Lane line matching positioning method, electronic device and storage medium

Abstract

The present invention discloses a lane line matching and positioning method, electronic device and storage medium, the method comprising: obtaining lane lines identified by a vehicle camera device, and using the points of the lane lines identified by the vehicle camera device as photographed lane line points; obtaining the map lane lines of the vehicle positioning position in the map, and using the points on the map lane lines as map lane line points; matching the photographed lane line points with the map lane line points; constructing an error function and an objective function; iteratively solving the objective function; correcting the vehicle positioning position based on the transformation matrix solved at the end of the iteration, and obtaining the confidence of the positioning position. The present invention obtains the optimal matching result by optimizing the objective function. In addition, the present invention can use the matching error after the optimization to calculate the confidence of the lane line recognition in real time, and adjust the weight of the camera lane line recognition result in the fusion positioning according to the real-time confidence, thereby reducing the impact of the randomness of the recognition error on the fusion positioning accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of automobiles, and in particular to a lane line matching and positioning method, electronic equipment, and storage medium. Background Art

[0002] High-level autonomous driving relies on high-precision maps. In order to ensure that the vehicle can be accurately positioned on the high-precision map, a multi-sensor fusion positioning method based on lane line and map matching can be used.

[0003] When using a camera to identify lane markings for vehicle positioning, the positioning effect is affected by lane marking recognition errors, so correction of these errors is necessary during positioning. As is well known, the errors in lane marking recognition by cameras are subject to randomness, especially in conditions such as weather, lighting, water accumulation, and blurred lane markings.

[0004] Current lane matching localization methods match a discrete set of map points with a set of lane marking points. However, lane markings on straight roads lack longitudinal localization features, which affects confidence calculations. Furthermore, the sampling distance of the discrete point set affects matching accuracy. Consequently, current localization evaluation methods cannot effectively evaluate localization performance in real time. Summary of the Invention

[0005] Based on this, it is necessary to provide a lane line matching and positioning method, electronic device and storage medium to address the technical problems of the existing lane line matching and positioning methods with low matching accuracy.

[0006] The present invention provides a lane line matching and positioning method, comprising:

[0007] Obtaining a lane line identified by a vehicle camera device, and using the lane line point identified by the vehicle camera device as a captured lane line point;

[0008] Get the map lane line of the vehicle's location in the map, and use the point on the map lane line as the map lane line point;

[0009] Match the captured lane marking points with the map lane marking points;

[0010] Constructing an error function for the lateral position error between the captured lane line point and the matched map lane line point using the transformation matrix as a variable, and constructing an objective function for the error function;

[0011] Iteratively solve the objective function through the gradient descent algorithm;

[0012] The vehicle positioning position is corrected based on the transformation matrix solved at the end of the iteration, and a corrected positioning position is obtained by correcting the vehicle positioning position based on the lane line recognized by the camera device, and a confidence level of the corrected positioning position is obtained.

[0013] Furthermore, matching the captured lane line points with the map lane line points specifically includes:

[0014] For each captured lane line point, draw a perpendicular line from the captured lane line point to the map lane line, and use the point of the foot of the perpendicular on the map lane line as the map lane line point corresponding to the captured lane line point.

[0015] Furthermore, the transformation matrix includes a rotation matrix and a lateral translation variable. The construction of an error function regarding the lateral position error between the captured lane line point and the matched map lane line point using the transformation matrix as a variable specifically includes:

[0016] Constructing the error function , where n is the number of lane line points, R is the rotation matrix, t is the lateral translation variable, and p i is the coordinate of the i-th lane line point, q i is the coordinate of the map lane line point corresponding to the i-th captured lane line point, and the captured lane line point and the corresponding map lane line point are taken as a point pair, w i is the weight of the i-th point pair, w i =(1-p ix / MaxX) / SumW, p ix is the longitudinal distance of the i-th lane line point, MaxX is the maximum longitudinal distance among all point pairs, .

[0017] Furthermore, constructing an objective function related to the error function specifically includes:

[0018] Constructing the objective function , where E is the error function.

