Intelligent vehicle target selection method, system, electronic device and storage medium

Abstract

The present invention discloses a method, system, electronic device and storage medium for intelligent vehicle target selection, and relates to the field of intelligent driving technology. The method comprises: obtaining information about the vehicle, the lane to which the vehicle belongs and the perceived target, determining the detection range according to a set prediction ratio, calculating the corner coordinates of the perceived target in the vehicle coordinate system and fitting the lane line, and obtaining the perceived target lane information. Subsequently, an attribute reduction algorithm is used to label and filter information to generate a training data set. A support vector machine algorithm is used to train a model to obtain a vehicle target selection model, which is used to detect the degree of influence of the perceived target on the vehicle's driving and determine the target selection result. The present invention improves the accuracy of target selection and the safety of intelligent vehicle driving. The present invention can improve the accuracy of target selection and the safety of intelligent vehicle driving.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving technology, and in particular to an intelligent vehicle target selection method, system, electronic device and storage medium. Background Art

[0002] The target selection system is part of the autonomous driving system. Its primary function is to process target information from the onboard perception system and, based on the target's attributes, select the targets that pose the greatest threat to the vehicle at the moment. Therefore, the accuracy of target selection directly impacts vehicle safety. This target selection method can significantly reduce the chip's computational burden and improve system efficiency. For example, the front-end perception algorithm identifies 50 targets during a vehicle's driving process. These targets include parked vehicles, motor vehicles, non-motor vehicles, and pedestrians. These targets have varying degrees of impact on the vehicle's driving. Therefore, the target selection system needs to calculate the target's impact on the vehicle based on its various attributes and select the targets with the greatest impact based on a ranking.

[0003] Since vehicles may encounter a variety of complex road conditions during driving, the current target selection method has the defects of incomplete consideration or over-reliance on a certain target attribute when considering target attributes, resulting in frequent misselection, omission or late selection of targets, which in turn causes the vehicle to brake incorrectly, not brake or brake late during driving, posing safety hazards. Summary of the Invention

[0004] The purpose of the present invention is to provide a method, system, electronic device and storage medium for intelligent vehicle target selection, which can improve the accuracy of target selection and the safety of intelligent vehicle driving.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] An intelligent vehicle target selection method, comprising:

[0007] Acquiring vehicle information, lane information, and perceived target information; the lane information includes the lane width, lane line fitting cubic polynomial, and lane line effective length corresponding to the vehicle's lane; and the perceived target information includes target attribute information.

[0008] Determining a detection range of a target selection algorithm according to a set prediction ratio based on the vehicle information, the lane information to which the vehicle belongs, and the perceived target information;

[0009] Within the detection range, the coordinates of the four corner points of the perceived target in the vehicle coordinate system are calculated, and a cubic polynomial is fitted based on the target rectangular frame generated by the four corner point coordinates and the lane line to calculate the lane information of the perceived target;

[0010] Based on an attribute reduction algorithm, the vehicle information, the lane information of the vehicle, the perceived target information, and the lane information of the perceived target are labeled and filtered to generate a training data set; the training data set includes a target selection attribute data set and a label data set; the label data set includes the degree of influence of the perceived target on the vehicle;

[0011] The support vector machine algorithm and the training data set are used to perform model training to obtain a trained vehicle target selection model, and the vehicle target selection model is used to perform detection to obtain the degree of influence of the perceived target on the vehicle's driving process, thereby determining the result of the target selection.

[0012] Optionally, the vehicle information includes the vehicle speed and steering wheel angle; the target attribute information includes the longitudinal distance, lateral distance, longitudinal speed, lateral speed, target length, target width and target orientation angle of the target rear bumper center relative to the rear axle center of the vehicle.

[0013] Optionally, the lane width includes the distance between two adjacent lane lines; the lane line fitting cubic polynomial expression is: y(x) = C0*x 3 +C1*x 2 +C2*x+C3, where C0, C1, C2, and C3 are the cubic polynomial coefficients of the fitted lane line, x is the longitudinal coordinate in the vehicle coordinate system, and y(x) is the lateral coordinate in the vehicle coordinate system.

