Method for self-vehicle positioning using lane line matching

Abstract

The present invention relates to a method for self-vehicle positioning using lane line matching, which includes the steps of: obtaining a camera fitting curve; the camera fitting curve is in the camera coordinate system; extracting camera shape points; packing to obtain a set of camera shape points; obtaining map lane lines; packing to obtain a set of map shape points; performing secondary processing on the set of camera shape points; performing secondary processing on the set of map shape points to convert the map lane lines into map lane line fitting curves; performing equally spaced shape point sampling to obtain a set of sampled camera shape points and a set of sampled map shape points; calculating the mean square error to obtain the mean square error of the sampled camera shape points and the mean square error of the sampled map shape points; performing ICP matching to obtain R and T error parameters; obtaining the corrected vehicle position information. The present invention achieves the technical effects of low cost but high precision; it can handle almost all road conditions with extremely small errors.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving technology, and particularly to a method for positioning a vehicle by lane line matching. Background Art

[0002] In the field of autonomous driving, vehicle positioning technology is a key technology. Through positioning technology and combined with the advanced information provided by high-precision maps, autonomous driving vehicles can perceive the situation of the road they are traveling on and the traffic information on the road in advance in a complex and changeable road environment, and make behavior plans in advance. Therefore, autonomous driving technology poses relatively high requirements for high-precision positioning technology. The industry expects the positioning accuracy to be maintained at 20 cm. Currently, the price of mainstream positioning products to achieve this accuracy is relatively high, which is not conducive to the mass production of autonomous driving products;

[0003] In intelligent driving technology, a basic task and one of the difficulties in this field is to real-time calibrate the position of the vehicle on a surveyed high-precision map. This is because there will be errors whether through GPS positioning or visual recognition by in-vehicle cameras. If the errors are not corrected in time, a cumulative effect will occur, causing the autonomous driving vehicle to gradually deviate from the original route and then an accident will occur.

[0004] Regarding error elimination and precise positioning on a high-precision map, the existing technologies are divided into two categories. The following will use a typical technical solution as a representative for each of these two categories of technologies to illustrate their defects, so as to further illustrate the object of the present invention:

[0005] The first category of technologies is typically represented by a Chinese patent application with the name "Lane Line Matching Positioning Method, Electronic Device and Storage Medium" and the application number CN202111646702.3, which discloses the following technical features:

[0006] Obtain the lane lines recognized by the vehicle camera device, and use the points of the lane lines recognized by the vehicle camera device as the photographed lane line points; obtain the map lane lines at the vehicle positioning position in the map, and use the points on the map lane lines as the map lane line points; match the photographed lane line points with the map lane line points; construct an error function with the transformation matrix as a variable for the lateral position error between the photographed lane line points and the matched map lane line points, and construct an objective function for the error function; iteratively solve the objective function through the gradient descent algorithm; correct the vehicle positioning position based on the transformation matrix obtained at the end of the iteration to obtain a corrected positioning position for correcting the vehicle positioning position based on the lane lines recognized by the camera device, and obtain the confidence of the corrected positioning position.

[0007] The defects of the first category of existing technologies are as follows:

[0008] Since this method uses the lane line points captured by the camera, draws perpendicular lines from the captured lane line points to the map lane lines, and takes the foot points on the map lane lines as the corresponding map lane line points of the captured lane line points, constructs an error function with the transformation matrix as the variable and regarding the lateral position error between the captured lane line points and the matched map lane line points. However, when the longitudinal distance of the road varies greatly and there are curves, this point set corresponding method will have abnormal matching situations, resulting in inevitable positioning errors and unable to truly eliminate errors;

[0009] The second type of technology is typically represented by the Chinese patent application with the name "Positioning Pose Calibration Method and Vehicle Based on Vision and Map Lane Line Matching", application number CN202210188413.1, which discloses the following technical features:

[0010] Convert the lane line shape points of the high-precision map to the vehicle coordinate system to form a point set, select N points from the point set to form a source point set, and sample the visual lane lines obtained by the vehicle's front-view camera to form a corresponding point set of the source point set, that is, a target point set; use the distance between two points to represent the deviation and construct an objective function to solve the rotation and translation transformation matrix from the source point set to the target point set; set the tolerance value of the deviation; use the rotation and translation transformation matrix to map all points in the point set to a new point set in the coordinate system of the target point set; calculate the deviation of all points in the new point set to the nearest point on the corresponding visual lane line, and find the point pairs with the deviation within the tolerance value; judge whether the number of point pairs with the deviation within the tolerance value is greater than the set value. If so, use the point pairs with the deviation within the tolerance value to form a new source point set and a new target point set, and recalculate the rotation and translation transformation matrix from the new source point set to the new target point set. This rotation and translation matrix is the correction amount and stop the calculation; if not, continue the iterative calculation. When the number of iterations reaches the maximum number of iterations, stop the iteration and output the correction amount of the vehicle's position and heading;

[0011] The defects of the second type of existing technology are as follows:

[0012] Since both the lane lines and the curb curves obtained by the vehicle's front-view camera are represented as cubic curves in the vehicle coordinate system, and then the map shape points of the lane lines and the curb curves in the high-precision map are converted to the vehicle coordinate system, and the distance deviation between the corresponding points after interpolation of the two point sets is used as the error function. At the same time, since the cubic polynomial of the vehicle line fitting output by the camera can only be used when it is close to the vehicle, and the fitting of the real road is not ideal in many scenarios, the application scenario is very limited and the promotion value is extremely low.