[0019] Furthermore, the vehicle positioning position is corrected based on the transformation matrix solved at the end of the iteration to obtain a corrected positioning position based on the lane line recognized by the camera device, specifically including:

[0020] A transformation matrix is constructed based on the rotation matrix R and the lateral translation variable t solved at the end of the iteration, and the vehicle positioning position is corrected according to the transformation matrix to obtain a corrected positioning position of the vehicle positioning position based on the lane line recognized by the camera device.

[0021] Furthermore, obtaining the confidence level of the corrected positioning position specifically includes:

[0022] The confidence of the transformation matrix is calculated as the confidence of the corrected positioning position.

[0023] Furthermore, the calculating the confidence of the transformation matrix specifically includes:

[0024] Obtain the error value of the error function when solving the objective function for the last iteration as the data error;

[0025] According to the data error, a positioning error solution equation is constructed based on the Hessian matrix representation of the objective function relative to the transformation matrix;

[0026] An equation is solved according to the positioning error to determine the positioning error, and a confidence level of the transformation matrix is determined according to the positioning error.

[0027] Furthermore, it also includes:

[0028] The corrected positioning position is fused with other positioning sources, wherein the fusion weight of the corrected positioning position is calculated based on the confidence level.

[0029] The present invention provides an electronic device, comprising:

[0030] at least one processor; and,

[0031] a memory communicatively connected to at least one of the processors; wherein,

[0032] The memory stores instructions that can be executed by at least one of the processors. The instructions are executed by at least one of the processors to enable the at least one processor to perform the lane matching and positioning method as described above.

[0033] The present invention provides a storage medium, which stores computer instructions. When a computer executes the computer instructions, it is used to execute all steps of the lane line matching and positioning method as described above.

[0034] The present invention uses the lateral matching position error between the map point set and the camera lane line recognition result to construct an error function, and obtains the optimal matching result by reducing the error function. The optimal matching result is combined with satellite positioning, inertial navigation positioning, etc. to obtain an estimate of the vehicle's position and posture as the vehicle fusion positioning result. Furthermore, the present invention can use the optimized residual matching error to calculate the confidence of lane line recognition in real time, and adjust the weight of the camera lane line recognition result in fusion positioning according to the real-time confidence, thereby reducing the impact of the randomness of the recognition error on the accuracy of fusion positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a workflow diagram of a lane line matching and positioning method of the present invention;

[0036] Figure 2 This is a workflow diagram of a lane matching and positioning method according to one embodiment of the present invention;

[0037] Figure 3 The following is a flowchart of the matching and positioning module of the best embodiment of the present invention;

[0038] Figure 4 This is a workflow diagram of the confidence assessment module of the best embodiment of the present invention;

[0039] Figure 5 A workflow diagram of lane line matching and positioning according to the best embodiment of the present invention;

[0040] Figure 6 The lane line correction simulation test results using the lane line matching and positioning of the present invention are as follows;

[0041] Figure 7 This is the GPS correction simulation test result using the lane matching positioning of the present invention;

[0042] Figure 8 The figure is a schematic diagram of the hardware structure of an electronic device of the present invention. DETAILED DESCRIPTION

[0043] The following further describes specific embodiments of the present invention with reference to the accompanying drawings. Like components are denoted by like reference numerals. It should be noted that the terms "front," "rear," "left," "right," "up," and "down" used in the following description refer to directions in the accompanying drawings, and the terms "inward" and "outward" refer to directions toward or away from the geometric center of a particular component, respectively.

[0044] Example 1

[0045] like Figure 1 The flowchart of a lane matching and positioning method of the present invention is shown, including:

[0046] Step S101, obtaining a lane line recognized by a vehicle camera device, and using the lane line point recognized by the vehicle camera device as a captured lane line point;

[0047] Step S102, obtaining a map lane line of the vehicle's location in the map, and using a point on the map lane line as a map lane line point;

[0048] Step S103, matching the captured lane line points with the map lane line points;

[0049] Step S104, constructing an error function for the lateral position error between the captured lane line point and the matched map lane line point using the transformation matrix as a variable, and constructing an objective function for the error function;

[0050] Step S105, iteratively solving the objective function using a gradient descent algorithm;

[0051] Step S106 , correcting the vehicle positioning position based on the transformation matrix solved at the end of the iteration, obtaining a corrected positioning position based on the lane line recognized by the camera device, and obtaining the confidence of the corrected positioning position.