[0014] Optionally, before calculating the coordinates of the four corner points of the perception target in the vehicle coordinate system, the method further includes: establishing a vehicle coordinate system;

[0015] The specific process of establishing the vehicle coordinate system includes:

[0016] The vehicle coordinate system is established with the center of the rear axle of the vehicle as the origin, the direction of the vehicle head as the positive direction of the vertical axis of the coordinate system, and the right direction perpendicular to the vertical axis of the coordinate system as the positive direction of the horizontal axis.

[0017] Optionally, the degree of influence of the perceived target on the vehicle specifically includes: the nearest target in front of the lane to which the vehicle belongs as level 1, the nearest target in front of the left and right adjacent lanes of the vehicle as level 2, the nearest target behind the left and right adjacent lanes of the vehicle as level 3, the nearest target behind the lane to which the vehicle belongs as level 4, the second nearest target in front of the lane to which the vehicle belongs as level 5, the second nearest target in front of the left and right adjacent lanes of the vehicle as level 6, and all other targets within the detection range of the target selection algorithm as level 7.

[0018] The present invention also provides an intelligent vehicle target selection system, comprising:

[0019] a data acquisition unit configured to acquire vehicle information, lane information, and perceived target information; the lane information includes the lane width, lane line fitting cubic polynomial, and lane line effective length corresponding to the vehicle's lane; and the perceived target information includes target attribute information;

[0020] an algorithm range determination unit, configured to determine a detection range of a target selection algorithm according to a set prediction ratio based on the vehicle information, the lane information to which the vehicle belongs, and the perceived target information;

[0021] a lane information calculation unit, configured to calculate the coordinates of four corner points of a perceived target in the vehicle coordinate system within the detection range, and to fit a cubic polynomial to a target rectangular box generated by the four corner point coordinates and the lane line to calculate the lane information of the perceived target;

[0022] a data set generation unit, configured to label and filter the ego vehicle information, the ego vehicle lane information, the perceived target information, and the perceived target lane information based on an attribute reduction algorithm to generate a training data set; the training data set includes a target selection attribute data set and a label data set; the label data set includes the degree of influence of the perceived target on the ego vehicle;

[0023] The target selection application unit is used to use the support vector machine algorithm and the training data set to perform model training to obtain a trained vehicle target selection model, and use the vehicle target selection model to perform detection to obtain the degree of influence of the perceived target on the vehicle's driving process, thereby determining the result of the target selection.

[0024] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the above-mentioned intelligent vehicle target selection method.

[0025] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the intelligent vehicle target selection method as described above.

[0026] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0027] The present invention discloses a method, system, electronic device and storage medium for intelligent vehicle target selection. The method comprises obtaining information about the vehicle, the lane to which the vehicle belongs and the perceived target, determining the detection range according to a set prediction ratio, calculating the corner coordinates of the perceived target in the vehicle coordinate system and fitting the lane line to obtain the perceived target lane information. Subsequently, an attribute reduction algorithm is used to label and filter the information to generate a training data set. A support vector machine algorithm is used to train the model to obtain a vehicle target selection model, which is used to detect the degree of influence of the perceived target on the vehicle's driving and determine the target selection result. The present invention improves the accuracy of target selection and the safety of intelligent vehicle driving. The present invention can improve the accuracy of target selection and the safety of intelligent vehicle driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0029] Figure 1 A schematic flow chart of the target selection method for an intelligent vehicle according to the present invention;

[0030] Figure 2 A coordinate diagram of the target selection method for an intelligent vehicle in this embodiment;

[0031] Figure 3 Schematic diagram of another coordinate system of the target selection method for an intelligent vehicle in this embodiment;

[0032] Figure 4 This is a schematic diagram of the detection range of the target selection algorithm in this embodiment. DETAILED DESCRIPTION

[0033] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0034] The purpose of the present invention is to provide a method, system, electronic device and storage medium for intelligent vehicle target selection, which can improve the accuracy of target selection and the safety of intelligent vehicle driving.