[0013] Furthermore, whether it is the first type or the second type of existing technology, its cost is still high, and there is still a long way to go before real marketization applications. Summary of the Invention

[0014] In view of the above problems, the present invention provides a method for self-vehicle positioning using lane line matching, aiming to achieve the technical effects of low cost but high precision. Compared with the prior art, it has great promotion value; it can adapt to almost all road conditions with extremely small errors.

[0015] To solve the above problems, the technical solution provided by the present invention is as follows:

[0016] A method for self-vehicle positioning using lane line matching includes the following steps:

[0017] S100. Obtain a camera fitting curve through a front-view camera installed in front of the vehicle; the camera fitting curve is in the camera coordinate system; the camera coordinate system is: with the front of the vehicle as the x-axis, and the head pointing direction as the positive direction, and the camera coordinate system satisfies the right-hand rule;

[0018] S200. Extract camera shape points from the camera fitting curve; then pack all the extracted camera shape points to obtain a camera shape point set;

[0019] S300. Obtain the map lane lines where the current vehicle is located through a map module installed on the vehicle; the map lane lines are composed of map shape points; obtain the initial pose of the vehicle through a positioning module installed on the vehicle;

[0020] S400. Transform the map lane lines into the vehicle coordinate system; then extract map shape points from the map lane lines; then pack all the extracted map shape points to obtain a map shape point set;

[0021] S500. Perform secondary processing on the camera shape point set to improve the matching accuracy and ensure that the shape points can be mapped one-to-one;

[0022] S600. Perform secondary processing on the map shape point set; according to the processed map shape point set, convert the map lane lines into a map lane line fitting curve;

[0023] S700. At an artificially preset distance interval and within the range of an artificially preset distance value parameter, perform equidistant shape point sampling on the camera fitting curve and the map lane line fitting curve according to the equidistant sampling principle to obtain a camera shape point sampling set and a map shape point sampling set; the number of shape points in the camera shape point sampling set is the same as the number of shape points in the map shape point sampling set; the distance value parameter includes a camera distance value parameter and a map distance value parameter;

[0024] Calculate the mean square error between the shape points in the camera-shaped point sampling set and the shape points in the terrain-shaped point sampling set respectively, to obtain the mean square error of the camera-shaped point sampling and the mean square error of the terrain-shaped point sampling; then subtract the mean square error of the terrain-shaped point sampling from the mean square error of the camera-shaped point sampling to obtain the difference of the sampling mean square error; then, based on the difference of the sampling mean square error, perform the following operations:

[0025] If the difference of the sampling mean square error is not less than the manually preset difference threshold, return to and execute S100 again;

[0026] If the difference of the sampling mean square error is less than the difference threshold, execute S900;

[0027] S900. Perform ICP matching on the shape points in the camera-shaped point sampling set and the corresponding shape points in the terrain-shaped point sampling set to obtain the R error parameter and T error parameter of the corresponding shape points in the camera-shaped point sampling set and the terrain-shaped point sampling set;

[0028] S1000. Rotate the ego-vehicle position in the camera coordinate system and the initial pose using the R error parameter, and translate the ego-vehicle position in the camera coordinate system and the initial pose using the T error parameter to obtain the corrected vehicle position information; the corrected vehicle position information in the vehicle coordinate system is the final output result of this method.

[0029] Preferably, the camera fitting curve is written as F: X → Y, and is specifically expressed by the following formula:

[0030] Y = C0 + C1 * X + C2 * X 2 + C3 * X 3

[0031] where: X is the abscissa of the points on the camera fitting curve in the camera coordinate system; Y is the ordinate of the points on the camera fitting curve in the camera coordinate system; c3 is 6 times the curvature change rate at the intersection of the camera fitting curve and the Y-axis of the camera coordinate system; c2 is 2 times the curvature at the intersection of the camera fitting curve and the Y-axis of the camera coordinate system; c1 is the curvature at the intersection of the camera fitting curve and the Y-axis of the camera coordinate system; c0 is the intercept of the camera fitting curve and the Y-axis of the camera coordinate system.

[0032] Preferably, S200 specifically includes the following steps:

[0033] S210. Along the X-axis of the camera fitting curve, within the manually preset camera intercept range starting from the vehicle head, equally spaced divide the shape point parameters according to the manually preset camera equal spacing, and equally spaced extract the camera shape points;

[0034] S220. Then pack all the extracted camera-shaped points to obtain the set of camera-shaped points.