[0052] Specifically, the present invention can be applied to an electronic control unit (ECU) of a vehicle.

[0053] In step S101, the lane lines outside the vehicle are photographed by, for example, a camera, and the points of the lane lines obtained by the photographing are used as photographed lane line points. All the photographed lane line points are regarded as a point set. Then, in step S102, the vehicle positioning position, such as the Global Positioning System (GPS) positioning information, is simultaneously obtained, and the map lane line of the positioning position is obtained from the preset high-precision map, and the points on the map lane line are used as map lane line points. All the map lane line points are regarded as another point set. Then, in step S103, the map lane line points corresponding to each photographed lane line point are determined, and an error function is constructed in step S104. The error function uses the transformation matrix as a variable to calculate the lateral position error between the photographed lane line point and the matched map lane line point. At the same time, an objective function of the error function is constructed, and the objective function is to solve the transformation matrix that minimizes the value of the error function.

[0054] In step S105, the objective function is iteratively solved using a gradient descent algorithm. When the number of iterations is greater than a threshold or the error is less than a threshold, the iteration ends. The transformation matrix obtained at the end of the iteration is the transformation matrix that minimizes the value of the error function.

[0055] Finally, step S106 corrects the vehicle positioning position based on the transformation matrix, for example, multiplying the transformation matrix by the vehicle positioning position to obtain a corrected positioning position, and calculates the confidence of the corrected positioning position.

[0056] This module uses lane lines identified by the camera, GPS positioning information, and preset high-precision maps to match the lane lines with the high-precision map in lateral position to obtain the vehicle's position on the map.

[0057] The present invention uses the lateral position error between point sets to construct an error function and estimate the confidence error, which can eliminate the confidence calculation error caused by the lack of longitudinal positioning features in lane lines and accurately measure the positioning effect.

[0058] Example 2

[0059] like Figure 2 FIG. 1 is a flowchart of a lane matching and positioning method according to an embodiment of the present invention, including:

[0060] Step S201: Acquire lane lines recognized by the vehicle camera device, and use the points of the lane lines recognized by the vehicle camera device as captured lane line points.

[0061] Step S202: Obtain the map lane line of the vehicle positioning position in the map, and use the point on the map lane line as the map lane line point.

[0062] Step S203: For each captured lane line point, draw a perpendicular line from the captured lane line point to the map lane line, and use the point of the foot of the perpendicular on the map lane line as the map lane line point corresponding to the captured lane line point.

[0063] Step S204: The transformation matrix includes a rotation matrix and a lateral translation variable, and an error function is constructed. , where n is the number of lane line points, R is the rotation matrix, t is the lateral translation variable, and p i is the coordinate of the i-th lane line point, q i is the coordinate of the map lane line point corresponding to the i-th captured lane line point, and the captured lane line point and the corresponding map lane line point are taken as a point pair, w i is the weight of the i-th point pair, w i =(1-p ix / MaxX) / SumW,p ix is the longitudinal distance of the i-th lane line point, MaxX is the maximum longitudinal distance among all point pairs, .

[0064] Step S205: constructing the objective function , where E is the error function.

[0065] Step S206: Iteratively solve the objective function using a gradient descent algorithm.

[0066] Step S207, constructing a transformation matrix based on the rotation matrix R and the lateral translation variable t solved at the end of the iteration, and correcting the vehicle positioning position according to the transformation matrix to obtain a corrected positioning position of the vehicle based on the lane line recognized by the camera device;

[0067] Step S208: Calculate the confidence of the transformation matrix as the confidence of the corrected positioning position.

[0068] In one embodiment, calculating the confidence of the transformation matrix specifically includes:

[0069] Obtain the error value of the error function when solving the objective function for the last iteration as the data error;

[0070] According to the data error, a positioning error solution equation is constructed based on the Hessian matrix representation of the objective function relative to the transformation matrix;

[0071] An equation is solved according to the positioning error to determine the positioning error, and a confidence level of the transformation matrix is determined according to the positioning error.