[0035] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0036] like Figure 1 As shown, the present invention provides a method for selecting a target for an intelligent vehicle, comprising:

[0037] Step 100: Acquire vehicle information, lane information, and perceived target information; the lane information includes the lane width, lane line fitting cubic polynomial, and lane line effective length corresponding to the vehicle's lane; and the perceived target information includes target attribute information.

[0038] Step 200: Based on the vehicle information, the lane information of the vehicle, and the perceived target information, a detection range of the target selection algorithm is determined according to a set prediction ratio.

[0039] Step 300: Within the detection range, calculate the coordinates of the four corner points of the perceived target in the vehicle coordinate system, and fit a cubic polynomial based on the target rectangular box generated by the four corner point coordinates and the lane line to calculate the lane information of the perceived target.

[0040] Step 400: Based on an attribute reduction algorithm, the vehicle information, the lane information of the vehicle, the perceived target information, and the lane information of the perceived target are labeled and filtered to generate a training data set; the training data set includes a target selection attribute data set and a label data set; the label data set includes the degree of influence of the perceived target on the vehicle.

[0041] Step 500: Use the support vector machine algorithm and the training data set to perform model training to obtain a trained vehicle target selection model, and use the vehicle target selection model to perform detection to obtain the degree of influence of the perceived target on the vehicle's driving process, thereby determining the result of the target selection.

[0042] As a specific implementation method, the calculation process of Example 1 is provided as follows:

[0043] Obtain vehicle information of the own vehicle, lane information of the own vehicle and the perceived target lane, and attribute information of the perceived target; wherein the own vehicle information includes the own vehicle's speed and steering wheel angle, and the own vehicle's lane information includes the own vehicle's lane, lane width, coefficients of a cubic polynomial fitting lane line, and effective length of the lane line; attribute information of the perceived target includes the longitudinal distance, lateral distance, longitudinal speed, lateral speed, target length, target width, and target heading angle of the target rear bumper center relative to the center of the own vehicle's rear axle; the perceived target lane information includes at least lane width, coefficients of a cubic polynomial fitting lane line, and effective length of the lane line.

[0044] Determining a detection range of a target selection algorithm based on the vehicle information, the lane information, and a set prediction ratio includes:

[0045] Based on the lane width, the coefficients of the cubic polynomial fitting lane line, and the effective length of the lane line, the target selection range is determined according to a certain prediction ratio, where the prediction ratio is set according to the specific situation; the lane width includes the distance between two adjacent lane lines; the environment perception system obtains the x and y coordinates of the lane line, and based on the coordinates, determines the coefficients of the cubic polynomial fitting lane line, specifically expressed as: y(x) = C0*x3+C1*x2+C2*x+C3, where C0, C1, C2, and C3 are the coefficients of the cubic polynomial fitting lane line, x is the longitudinal coordinate in the vehicle coordinate system, and y(x) is the lateral coordinate in the vehicle coordinate system; the effective length of the lane line is obtained by detection and identification by the intelligent vehicle environment perception system.

[0046] Based on the target length, width, longitudinal coordinate, transverse coordinate, and target orientation angle information, the coordinates of the corner points of the target relative to the vehicle coordinate system are calculated, specifically including:

[0047] Based on the target length, the target width, the target longitudinal coordinate, the target lateral coordinate and the target heading angle information, the corner point coordinates of the target relative to the vehicle are calculated; wherein the corner point coordinates include the left front corner point coordinates, the right front corner point coordinates, the left rear corner point coordinates and the right rear corner point coordinates of the target.

[0048] Before obtaining the target information and calculating the target corner coordinates, the vehicle coordinate system must be determined, including:

[0049] The vehicle coordinate system is established with the center of the rear axle of the vehicle as the origin, the direction of the vehicle head as the positive direction of the vertical axis of the coordinate system, and the right direction perpendicular to the vertical axis of the coordinate system as the positive direction of the horizontal axis.

[0050] When the rectangular box defined by the corner point coordinates does not intersect with the two adjacent lane lines, the lane number of the current target is determined to be between the two lane lines; when the rectangular box defined by the corner point coordinates intersects with one of the two adjacent lane lines, the lane number of the current target is determined based on the ratio of the lane lines that divide the rectangular box.