[0035] Preferably, in S300, the map lane lines where the current vehicle is located are obtained through the map module installed on the vehicle, and the specific steps are as follows:

[0036] S310. Obtain the initial pose through the positioning module;

[0037] S320. The map module extracts high-precision map data within a preset range around the vehicle according to the initial pose; the high-precision map data includes the map lane lines;

[0038] S330. Output the high-precision map data.

[0039] Preferably, in S400, the graphic points are extracted from the map lane lines, and the specific steps are as follows:

[0040] S410. Intercept a length of the specified distance of the side line on the map lane line that has been transformed into the vehicle coordinate system;

[0041] S420. Extract all the graphic points within the length of the specified distance of the side line on the map lane line; the graphic points are expressed by the following formula:

[0042] (A i , B i )

[0043] Where: A is the abscissa of the point on the map lane line in the vehicle coordinate system; B is the ordinate of the point on the map lane line in the vehicle coordinate system; i is the counter;

[0044] S430. Pack all the extracted graphic points to obtain the set of graphic points; then output the set of graphic points.

[0045] Preferably, in S500, a secondary process for improving the matching accuracy and ensuring one-to-one mapping of the shape points is performed on the set of camera-shaped points, and the specific steps are as follows:

[0046] S510. Construct a camera distance value parameter with the X value and Y value of the camera-shaped points in the set of camera-shaped points as the dependent variables; the camera distance value parameter is expressed by the following formula

[0047]

[0048] Where: S is the camera distance value parameter; N is the number of camera-shaped points in the set of camera-shaped points;

[0049] Construct a function of X value and Y value with the camera distance value parameter as the independent variable, expressed as follows:

[0050]

[0051] Then calculate two sets of camera form point parameters, expressed as follows:

[0052]

[0053] S530. According to the camera form point parameters, transform the camera fitting curve into a fifth-degree polynomial, expressed as follows:

[0054]

[0055] Where: The X and Y corresponding to S = 0 are the coordinates of the projection point where the origin of the camera coordinate system is vertically projected onto the lane line in the camera coordinate system;

[0056] S540. According to the camera form points in the camera form point set and the fifth-degree polynomial in S530, calculate two sets of camera equation systems, corresponding to F: S → X and F: S → Y respectively, expressed as follows:

[0057]

[0058]

[0059] S550. Solve for the two sets of camera form point parameters (C0, C1, C2, C3, C4, C5) and (C'0, C'1, C'2, C'3, C'4, C'5) in the camera equation systems in S540 to obtain the optimal solutions of the two sets of camera equation systems;

[0060] S560. Output the optimal solutions of the two sets of camera equation systems.

[0061] Preferably, S600 specifically includes the following steps:

[0062] S610. Construct a map distance value parameter with the A value and B value of the terrain form points in the terrain form point set as the dependent variables; the map distance value parameter is expressed as follows:

[0063]

[0064] Where: L is the map distance value parameter; N is the number of terrain form points in the terrain form point set;

[0065] S620. Construct a function of X value and Y value with the map distance value parameter as the independent variable, expressed as follows:

[0066]

[0067] Then two sets of camera-shaped point parameters are calculated and expressed by the following formula:

[0068]

[0069] S630. According to the topographic point parameters, the map lane line fitting curve is obtained, which is a fifth-degree polynomial and is expressed by the following formula:

[0070]

[0071] Where: The coordinates of points A and B corresponding to L = 0 are the coordinate values of the projection points of the vehicle's own coordinates in the vehicle coordinate system projected onto the lane boundary lines in the vehicle coordinate system;

[0072] S640. According to the topographic points in the topographic point set and the fifth-degree polynomial in S630, two sets of map equations are calculated, corresponding to F: L → A and F: L → B respectively, and are expressed by the following formula:

[0073]

[0074]

[0075] S650. Solve the map equations in S640 to obtain the parameter values (D0, D1, D2, D3, D4, D5) and (D'0, D'1, D'2, D'3, D'4, D'5) corresponding to F: L → A and F: L → B;

[0076] S660. Output (D0, D1, D2, D3, D4, D5) and (D'0, D'1, D'2, D'3, D'4, D'5); output the map lane line fitting curve.

[0077] Preferably, the value range of the distance interval in S700 is [0.2 m, 0.8 m]; the range of the distance value parameter is [0 m, 50 m]; the number of shape points in the camera-shaped point sampling set is the same as the number of shape points in the topographic point sampling set, both being 40.

[0078] Preferably, in S900, the shape points in the camera-shaped point sampling set are matched with the corresponding shape points in the topographic point sampling set by ICP, and are expressed by the following formula:

[0079]

[0080] Where: Where and is the corresponding shape point in the camera-shaped point sampling set and the terrain-shaped point sampling set.

[0081] Preferably, the corrected vehicle position information is expressed by the following formula:

[0082] L a = R * L b + T

[0083] where: L a is the corrected vehicle position information; L b is the initial pose of the vehicle.