[0072] Step S209: fusing the corrected positioning position with other positioning sources, wherein the fusion weight of the corrected positioning position is calculated based on the confidence level.

[0073] Specifically, this embodiment consists of a matching and positioning module and a confidence assessment module, wherein:

[0074] Matching and positioning module: This module uses lane lines identified by the camera, GPS positioning information, and preset high-precision maps to match the lane lines with the high-precision map in lateral position to obtain the vehicle's position on the map.

[0075] 1) Matching Lane Points with Map Vectors: Steps S201 through S203 match the captured lane points with map lane points using map vectors. Specifically, the captured lane points are matched with vector features in a pre-set high-precision map. Lane features in the map are represented as line segments, which are represented as endpoints and equations. The lane points captured by the camera are output in the vehicle coordinate system. This is then transferred to the map coordinate system via the vehicle coordinate system and GPS. A perpendicular line is then drawn between the captured lane point and the line segment in the map coordinate system. The point corresponding to the foot of the perpendicular on the line segment is the matching point on the map corresponding to the captured lane point. Generally speaking, GPS positioning error does not exceed one lane. Therefore, after GPS positioning is performed on the map, matching points are searched within that lane. The camera distinguishes between left and right lanes when identifying lane lines, and then selects the corresponding left or right lane line in the map as the perpendicular line.

[0076] 2) Error function construction: Steps S204 to S206 construct the error function and the objective function and perform iterative calculations to obtain the transformation matrix. Specifically, the error function E is constructed using all the above-mentioned lane line points p and their corresponding map lane line points q. The error function is the lateral position error between the two point sets. The error function , where n is the number of lane line points, R is the rotation matrix, t is the lateral translation variable, and p i is the coordinate of the i-th lane line point, q iis the coordinate of the map lane line point corresponding to the i-th captured lane line point, and the captured lane line point and the corresponding map lane line point are taken as a point pair, w i is the weight of the i-th point pair, w i =(1-p ix / MaxX) / SumW, p ix is the longitudinal distance of the i-th lane line point, MaxX is the maximum longitudinal distance among all point pairs, .

[0077] Objective function , using the gradient descent algorithm, the objective function is iteratively solved. When the error value of the error function is less than the threshold or reaches the number of iterations, the solution is considered complete and the optimal transformation matrix is obtained, that is, the required transformation matrix T. The initial GPS position is transformed using the transformation matrix T to obtain the vehicle's positioning position on the map.

[0078] Because lane markings are linear features with unobservable longitudinal displacement, the point set matching method is changed to lateral distance matching between the captured lane marking points and the mapped lane marking vectors during the solution process. The unobservable longitudinal direction is removed during the iterative solution and covariance solution. The lateral direction refers to the direction perpendicular to the vehicle's current direction of travel, while the longitudinal direction refers to the vehicle's current direction of travel.

[0079] Step S207 is the confidence assessment module: This module uses the desired positioning result and the positioning data error COV(Z) to obtain the positioning error COV(X) and determines the positioning confidence based on the positioning error COV(X). The positioning error can be converted to the corresponding confidence using existing confidence-error conversion methods.

[0080] Specifically:

[0081] 1) Obtaining Data Error COV(Z): Data error includes lane recognition error and lane-to-map matching error. When the matching and positioning module meets the target function iteration requirements or the error value of the error function during an iteration meets the threshold, the iteration is terminated. The error function value at the last iteration is assumed to include irreversible recognition error and algorithmic mathematical error. Therefore, the data error COV(Z) is set to the normalized error value of the error function at the last iteration.

[0082] 2) Since the objective function is expressed as least squares, the relationship between the data error COV(Z) and the positioning error COV(X) can be expressed using the Hessian matrix of the objective function J relative to the transformation matrix T, and the following positioning error solution equation can be obtained:

[0083] .

[0084] The specific solution is as follows:

[0085] Let the overall objective function J=Σ||G||^2, let F=||G||;

[0086] Translation vector and rotation matrix:

[0087] T = [x; y;];

[0088] R = [cos(a) -sin(a);

[0089] sin(a) cos(a)];

[0090] Transformation matrix T= ,in,

[0091] X is the longitudinal translation of the vehicle in map coordinates, Y is the lateral translation of the vehicle in map coordinates, and a is the change in the vehicle heading angle, which is obtained through two iterations.