[0051] Furthermore, a target selection attribute dataset is established based on vehicle driving data through manual labeling, wherein the labeled target attributes include target type, target orientation, lane number of the target, target longitudinal position, target lateral position, target longitudinal speed, target lateral speed, target motion state, and whether the target is braking.

[0052] Based on the target selection attribute dataset, a target selection label dataset is established by manual labeling, wherein the content of the label includes the degree of influence of the target on the ego vehicle at the current moment, wherein the degree of influence is divided into 7 levels, specifically including: the nearest target in front of the lane to which the ego vehicle belongs as level 1, the nearest target in front of the left and right adjacent lanes of the ego vehicle as level 2, the nearest target behind the left and right adjacent lanes of the ego vehicle as level 3, the nearest target behind the lane to which the ego vehicle belongs as level 4, the second nearest target in front of the lane to which the ego vehicle belongs as level 5, the second nearest target in front of the left and right adjacent lanes of the ego vehicle as level 6, and all other targets within the detection range of the target selection algorithm as level 7.

[0053] Given that there are many attributes involved in the calculation during target selection, attribute reduction is performed on the attribute dataset based on the attribute reduction method, specifically including:

[0054] Based on the attribute data set and the label data set, attributes having a greater impact in the target selection process are obtained according to an attribute reduction method.

[0055] In this embodiment, based on the attribute reduction results, attributes with greater influence are selected to form a new target selection attribute dataset, which is then used as the corresponding target selection label dataset. A support vector machine algorithm is used to train and validate the dataset. Specifically, based on the attribute dataset and the label dataset, 80% is randomly selected as training data and the remaining 20% ​​as validation data to verify the training effect of the support vector machine.

[0056] As a specific implementation method, the calculation process of Example 2 is provided as follows:

[0057] S1: Obtain information about the vehicle, the lane it belongs to, and the perceived target.

[0058] S2: Determine a detection range of a target selection algorithm based on the vehicle information, lane information, target attributes, and the predicted ratio.

[0059] S3: Calculate the coordinates of the four corner points of the target in the vehicle coordinate system based on the target information.

[0060] S4: Determine whether the rectangular frame formed by the corner point coordinates and the cubic polynomial fitted based on the lane line intersect, and determine the lane to which the target belongs according to the ratio of the lane line segmentation target frame.

[0061] S5: Based on the lane to which the target belongs, the target information and the lane to which the vehicle belongs, a target selection attribute dataset and a label dataset are constructed by manual labeling.

[0062] S6: Based on the target selection attribute data set and the label data set, the attribute reduction algorithm is used to obtain the target attributes that have a greater impact on the driving of the vehicle, and the new target selection attribute data set is established.

[0063] S7: Based on the new target selection attribute data set and the label data set, a support vector machine algorithm is used to train and verify the data set, and a target selection result is determined based on the target influence degree level.

[0064] In a specific implementation, the vehicle information, the lane information of the vehicle and the target information are first obtained; the vehicle information includes the vehicle speed and steering wheel angle; the attribute information and lane information of the target are obtained through the perception system, which may include an inertial navigation system, a lidar sensor, a camera sensor or a millimeter-wave radar and other sensors, which are set on the vehicle to collect target information from the vehicle and the surrounding environment. The number and distribution of sensors are determined by the actual vehicle model. For example, a front-view camera, four side-view cameras and a front millimeter-wave radar can be set, and no specific limitation is made here.

[0065] Based on the lane information, the target attributes and the predicted ratio, the detection range of the target selection algorithm of the target detection area is determined, the coordinates of the four corner points of the target in the vehicle coordinate system are calculated based on the target information, and the lane information to which the target belongs is calculated based on the cubic polynomial fitted by the lane line and the target corner point coordinates. The target selection attribute dataset and the label dataset are composed based on the manual labeling method.