[0084] Compared with the prior art, the present invention has the following advantages:

[0085] 1. Since the present invention completes high-precision positioning work based on low-cost rough-precision positioning devices, it achieves the technical effect of low cost but high precision. Compared with the prior art, it has great promotion value;

[0086] 2. Since the present invention realizes high-precision positioning of an autonomous driving vehicle through rough-precision positioning devices, a vehicle front-view camera capable of recognizing lane lines, and a high-precision map module, it can handle almost all road conditions with extremely small errors; verified by actual vehicles, in a specific scenario where the lane lines are clear, the positioning accuracy can be effectively guaranteed within 20 cm. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 is a schematic flowchart of a specific embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0088] The present invention will be further clarified below with reference to specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention. After reading the present invention, various equivalent modifications made by those skilled in the art fall within the scope defined by the appended claims of this application.

[0089] As Figure 1 shown, a method for self-vehicle positioning using lane line matching includes the following steps:

[0090] S100. Obtain a camera fitting curve through a front-view camera installed in front of the vehicle; the camera fitting curve is in the camera coordinate system; the camera coordinate system is: with the front of the vehicle as the x-axis, and the direction of the vehicle head as the positive direction, and the camera coordinate system satisfies the right-hand rule.

[0091] In this specific embodiment, the camera fitting curve is written as F: X → Y and is specifically expressed by Equation (1):

[0092] Y = C0 + C1 * X + C2 * X2 +C3*X 3 (1)

[0093] Where: X is the abscissa of the points on the camera fitting curve in the camera coordinate system; Y is the ordinate of the points on the camera fitting curve in the camera coordinate system; c3 is 6 times the curvature change rate at the intersection of the camera fitting curve and the Y-axis in the camera coordinate system; c2 is 2 times the curvature at the intersection of the camera fitting curve and the Y-axis in the camera coordinate system; c1 is the curvature at the intersection of the camera fitting curve and the Y-axis in the camera coordinate system; c0 is the intercept of the camera fitting curve and the Y-axis in the camera coordinate system.

[0094] It should be noted that the forward-looking camera installed in the current prior art can already stably output the camera fitting curve, so there is no need to convert the collected video data to obtain the camera fitting curve.

[0095] S200. Extract camera shape points from the camera fitting curve; then pack all the extracted camera shape points to obtain a set of camera shape points.

[0096] S200 specifically includes the following steps:

[0097] S210. Along the X-axis of the camera fitting curve, within the artificially preset camera intercept range starting from the vehicle head, equally spaced divide the shape point parameters according to the artificially preset camera, and equally spaced extract camera shape points.

[0098] S220. Then pack all the extracted camera shape points to obtain a set of camera shape points.

[0099] It should be noted that the forward-looking camera will have a distortion problem in fitting the lane lines at a long distance. Therefore, the X range of the forward-looking camera, that is, the camera intercept range, is 0-20 m; then the equally spaced division of the shape point parameters of the camera is selected as 0.1 m. Therefore, the number of camera shape points that can be extracted from a camera fitting curve of a forward-looking camera is 200; then the set of camera shape points is written as ∑(X i ,Y i ).

[0100] S300. Through the map module installed on the vehicle, obtain the map lane lines where the current vehicle is located; the map lane lines are composed of map shape points; through the positioning module installed on the vehicle, obtain the initial pose of the vehicle.

[0101] In this specific embodiment, obtaining the map lane lines where the current vehicle is located through the map module installed on the vehicle specifically includes the following steps:

[0102] S310. Through the positioning module, obtain the initial pose.

[0103] S320. The map module extracts high-precision map data within a preset range around the vehicle according to the initial pose; the high-precision map data includes map lane lines.

[0104] S330. Output the high-precision map data.

[0105] S400. Transform the map lane lines into the vehicle coordinate system; then extract the map feature points from the map lane lines; then pack all the extracted map feature points to obtain a set of map feature points;

[0106] In this specific embodiment, extracting the map feature points from the map lane lines specifically includes the following steps:

[0107] S410. On the map lane lines that have been transformed into the vehicle coordinate system, intercept the map feature points on the sidelines of the vehicle's lane at a preset specified distance length of the sidelines.

[0108] In this specific embodiment, the specified distance length of the sidelines is the map feature points on the sidelines of the map lane lines at a distance of 200 m in front of the vehicle and the map feature points on the sidelines of the map lane lines at a distance of 20 m behind the vehicle.

[0109] It should be noted that the reason for taking this distance length is to ensure that there is enough data support for subsequent feature point fitting and to ensure the accuracy of feature point fitting.

[0110] S420. Extract all the map feature points within the specified distance length of the sidelines of the map lane lines; express the map feature points by the following formula:

[0111] (A i ,B i )

[0112] where: A is the abscissa of the point on the map lane line in the vehicle coordinate system; B is the ordinate of the point on the map lane line in the vehicle coordinate system; i is the counter.

[0113] S430. Pack all the extracted map feature points to obtain a set of map feature points; then output the set of map feature points.