[0092] Corresponding lane line points and map points: P i = [p ix ;p iy ]; Q i = [q ix ;q iy ]; p i and Q i Represents the coordinates of a point on the map.

[0093] The error between two points: ; Here R is the same as R in the matrix, wi is the weight of each point pair, w i =(1-p ix / MaxX) / SumW, p ix is the longitudinal distance of the i-th lane line point, MaxX is the maximum longitudinal distance among all point pairs, .

[0094] Let X = [y, a], J for X = [y, a] Hessian matrix is expressed as:

[0095] Here X represents the state quantity, which is a vector;

[0096] (1) Since the longitudinal displacement is not considered, we can only solve :

[0097] Since there are multiple pairs of points, , and the same applies to other cases.

[0098] Then for J i Solving for partial derivatives

[0099] , • is the dot product symbol.

[0100]

[0101] Furthermore, since the longitudinal displacement is not considered, x is 0, which can be expressed as:

[0102]

[0103] Expand the derivative and bring it in. Since the vertical x is not considered, the Jacobian matrix of y and a can be solved in the same way, for example:

[0104] Similarly, the Hessian matrix can be solved according to the Jacobian matrix:

[0105]

[0106]

[0107] but

[0108] In this formula, X represents the state quantity

[0109] Solution In this formula, X represents the state quantity, and Z represents the state quantity of data error.

[0110] Similarly, solve the Jacobian matrix of G for pi and qi , and then solve its second-order derivative

[0111]

[0112]

[0113] Finally, construct cov(Z)

[0114]

[0115] And calculate .

[0116] Finally, step S208 uses the revised position for multi-sensor fusion positioning, with the confidence level used as a weight for the fusion positioning. For example, the confidence level can be used directly as a weight for the revised position. Multi-sensor fusion positioning can utilize existing fusion positioning methods, with the revised position as one of the positioning sources.

[0117] The specific fusion method can be implemented using the existing position fusion method.

[0118] The proposed method uses lane line points to match map features to obtain accurate matching points. An error function is constructed using the lateral matching position error between the map point set and the camera lane line recognition results. By reducing this error function, the optimal matching result is achieved. Furthermore, the optimized residual matching error is used to calculate the confidence level of lane line recognition in real time. This confidence level is then used to adjust the weight of the camera lane line recognition results in subsequent fusion positioning, reducing the impact of random recognition errors on fusion positioning accuracy.

[0119] like Figure 3 The following is a flowchart of the matching and positioning module according to the preferred embodiment of the present invention, including:

[0120] Step S301, receiving lane lines recognized by the camera;

[0121] Step S302, initializing positioning using GPS positioning information;

[0122] Step S303, obtaining map vector data of the corresponding location;

[0123] Step S304: Match the lane line points with the map vectors one by one to obtain matching map points;

[0124] Step S305: construct an error function, where the error function is the lateral position error between the lane point set and the map point set;

[0125] Step S306, using a gradient descent algorithm, iteratively solves the transformation matrix from lane lines to the map;

[0126] Step S307: If the error is less than the threshold, execute step S308; otherwise, execute step S306 again.

[0127] Step S308: Obtain the required transformation matrix, and use the transformation matrix and the initial position of the GPS to obtain the position of the vehicle on the map;

[0128] Step S309: Calculate the position reliability.

[0129] like Figure 4 The following is a flowchart of the confidence assessment module according to the preferred embodiment of the present invention, including:

[0130] Step S401: The data error in lane line recognition and matching is expressed as COV(Z);

[0131] In step S402, the desired positioning error is expressed as COV(X). Since the objective function is least squares, the relationship between the data error COV(Z) and the positioning error COV(X) can be expressed using the Hessian matrix of the objective function J for the positioning result X:

[0132]

[0133] In step S403, the error function result and the data error are reflected in the positioning error COV(X), and the confidence level is determined based on the positioning error COV(X). When the positioning error is large, the confidence level is relatively small, and the fusion is based on a smaller weight, making the fused positioning result more accurate.