[0066] The target selection attribute dataset and the label dataset are processed based on the attribute reduction algorithm, and a new target selection attribute dataset is established. The new target selection attribute dataset and the label dataset are trained and verified based on the support vector machine algorithm, and the result of target selection is determined based on the impact level of the target on the driving process of the vehicle.

[0067] In summary, the embodiment of the present invention calculates the detection range and target corner coordinates of the target selection algorithm based on the intelligent vehicle driving data, establishes a target selection attribute dataset and a target selection label dataset through manual labeling, and obtains a new target selection attribute dataset based on the attribute reduction algorithm. The support vector machine algorithm is used to calculate the degree of influence of the target on the vehicle's driving process, thereby determining the selected target, improving the accuracy and timeliness of target selection and the driving safety of the intelligent vehicle.

[0068] In this embodiment, the detection range of the target selection algorithm is determined based on the lane information, the target attributes, and the predicted ratio. Specifically, the detection range of the target selection algorithm is calculated using the following formula:

[0069] Left edge of the target selection algorithm's detection range: y L (x) = C0 L *x 3 +C1 L *x 2 +C2 L *x+C3 L .

[0070] Right edge of the target selection algorithm's detection range: y R (x) = C0 R *x 3 +C1 R *x 2 +C2 R *x+C3 R .

[0071] The upper boundary of the target selection algorithm's detection range: x T =k T *L T .

[0072] Lower boundary of the target selection algorithm detection range: x B =k B *L B .

[0073] Among them, x is the longitudinal coordinate in the vehicle coordinate system, y L (x), y R (x) is the lateral coordinate in the corresponding vehicle coordinate system, C0 L 、C1 L 、C2 L 、C3 L 、C0 R 、C1 R 、C2 R 、C3 R is the cubic polynomial coefficient of the lane line fitting, x T 、xB The upper and lower boundary ordinates of the lane line are taken as values, L T , L B are the effective lengths of the lane lines in front and behind the vehicle, respectively, k T 、k B They are the predicted ratios of the lane lines in front and behind the vehicle, respectively. They can be adjusted according to the actual lane line conditions and are not specifically limited here.

[0074] In this embodiment, the four corner point coordinates of the target relative to the vehicle are calculated based on the target length, the target width, the target longitudinal coordinate, the target lateral coordinate and the target orientation angle information, wherein the corner point coordinates include the left front corner point coordinates, the right front corner point coordinates, the left rear corner point coordinates and the right rear corner point coordinates of the target.

[0075] In this embodiment, before calculating the coordinates of the target corner point, it is necessary to first establish a coordinate system, such as Figure 2 As shown, Figure 2 The figure shows a schematic diagram of an intelligent vehicle target selection method based on support vector machine and attribute reduction provided by the present invention, with the center of the rear axle of the vehicle as the origin of the coordinate system, the direction of the front of the vehicle as the positive direction of the vertical axis, and the right direction perpendicular to the vertical axis as the positive direction of the horizontal axis.

[0076] As an example, Figure 3 As shown, based on the longitudinal coordinates, lateral coordinates, target length, target width, and target orientation angle of the target rear bumper midpoint relative to the vehicle obtained by the perception system, the coordinates of the left front corner, right front corner, left rear corner, and right rear corner of the target can be calculated. The specific calculation method can be referred to the following calculation formula:

[0077] Coordinates of the left front corner (x FL ,y FL ):

[0078] x FL =x v +length*sin(θ)+0.5*width*sin(θ);

[0079] y FL =y v +length*cos(θ)-0.5*width*cos(θ);

[0080] Coordinates of the right front corner (x FR ,y FR ):

[0081] x FR =x v +length*sin(θ)-0.5*width*sin(θ);

[0082] y FR =y v +length*sin(θ)+0.5*width*cos(θ);

[0083] Left rear corner coordinates (x RL ,y RL ):

[0084] x RL =x v +0.5*width*sin(θ);

[0085] y RL =y v -0.5*width*cos(θ);

[0086] Coordinates of the right rear corner (x RR ,y RR ):

[0087] x RR =x v -0.5*width*sin(θ);

[0088] y RR =y v +0.5*width*cos(θ);

[0089] Among them, length and width are the length and width of the target respectively, x v 、y v are the longitudinal coordinate and lateral coordinate of the target rear bumper center relative to the vehicle, and θ is the target's heading angle.