[0114] Before elaborating on S500 - S600, it should be noted in advance that the reason for performing the steps of S500 - S600 is that camera - shaped points sampled from the camera - fitting curve on the same side of the lane output by the camera and map - lane - shaped points are obtained. Since the segmentation of high - precision map - shaped points is not fixed, the number of map - shaped points extracted from the map is also not fixed. To ensure the accuracy of subsequent shape - point matching and reduce the error of shape - point matching, the two sets of already - extracted shape - point sets, namely the map - shaped point set and the camera - shaped point set, need to be processed to ensure that the shape points can be mapped one - to - one point - to - point. Therefore, the method of the present invention performs secondary processing on the map - shaped point set and the camera - shaped point set respectively.

[0115] S500. Perform secondary processing on the camera - shaped point set to improve the matching accuracy and ensure that the shape points can be mapped one - to - one.

[0116] In this specific embodiment, performing secondary processing on the camera - shaped point set to improve the matching accuracy and ensure that the shape points can be mapped one - to - one specifically includes the following steps:

[0117] S510. Taking the X - value and Y - value of the camera - shaped points in the camera - shaped point set as dependent variables, construct a camera - distance value parameter; the camera - distance value parameter is expressed by Equation (2)

[0118]

[0119] Where: S is the camera - distance value parameter; N is the number of camera - shaped points in the camera - shaped point set.

[0120] S520. Taking the camera - distance value parameter as an independent variable to construct a function about the X - value and Y - value, expressed by Equation (3):

[0121]

[0122] Then calculate to obtain two sets of camera - shaped point parameters, expressed by Equation (4):

[0123]

[0124] S530. According to the camera - shaped point parameters, transform the camera - fitting curve into a fifth - degree polynomial, expressed by Equation (5):

[0125]

[0126] Where: The X and Y corresponding to S = 0 are the coordinates of the projection point where the origin of the camera coordinate system is vertically projected onto the lane line in the camera coordinate system.

[0127] It should be noted that the reason for taking a fifth - degree polynomial is to avoid distortion as much as possible when fitting the curve.

[0128] According to the camera-shaped points in the camera-shaped point set, and the fifth-degree polynomial in S530, plus the shaped point data obtained by previous equidistant sampling, two sets of camera equations are calculated, corresponding to F: S → X and F: S → Y respectively, and expressed by equations (6) and (7):

[0129]

[0130] This set of equations corresponds to F: S → X.

[0131]

[0132] This set of equations corresponds to F: S → Y.

[0133] S550. Solve for the two sets of camera-shaped point parameters (C0, C1, C2, C3, C4, C5) and (C'0, C'1, C'2, C'3, C'4, C'5) in the camera equations in S540 to obtain the optimal solutions of the two sets of camera equations.

[0134] In this specific embodiment, the least squares method is used to solve respectively to obtain the optimal solutions of the two sets of parametric equations.

[0135] S560. Output the optimal solutions of the two sets of camera equations.

[0136] S600. Perform secondary processing on the ground-shaped point set; according to the processed ground-shaped point set, convert the map lane lines into map lane line fitting curves.

[0137] S600 specifically includes the following steps:

[0138] S610. Construct a map distance value parameter with the A value and B value of the ground-shaped points in the ground-shaped point set as the dependent variables; the map distance value parameter is expressed by equation (8):

[0139]

[0140] Where: L is the map distance value parameter; N is the number of ground-shaped points in the ground-shaped point set.

[0141] S620. Construct functions regarding the X value and Y value with the map distance value parameter as the independent variable, and express them by equation (9):

[0142]

[0143] Then calculate two sets of camera-shaped point parameters, and express them by equation (10):

[0144]

[0145] Based on the map graphic point parameters, obtain the map lane line fitting curve, which is a fifth-degree polynomial and is expressed by Equation (11):

[0146]

[0147] Where: The A and B corresponding to L = 0 are the coordinate values of the projection points of the vehicle's own coordinates in the vehicle coordinate system projected onto the lane boundary lines in the vehicle coordinate system.

[0148] It should be noted that the reason for taking a fifth-degree polynomial is to avoid distortion as much as possible during curve fitting.

[0149] S640. Based on the map graphic points in the map graphic point set and the fifth-degree polynomial in S630, calculate two sets of map equation systems, corresponding to F: L → A and F: L → B respectively, and are expressed by Equations (12) and (13):

[0150]

[0151] This set of equations corresponds to F: L → A.

[0152]

[0153] This set of equations corresponds to F: L → B.

[0154] S650. Solve the map equation systems in S640 to obtain the corresponding parameter values (D0, D1, D2, D3, D4, D5) and (D'0, D'1, D'2, D'3, D'4, D'5) of F: L → A and F: L → B.

[0155] S660. Output (D0, D1, D2, D3, D4, D5) and (D'0, D'1, D'2, D'3, D'4, D'5); output the map lane line fitting curve.