[0134] like Figure 5 As shown, a workflow diagram of lane matching and positioning according to the best embodiment of the present invention includes:

[0135] Step S501, obtaining lane lines recognized by the camera;

[0136] Step S502: Matching lane points with map vectors to obtain matching points, and constructing a lateral position error function and an objective function for the two point sets;

[0137] Step S503, iteratively solving the objective function to obtain a positioning result;

[0138] Step S504: Obtain the location reliability using the Hessian matrix of the objective function.

[0139] like Figure 6 The figure shows the lane line correction simulation test results using the lane line matching positioning of the present invention, including the map lane line 61, the identified lane line 62 after GPS disturbance, and the identified lane line 63 after correction.

[0140] Map lane lines 61 are projected onto the image; GPS perturbation-detected lane lines 62 are camera-detected lane lines, located on the map using GPS with added errors, and then projected back onto the image; and corrected lane lines 63 are camera-detected lane lines, located on the map using corrected GPS, and then projected back onto the image. It can be seen that after GPS errors, the corrected lane lines 63, obtained using the lane line matching and positioning method of the present invention, are consistent with map lane lines 61.

[0141] like Figure 7 The figure shows the GPS correction simulation test results using the lane matching positioning of the present invention, including: true value GPS 71, corrected GPS 72, and disturbed GPS 73.

[0142] The true GPS 71 represents the true position. The perturbed GPS 72 represents the GPS position with the error added. The corrected GPS 72 represents the position corrected after the camera recognizes the lane markings. It can be seen that after the GPS error, the corrected GPS 72, obtained using the lane matching positioning method of the present invention, overlaps with the true GPS 71.

[0143] In high-level autonomous driving, multiple sensors will be integrated for positioning, and the fusion weights of different sensors need to be determined. Therefore, positioning information from different sources is needed to provide its confidence.

[0144] The proposed method uses lane line points to match map features to obtain accurate matching points. An error function is constructed using the lateral matching position error between the map point set and the camera lane line recognition results. By reducing this error function, the optimal matching result is achieved. Furthermore, the optimized residual matching error is used to calculate the confidence level of lane line recognition in real time. This confidence level is then used to adjust the weight of the camera lane line recognition results in subsequent fusion positioning, reducing the impact of random recognition errors on fusion positioning accuracy.

[0145] Example 6

[0146] like Figure 8 FIG. 1 is a schematic diagram of the hardware structure of an electronic device of the present invention, comprising:

[0147] at least one processor 801; and,

[0148] A memory 802 in communication with at least one of the processors 801; wherein,

[0149] The memory 802 stores instructions that can be executed by at least one of the processors. The instructions are executed by at least one of the processors to enable the at least one processor to perform the lane matching and positioning method as described above.

[0150] Figure 8 A processor 801 is taken as an example.

[0151] The electronic device may further include an input device 803 and a display device 804 .

[0152] The processor 801, the memory 802, the input device 803 and the display device 804 may be connected via a bus or other means, with the bus connection being used as an example in the figure.

[0153] The memory 802 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs, and modules, such as program instructions / modules corresponding to the lane matching and positioning method in the embodiment of the present application, for example, Figure 1 The processor 801 executes the non-volatile software programs, instructions and modules stored in the memory 802 to perform various functional applications and data processing, that is, to implement the lane matching and positioning method in the above embodiment.

[0154] Memory 802 may include a program storage area and a data storage area. The program storage area may store an operating system and applications required for at least one function; the data storage area may store data generated by the lane matching and positioning method. Furthermore, memory 802 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state memory device. In some embodiments, memory 802 may optionally include a remote memory device relative to processor 801. Such remote memory device may be connected to the device executing the lane matching and positioning method via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0155] The input device 803 can receive user clicks and generate signal input related to user settings and function control of the lane line matching positioning method. The display device 804 can include a display device such as a display screen.

[0156] The one or more modules are stored in the memory 802 and, when executed by the one or more processors 801 , execute the lane matching and positioning method in any of the above method embodiments.