[0090] Based on the above-obtained vehicle information, lane information, target information, and lane information of the target, we integrate them together to establish the attribute dataset for target selection. We calculate the target selected for each frame of driving data through manual labeling. The labeling content is the impact level of the target on the driving process of the vehicle, thereby establishing the label dataset for target selection.

[0091] Based on the attribute reduction algorithm, attribute reduction is performed on the attribute data set of the target selection. The definition of the attribute reduction algorithm includes:

[0092] Define a decision system S for the impact level of a target on the vehicle's driving. The expression formula of the decision system is S = (U, C, D), where U is the entire set of research objects, which in this invention is the entire set of labeled target selection attribute datasets and target selection label datasets; C is the conditional attribute, which in this invention is the set of target selection attribute datasets; D is the decision attribute, which in this invention is the set of target selection label datasets. The regular domain expression of the decision attribute D under the conditional attribute C is as follows:

[0093] The regular expression of decision attribute D under condition attribute C is:

[0094] For a given decision system S = (U, C, D), if POS C (D)=POS C-a (D), then attribute a is said to be omitable for decision attribute D in condition attribute C, otherwise attribute a is said to be non-omissible for decision attribute D in condition attribute C, where a is one of the attributes included in the condition attribute.

[0095] For a given decision system S = (U, C, D), when the condition attribute C is reduced to the decision attribute D, there exists a non-empty subset P. If the following conditions are met, the non-empty subset P is called the reduction result of the condition attribute C:

[0096] First, for any conditional attribute a∈P, the conditional attribute a cannot be omitted for the decision attribute D.

[0097] Second, POS P (D)=POS C (D).

[0098] In summary, the target selection condition attribute dataset is processed based on the attribute reduction algorithm combined with the target selection label dataset to obtain a reduced target selection condition attribute set P, and P is used as a new target selection condition attribute dataset.

[0099] The target selection condition attribute dataset and the target selection label attribute dataset are trained and validated based on the support vector machine algorithm. The ratio of the training dataset data volume to the validation dataset data volume is set to 8:2. The support vector machine algorithm obtains the degree of influence of the target on the driving process of the vehicle by finding the classification hyperplane. The specific expression is as follows:

[0100] The hyperplane expression is: w1*x1+w2*x2+…+w n *x n +b=0.

[0101] Based on the degree of influence of the target on the driving process of the vehicle, there are 7 levels in total, so 6 hyperplane expressions need to be established; the calculation is based on the support vector machine algorithm to complete the degree of influence of the target on the driving process of the vehicle, thereby completing the intelligent vehicle target selection based on support vector machine and attribute reduction.

[0102] For the determination of the degree of influence of the target on the driving of the vehicle, see Figure 4 ,like Figure 4 As shown, after the detection range is determined based on the target selection algorithm, the nearest target in front of the lane to which the vehicle belongs is set as level 1, the nearest target in front of the left and right adjacent lanes of the vehicle is set as level 2, the nearest target behind the left and right adjacent lanes of the vehicle is set as level 3, the nearest target behind the lane to which the vehicle belongs is set as level 4, the next closest target in front of the lane to which the vehicle belongs is set as level 5, the next closest target in front of the left and right adjacent lanes of the vehicle is set as level 6, and all other targets within the detection range of the target selection algorithm are set as level 7.