[0156] S700. At the artificially preset distance intervals and within the range of the artificially preset distance value parameters, according to the equal-spacing sampling principle, perform equal-spacing graphic point sampling on the camera fitting curve and the map lane line fitting curve respectively to obtain the camera graphic point sampling set and the map graphic point sampling set; the number of graphic points in the camera graphic point sampling set is the same as the number of graphic points in the map graphic point sampling set; the distance value parameters include the camera distance value parameter and the map distance value parameter.

[0157] In this specific embodiment, when obtaining the parameter equations of S-A, S-B, S-X, and S-Y respectively, in order to ensure the accuracy of graphic point matching, equal-spacing is performed on these four groups of parameter equations, and the value range of the distance interval is [0.2 m, 0.8 m]; the range of the distance value parameter is [0 m, 50 m]; the camera graphic point sampling set is written as ∑(X'i , Y' i ), the number of shape points in is the same as that in the ground map shape point sampling set, written as ∑(A' i , B' i ), and the number of shape points in is the same, both 40.

[0158] S800. Calculate the mean square error of the shape points in the camera shape point sampling set and the shape points in the ground map shape point sampling set respectively to obtain the camera shape point sampling mean square error and the ground map shape point sampling mean square error; then subtract the ground map shape point sampling mean square error from the camera shape point sampling mean square error to obtain the difference of the sampling mean square error; then perform the following operations according to the difference of the sampling mean square error:

[0159] If the difference of the sampling mean square error is not less than the difference threshold preset manually, go back to and execute S100 again.

[0160] If the difference of the sampling mean square error is less than the difference threshold, execute S900.

[0161] It should be noted that the principle of the steps of S800 is as follows:

[0162] Before performing shape point matching, it is necessary to calculate the error parameters for the shape point data of the camera fitting curve and the shape point data of the map lane line fitting curve that have been extracted; by calculating the mean square error of the shape point data of the two groups of samplings respectively, and then comparing the difference of the mean square errors of the two groups of shape points, if the difference of the mean square errors of the two groups is within the specified threshold range, it can be determined that the two groups of shape point parameters have performed normal curve fitting, and the two groups of shape point data extracted can perform normal shape point matching. The purpose of solving the error parameters is to ensure the consistency of the curve fitting of the camera lane boundary and the high-precision map lane boundary, prevent the occurrence of overfitting in some parts, and the difference between the fitting curve and the actual road lane boundary is large, resulting in calculation errors.

[0163] S900. Perform ICP matching on the shape points in the camera shape point sampling set and the corresponding shape points in the ground map shape point sampling set to obtain the R error parameter and T error parameter of the shape points in the camera shape point sampling set and the corresponding shape points in the ground map shape point sampling set.

[0164] In this specific embodiment, performing ICP matching on the shape points in the camera shape point sampling set and the corresponding shape points in the ground map shape point sampling set is expressed by Equation (14):

[0165]

[0166] Where: Where and are the corresponding shape points in the camera shape point sampling set and the ground map shape point sampling set.

[0167] It should be noted that the principle of the steps of S900 is as follows:

[0168] Through the previous curve function conversion of the camera fitting curve and the map lane line fitting curve and the fitting of the quintic polynomial curve, and then performing form point sampling on the camera lane boundary line and the map lane boundary line at the same S distance range and the same spacing, this can ensure that the form point data sampled in this way can have a one-to-one mapping relationship, reducing the increase in matching error caused by the inconsistency of form point sampling between the camera lane boundary line and the map lane boundary line. Then, perform form point ICP matching on the equally spaced sampled form points of the lane boundary lines obtained by the two groups of sensors extracted previously. Through the ICP matching algorithm of the form points, the R (rotation) error parameter and the T (translation) error parameter between the two groups of form points can be calculated. This set of parameter errors is the error between the camera fitting curve obtained by the camera sensor in the camera coordinate system and the map lane line obtained by the high-precision map sensor in the vehicle coordinate system converted by the positioning module.

[0169] S1000. Rotate the ego-vehicle position and initial pose in the camera coordinate system using the R error parameter, and translate the ego-vehicle position and initial pose in the camera coordinate system using the T error parameter to obtain the corrected vehicle position information; the corrected vehicle position information in the vehicle coordinate system is the final output result of this method.

[0170] In this specific embodiment, the corrected vehicle position information is expressed by Equation (15):

[0171] L a =R*L b +T (15)

[0172] Where: L a is the corrected vehicle position information; L b is the initial pose of the vehicle.

[0173] It should be noted that the principle of the steps of S1000 is as follows:

[0174] Through the previous minimum error matching method, the error between the camera fitting curve obtained by the front view camera in the camera coordinate system and the map lane line obtained by the map module in the vehicle coordinate system converted by the positioning module can be calculated. The rotation and translation of the ego-vehicle position and attitude L b can be performed, and then the position information of the ego-vehicle position of the map lane line converted to the vehicle coordinate system of the high-precision map sensor can be obtained.

[0175] It has been proven by experiments that the present invention can effectively ensure the positioning accuracy within 20 cm on the premise of low cost.

