μ(u) is introduced,

(5) Using rk as the input variable of the raised half-normal distribution function μ(u), the variance adjustment factor η _k can be obtained h _k = μ(r _k ) (24); Step 3: Comprehensively evaluate the relative position characteristics, position accuracy characteristics and relative motion characteristics of RB; use the evaluated optimal RB position coordinates and the relative distance from VTBP to establish a cooperative positioning equation group, and solve the VTBP position coordinates; Step 4: Establish a dead reckoning model with the solved VTBP position coordinates as observations, and correct the VTBP position coordinates to obtain the final VTBP position coordinates to complete vehicle collaborative positioning. 2. A kind of vehicle collaborative positioning method based on brain selective attention mechanism according to claim 1, is characterized in that: the calculation process of the geometric precision eigenvalue described in step S1.3 comprises: (1) Introduce the measurement error e _i , then the position error of the corresponding VTBP is (e _x , e _y ), then the formula (1) after adding the error is:

Transform (2) into a linear equation to get

(2) order

a _i =(xx _i )/d _i , b _i =(yy _i )/d _i , Formula (3) can be transformed into matrix form: L=HX+E (4) In formula (4): L=[l ₁ ,l ₂ ,…l _n ] ^T , X=[E _x ,E _y ] ^T , E=[E ₁ ,E ₂ ,…E _n ] ^T ,

(3) When H is full rank, H ^T H is reversible, then the error between the estimated value and the real value of position is:

The size of the error in formula (5) is measured by covariance, and we get:

In formula (6): σ ² is the variance when the noises in e are uncorrelated; (4) It can be seen from formula (6) that (H ^T H) ^-1 represents the magnification of the distance measurement error. Therefore, the geometric position precision factor of vehicle cooperative positioning is defined as G, then

Wherein, in formula (7), tr[] represents the inverse operation of seeking matrix; (5) Simplify the operation of formula (7), matrix inversion is equivalent to the sum of matrix eigenvalues, and the corresponding formula (7) can be written as

In formula (8): λ ₁ and λ ₂ are the eigenvalues of the matrix H ^T H; Finally, the geometric position precision factor of vehicle cooperative positioning can be obtained:

In formula (9), det[] means to find the determinant of the matrix. 3. a kind of vehicle collaborative positioning method based on brain selective attention mechanism according to claim 1, is characterized in that: the concrete process of step 3 comprises: S3.1 Use the fuzzy evaluation method to comprehensively evaluate the relative position characteristics, position accuracy characteristics and relative motion characteristics of the RB combination, and select the optimal RB cooperative positioning combination; S3.2 Use the position coordinates of the optimal RB and the relative distance to the VTBP to establish a cooperative positioning equation group, and use the Taylor series analysis method to solve the VTBP position coordinates to establish a cooperative positioning equation. 4. a kind of vehicle cooperative positioning method based on brain selective attention mechanism according to claim 1, is characterized in that: the concrete process of step 4 comprises: S4.1 Establish an observation equation with the solved VTBP position coordinates as the observation quantity, and establish a dead reckoning model with the velocity and heading angle measured by the VTBP itself; S4.2 Take the relative distance and relative angle of the two workshops as observations, and establish a measurement model in conjunction with the reference vehicle position coordinates transmitted by the reference vehicle; S4.3 Use the extended Kalman to correct the VTBP position coordinates, so as to obtain the final VTBP position coordinates, and complete the vehicle collaborative positioning.

Images