[0103] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0104] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for selecting a target for an intelligent vehicle, characterized in that: include: Obtain information about the vehicle, its lane, and the perceived target; The lane information of the vehicle includes the lane width, lane line fitting cubic polynomial and lane line effective length corresponding to the lane of the vehicle; the perceived target information includes target attribute information; Determining a detection range of a target selection algorithm according to a set prediction ratio based on the vehicle information, the lane information to which the vehicle belongs, and the perceived target information; Within the detection range, the coordinates of the four corner points of the perceived target in the vehicle coordinate system are calculated, and a cubic polynomial is fitted based on the target rectangular frame generated by the four corner point coordinates and the lane line to calculate the lane information of the perceived target; Based on an attribute reduction algorithm, the vehicle information, the lane information of the vehicle, the perceived target information, and the lane information of the perceived target are labeled and filtered to generate a training data set; The training data set includes a target selection attribute data set and a label data set; the label data set includes the degree of influence of the perceived target on the vehicle; The support vector machine algorithm and the training data set are used to perform model training to obtain a trained vehicle target selection model, and the vehicle target selection model is used to perform detection to obtain the degree of influence of the perceived target on the vehicle's driving process, thereby determining the result of the target selection.

2. The intelligent vehicle target selection method according to claim 1, characterized in that: The vehicle information includes the vehicle speed and steering wheel angle; the target attribute information includes the longitudinal distance, lateral distance, longitudinal speed, lateral speed, target length, target width and target heading angle of the target rear bumper center relative to the rear axle center of the vehicle.

3. The intelligent vehicle target selection method according to claim 1, characterized in that: The lane width includes the distance between two adjacent lane lines; the lane line fitting cubic polynomial expression is: y(x) = C0*x 3 +C1*x 2 +C2*x+C3, where C0, C1, C2, and C3 are the cubic polynomial coefficients of the fitted lane line, x is the longitudinal coordinate in the vehicle coordinate system, and y(x) is the lateral coordinate in the vehicle coordinate system.

4. The intelligent vehicle target selection method according to claim 1, characterized in that: Before calculating the coordinates of the four corner points of the perceived target in the vehicle coordinate system, the process also includes: establishing a vehicle coordinate system; The specific process of establishing the vehicle coordinate system includes: The vehicle coordinate system is established with the center of the rear axle of the vehicle as the origin, the direction of the vehicle head as the positive direction of the vertical axis of the coordinate system, and the right direction perpendicular to the vertical axis of the coordinate system as the positive direction of the horizontal axis.

5. The intelligent vehicle target selection method according to claim 1, characterized in that: The degree of influence of the perceived target on the ego vehicle specifically includes: the nearest target in front of the ego vehicle's lane as level 1, the nearest target in front of the ego vehicle's left and right adjacent lanes as level 2, the nearest target behind the ego vehicle's left and right adjacent lanes as level 3, the nearest target behind the ego vehicle's lane as level 4, the next closest target in front of the ego vehicle's lane as level 5, the next closest target in front of the ego vehicle's left and right adjacent lanes as level 6, and all other targets within the detection range of the target selection algorithm as level 7.

6. An intelligent vehicle target selection system, characterized in that: include: A data acquisition unit is used to obtain information about the vehicle, the lane it belongs to, and the perceived target; The lane information of the vehicle includes the lane width, lane line fitting cubic polynomial and lane line effective length corresponding to the lane of the vehicle; the perceived target information includes target attribute information; an algorithm range determination unit, configured to determine a detection range of a target selection algorithm according to a set prediction ratio based on the vehicle information, the lane information to which the vehicle belongs, and the perceived target information; a lane information calculation unit, configured to calculate the coordinates of four corner points of a perceived target in the vehicle coordinate system within the detection range, and to fit a cubic polynomial to a target rectangular box generated by the four corner point coordinates and the lane line to calculate the lane information of the perceived target; a data set generating unit, configured to label and filter the vehicle information, the lane information of the vehicle, the perceived target information, and the lane information of the perceived target based on an attribute reduction algorithm to generate a training data set; The training data set includes a target selection attribute data set and a label data set; the label data set includes the degree of influence of the perceived target on the vehicle; The target selection application unit is used to use the support vector machine algorithm and the training data set to perform model training to obtain a trained vehicle target selection model, and use the vehicle target selection model to perform detection to obtain the degree of influence of the perceived target on the vehicle's driving process, thereby determining the result of the target selection.

7. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the intelligent vehicle target selection method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that It stores a computer program, which, when executed by a processor, implements the intelligent vehicle target selection method according to any one of claims 1 to 5.

Images