[0176] In the foregoing detailed description, various features are combined in a single embodiment to simplify the present disclosure. This method of disclosure should not be interpreted as reflecting an intention that the embodiments of the claimed subject matter require more features than are clearly recited in each claim. On the contrary, as reflected in the appended claims, the invention lies in less than the full scope of features of the single disclosed embodiment. Accordingly, the appended claims are hereby expressly incorporated into the detailed description, where each claim stands on its own as a separate preferred embodiment of the invention.

[0177] The above-described disclosed embodiments are described to enable any person skilled in the art to make or use the present invention. For those skilled in the art, various modifications to these embodiments are obvious, and the general principles defined herein can be applied to other embodiments without departing from the spirit and scope of the present disclosure. Therefore, the present disclosure is not limited to the embodiments given herein, but is consistent with the broadest scope of the principles and novel features disclosed in this application.

[0178] The foregoing description includes examples of one or more embodiments. Of course, it is not possible to describe all possible combinations of components or methods for the purpose of describing the above embodiments, but those of ordinary skill in the art should recognize that the various embodiments can be further combined and arranged. Therefore, the embodiments described herein are intended to cover all such changes, modifications, and variations that fall within the scope of the appended claims. In addition, with respect to the term "comprising" used in the specification or claims, the manner in which this term is encompassed is similar to the term "including," as interpreted when "including" is used as a transitional word in a claim. Further, any use of the term "or" in the claims or specification is to mean "non-exclusive or."

[0179] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is only for the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for self-vehicle positioning using lane line matching, characterized in that: It includes the following steps: S100. Obtain a camera fitting curve through a front-view camera installed in front of the vehicle; the camera fitting curve is in the camera coordinate system; the camera coordinate system is: with the front of the vehicle as the x-axis, and the direction of the vehicle head as the positive direction, and the camera coordinate system satisfies the right-hand rule; S200. Extract camera shape points from the camera fitting curve; then pack all the extracted camera shape points to obtain a camera shape point set; S300. Obtain the map lane lines where the current vehicle is located through a map module installed on the vehicle; the map lane lines are composed of map shape points; obtain the initial pose of the vehicle through a positioning module installed on the vehicle; S400. Transform the map lane lines into the vehicle coordinate system; Then extract map shape points from the map lane lines; then pack all the extracted map shape points to obtain a map shape point set; S500. Perform secondary processing on the camera shape point set to improve the matching accuracy and ensure that the shape points can be mapped one-to-one; S600. Perform secondary processing on the map shape point set; according to the processed map shape point set, convert the map lane lines into a map lane line fitting curve; S700. At an artificially preset distance interval and within the range of an artificially preset distance value parameter, perform equidistant shape point sampling on the camera fitting curve and the map lane line fitting curve according to the equidistant sampling principle to obtain a camera shape point sampling set and a map shape point sampling set; the number of shape points in the camera shape point sampling set is the same as the number of shape points in the map shape point sampling set; the distance value parameter includes a camera distance value parameter and a map distance value parameter; S800. Calculate the mean square error of the shape points in the camera shape point sampling set and the shape points in the map shape point sampling set respectively to obtain the camera shape point sampling mean square error and the map shape point sampling mean square error; then subtract the map shape point sampling mean square error from the camera shape point sampling mean square error to obtain the difference of the sampling mean square error; then perform the following operations according to the difference of the sampling mean square error: If the difference of the sampling mean square error is not less than an artificially preset difference threshold, return to and execute S100 again; If the difference of the sampling mean square error is less than the difference threshold, execute S900; S900. Perform ICP matching on the shape points in the camera shape point sampling set and the corresponding shape points in the map shape point sampling set to obtain the R error parameter and the T error parameter of the shape points in the camera shape point sampling set and the corresponding shape points in the map shape point sampling set; S1000. Rotate the position of the vehicle itself in the camera coordinate system and the initial pose using the R error parameter, and translate the position of the vehicle itself in the camera coordinate system and the initial pose using the T error parameter to obtain the corrected vehicle position information; The corrected vehicle position information is in the vehicle coordinate system and is the final output result.

2. The method for self-vehicle positioning using lane line matching according to claim 1, characterized in that: The camera fitting curve is written as F: X → Y, and is specifically expressed as follows: Y = C0 + C1*X + C2*X 2 + C3*X 3 Where: X is the abscissa of the points on the camera fitting curve in the camera coordinate system; Y is the ordinate of the points on the camera fitting curve in the camera coordinate system; c3 is 6 times the curvature change rate at the intersection of the camera fitting curve and the Y-axis in the camera coordinate system; c2 is 2 times the curvature at the intersection of the camera fitting curve and the Y-axis in the camera coordinate system; c1 is the curvature at the intersection of the camera fitting curve and the Y-axis in the camera coordinate system; c0 is the intercept of the camera fitting curve and the Y-axis in the camera coordinate system.

3. The method for vehicle self-positioning using lane line matching according to claim 2, wherein: S200 specifically includes the following steps: S210. Along the X-axis of the camera fitting curve, within the artificially preset camera intercept range starting from the vehicle head, divide the form point parameters at equal intervals according to the artificially preset camera, and extract the camera form points at equal intervals. S220. Then pack all the extracted camera form points to obtain the camera form point set.