[0157] The proposed method uses lane line points to match map features to obtain accurate matching points. An error function is constructed using the lateral matching position error between the map point set and the camera lane line recognition results. By reducing this error function, the optimal matching result is achieved. Furthermore, the optimized residual matching error is used to calculate the confidence level of lane line recognition in real time. This confidence level is then used to adjust the weight of the camera lane line recognition results in subsequent fusion positioning, reducing the impact of random recognition errors on fusion positioning accuracy.

[0158] An embodiment of the present invention provides a storage medium storing computer instructions. When a computer executes the computer instructions, it is used to perform all steps of the lane matching and positioning method as described above.

[0159] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A lane line matching and positioning method, characterized in that: include: Obtaining a lane line identified by a vehicle camera device, and using the lane line point identified by the vehicle camera device as a captured lane line point; Get the map lane line of the vehicle's location in the map, and use the point on the map lane line as the map lane line point; Match the captured lane marking points with the map lane marking points; Constructing an error function for the lateral position error between the captured lane line point and the matched map lane line point using the transformation matrix as a variable, and constructing an objective function for the error function; Iteratively solve the objective function through the gradient descent algorithm; Correcting the vehicle positioning position based on the transformation matrix solved at the end of the iteration, obtaining a corrected positioning position based on the lane line recognized by the camera device, and obtaining a confidence level of the corrected positioning position; The transformation matrix includes a rotation matrix and a lateral translation variable. The error function of the lateral position error between the captured lane line point and the matched map lane line point is constructed with the transformation matrix as a variable, specifically including: Constructing the error function , where n is the number of lane line points, R is the rotation matrix, t is the lateral translation variable, and p i is the coordinate of the i-th lane line point, q i is the coordinate of the map lane line point corresponding to the i-th captured lane line point, and the captured lane line point and the corresponding map lane line point are taken as a point pair, w i is the weight of the i-th point pair, w i =(1-p ix / MaxX) / SumW, p ix is the longitudinal distance of the i-th lane line point, MaxX is the maximum longitudinal distance among all point pairs, .

2. The lane line matching and positioning method according to claim 1, characterized in that: The matching of the captured lane line points with the map lane line points specifically includes: For each captured lane line point, draw a perpendicular line from the captured lane line point to the map lane line, and use the point of the foot of the perpendicular on the map lane line as the map lane line point corresponding to the captured lane line point.

3. The lane line matching and positioning method according to claim 1, characterized in that: The constructing of the objective function about the error function specifically includes: Constructing the objective function , where E is the error function.

4. The lane line matching and positioning method according to claim 1, characterized in that: The method of correcting the vehicle positioning position based on the transformation matrix obtained at the end of the iteration to obtain a corrected positioning position based on the lane line recognized by the camera device specifically includes: A transformation matrix is constructed based on the rotation matrix R and the lateral translation variable t solved at the end of the iteration, and the vehicle positioning position is corrected according to the transformation matrix to obtain a corrected positioning position of the vehicle positioning position based on the lane line recognized by the camera device.

5. The lane line matching and positioning method according to claim 1, characterized in that: The confidence level of the corrected positioning position is obtained, specifically including: The confidence of the transformation matrix is calculated as the confidence of the corrected positioning position.

6. The lane line matching and positioning method according to claim 5, characterized in that: The calculating the confidence of the transformation matrix specifically includes: Obtain the error value of the error function when solving the objective function for the last iteration as the data error; According to the data error, a positioning error solution equation is constructed based on the Hessian matrix representation of the objective function relative to the transformation matrix; An equation is solved according to the positioning error to determine the positioning error, and a confidence level of the transformation matrix is determined according to the positioning error.

7. The lane line matching and positioning method according to claim 5, characterized in that: Also includes: The corrected positioning position is fused with other positioning sources, wherein the fusion weight of the corrected positioning position is calculated based on the confidence level.

8. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by at least one of the processors, and the instructions are executed by at least one of the processors to enable at least one of the processors to execute the lane matching and positioning method according to any one of claims 1 to 7.

9. A storage medium, characterized in that: The storage medium stores computer instructions, and when a computer executes the computer instructions, it is used to execute all steps of the lane line matching and positioning method according to any one of claims 1 to 7.

Images