4. The method for positioning the host vehicle by using lane line matching according to claim 3, wherein: In S300, the map lane line where the current vehicle is located is obtained through the map module installed on the vehicle, which specifically includes the following steps: S310. Obtain the initial pose through the positioning module. S320. The map module extracts high-precision map data within the artificially set vehicle surrounding set range according to the initial pose; the high-precision map data includes the map lane line. S330. Output the high-precision map data.

5. The method for self-vehicle positioning using lane line matching according to claim 4, wherein: In S400, the map form points are extracted from the map lane line, which specifically includes the following steps: S410. On the map lane line that has been transformed into the vehicle coordinate system, intercept the artificially preset specified distance length of the side line. S420. Extract all the map form points within the specified distance length of the side line on the map lane line. The map form point is expressed by the following formula: (A i ,B i ) Where: A is the abscissa of the points on the map lane line in the vehicle coordinate system; B is the ordinate of the points on the map lane line in the vehicle coordinate system. i is a counter. S430. Pack all the extracted map form points to obtain the map form point set; then output the map form point set.

6. The method for self-vehicle positioning using lane line matching according to claim 5, wherein: In S500, a secondary process for improving the matching accuracy and ensuring one-to-one mapping of the form points is performed on the camera form point set, which specifically includes the following steps: S510. Using the X value and Y value of the camera form points in the camera form point set as the dependent variables, construct a camera distance value parameter; the camera distance value parameter is expressed by the following formula Where: S is the camera distance value parameter; N is the number of camera form points in the camera form point set. S520. Construct a function about the X value and Y value with the camera distance value parameter as the independent variable, which is expressed by the following formula: Then calculate and obtain two sets of camera form point parameters, which are expressed by the following formula: S530. According to the camera form point parameters, transform the camera fitting curve into a fifth-degree polynomial, which is expressed by the following formula: Where: The X and Y corresponding to S = 0 are the coordinates of the projection point of the origin of the camera coordinate system vertically projected onto the lane side line in the camera coordinate system. Based on the camera-shaped points in the set of camera-shaped points and the fifth-degree polynomial in S530, two sets of camera equations are calculated, corresponding to F: S → X and F: S → Y respectively, and are expressed as follows: Solve for the two sets of camera-shaped point parameters (C0, C1, C2, C3, C4, C5) and (C'0, C'1, C'2, C'3, C'4, C'5) in the camera equations in S540 to obtain the optimal solutions of the two sets of camera equations; Output the optimal solutions of the two sets of camera equations.

7. The method for self-vehicle positioning using lane line matching according to claim 6, characterized in that: S600 specifically includes the following steps: Construct a map distance value parameter with the A value and B value of the terrain-shaped points in the set of terrain-shaped points as dependent variables; the map distance value parameter is expressed as follows: Where: L is the map distance value parameter; N is the number of terrain-shaped points in the set of terrain-shaped points; Construct a function regarding the X value and Y value with the map distance value parameter as the independent variable, and it is expressed as follows: Then two sets of camera-shaped point parameters are calculated, and are expressed as follows: Based on the terrain-shaped point parameters, obtain the map lane line fitting curve, which is a fifth-degree polynomial and is expressed as follows: Where: The A and B corresponding to L = 0 are the coordinate values of the projection points of the vehicle's own coordinates in the vehicle coordinate system projected onto the lane boundary line in the vehicle coordinate system; Based on the terrain-shaped points in the set of terrain-shaped points and the fifth-degree polynomial in S630, calculate two sets of map equations, corresponding to F: L → A and F: L → B respectively, and are expressed as follows: Solve the map equations in S640 to obtain the parameter values (D0, D1, D2, D3, D4, D5) and (D'0, D'1, D'2, D'3, D'4, D'5) corresponding to F: L → A and F: L → B; Output (D0, D1, D2, D3, D4, D5) and (D'0, D'1, D'2, D'3, D'4, D'5); output the map lane line fitting curve.

8. The method for vehicle self-localization using lane line matching according to claim 7, characterized in that: The value range of the distance interval in S700 is [0.2m, 0.8m]; the range of the distance value parameter is [0m, 50m]; the number of shape points in the set of sampled camera-shaped points is the same as the number of shape points in the set of sampled terrain-shaped points, both are 40.

9. The method for self-vehicle positioning using lane line matching according to claim 8, characterized in that: Perform ICP matching on the shape points in the set of sampled camera-shaped points and the corresponding shape points in the set of sampled terrain-shaped points, and it is expressed as follows: Wherein: wherein and are corresponding shape points in the camera-shaped point sampling set and the terrain-shaped point sampling set.

10. The method for self-vehicle positioning using lane line matching according to claim 1, characterized in that: The corrected vehicle position information is expressed as follows: L a = R * L b + T Where: L a is the corrected vehicle position information; L b is the initial pose of the vehicle.

Images