Method and device for generating lane markings

By approximating lane markings with virtual circles and correction points, the method generates smooth lane markings with lower computational load, addressing the instability issues of polynomial-based methods and ensuring stable vehicle behavior.

JP7751790B2Active Publication Date: 2025-10-09MAZDA MOTOR CORP
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
JP2021192003
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2021-11-26
Publication Date
2025-10-09
Estimated Expiration
2041-11-26

AI Technical Summary

Technical Problem

Existing methods for generating lane markings using polynomial approximation in autonomous vehicles face high computational loads and require complex pre- and post-processing, especially when dealing with implicit functions or straight lines, leading to unstable vehicle behavior.

Method used

A method and device that generate smoothed lane markings by identifying at least three points within a window, calculating a geometric centroid, approximating these points with a virtual circle, and setting correction points at the intersection of this circle and a line connecting the centroid, thereby avoiding complex polynomial approximations and reducing computational load.

Benefits of technology

This approach allows for the generation of smooth lane markings with reduced computational effort, ensuring stable vehicle behavior by aligning lane markings without the need for high-load processes like polynomial approximation and coordinate transformation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007751790000001
    Figure 0007751790000001
  • Figure 0007751790000002
    Figure 0007751790000002
  • Figure 0007751790000003
    Figure 0007751790000003
Patent Text Reader

Abstract

To generate a compartment line which is smoothed from point group data on the compartment line without employing a complicated polynominal approximation.SOLUTION: A compartment line generation method comprises: specifying at least three or more data points Pm located within a window Wk on the basis of, map information including a plurality of data points WPn representing a compartment line of a lane; calculating a geometric point Gk of gravity of the plurality of data points Pm; calculating a virtual circle VCk approximating the plurality of data points Pm; setting as a correction data point Qk an intersection shorter in a distance to the point Gk of gravity between two intersections of a straight line Lk connecting the center point Ck of the virtual circle VCk and the point Gk of gravity and the virtual circle VCk; moving the window Wk along the lane and repeating said processes to generate a point group of a plurality of correction data points Qk; and setting a point group of a plurality of correction data points Qk as a compartment line.SELECTED DRAWING: Figure 3
Need to check novelty before this filing date? Find Prior Art

Description

[Technical Field]

[0001] The present invention relates to a lane marking generation method and device, and more particularly to a lane marking generation method and device for generating lane marks for vehicle travel lanes. [Background technology]

[0002] Conventionally, technologies have been proposed for calculating a vehicle's target driving route based on lane marking information. The lane marking information included in map information is the position information or coordinate values ​​(latitude, longitude, altitude) of each point in point cloud data (a set of discrete points) that constitutes the lane markings. Alternatively, the lane marking information is the position information or coordinate values ​​of each point in point cloud data that constitutes the lane markings calculated from images captured by a vehicle camera. For example, the point cloud (waypoints) of lane marking information provided by a high-precision map may contain missing points, may not be arranged at regular intervals, or may have variations in positional accuracy in the longitudinal and lateral directions.

[0003] For this reason, if point cloud data of lane marking information is used to generate a target driving route for an autonomous vehicle without smoothing, the behavior of the vehicle during autonomous driving will become unstable. Therefore, it is necessary to smooth the point cloud data to reduce variations in data accuracy. For example, Patent Document 1 describes a method of calculating an approximate curve of lane marking lines by polynomial approximation of point cloud data of lane marking information using the least squares method or the like. A target driving route is generated based on this approximate curve. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 6884173 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the above-mentioned method using polynomial approximation has the problem that the complex polynomial approximation itself requires a large computational load. Furthermore, when using a straight line where x=k (a constant value) in the xy coordinate system or an implicit function (a function that has multiple y values ​​for the same x value in the xy coordinate system), it is not possible to specify the y value for the x value, so pre-processing and post-processing such as coordinate transformation processing are required. This further increases the computational load.

[0006] The present invention has been made to solve such problems, and aims to provide a method and device for generating demarcation lines that can generate smoothed demarcation lines from point cloud data of demarcation lines without using complex polynomial approximations. [Means for solving the problem]

[0007] In order to achieve the above object, the present invention provides: Used by the vehicle's computer to calculate the target driving path Lane markings Data point cloud A vehicle lane marking generation method for generating a lane marking, comprising: The vehicle is equipped with an on-board camera that provides image data of the surroundings of the vehicle or a navigation system that provides map information, and the computer executes a navigation system based on the image data or map information. Multiple data points indicating the position information of multiple discrete points on the lane markings are Get At least three points located within a window that specifies a predetermined distance range of the lot line Multiple a data point identification step of identifying data points; The computer Identified in the window At least 3 points or more a centroid generating step of calculating a geometric centroid of a plurality of data points; The computer Identified in the window At least 3 points or more a virtual circle generation step of calculating a virtual circle that approximates a plurality of data points; The computer a correction data point generating step of setting, as a correction data point, one of two intersection points between the virtual circle and a line connecting the center point of the virtual circle and the center of gravity, which is closer to the center of gravity; The computer The window is moved along the lane, and the data point specifying step, the center of gravity point generating step, the virtual circle generating step, and the corrected data point generating step are repeated to generate a point cloud of a plurality of corrected data points, and the point cloud of the plurality of corrected data points is used to align the lane markings. Data point cloud and a division line generating step of setting the division line to the above.

[0008] According to the present invention configured in this manner, the lane marking data (multiple data points) from the map information is divided into sections using windows, approximated by virtual circles, and corrected data points for each window are calculated. Then, by moving the window along the lane, the present invention can generate corrected lane markings consisting of multiple corrected data points over a predetermined range of the lane. Therefore, even if the lane marking data points in the map information are missing or vary, the present invention can generate smooth lane marking data with a low computational load while avoiding processes that require high computational loads, such as complex polynomial approximation and coordinate transformation processing.

[0009] In the present invention, preferably, in the virtual circle generating step, The computer The virtual circle is calculated by the least squares method. According to the present invention configured in this way, the virtual circle can be calculated with a low calculation load using a quadratic polynomial that represents the circle.

[0010] In the present invention, preferably, The navigation system The method further includes a step of receiving map information from a map information source. According to the present invention configured in this manner, since map information is received from an external map information source, calculations for newly generating map information from image data captured by a camera or the like within the vehicle are not required.

[0011] In the present invention, the lane preferably includes the lane on which the vehicle is traveling and lanes adjacent to or merging with the lane on which the vehicle is traveling. According to the present invention configured in this manner, it is possible to easily generate dividing lines for lanes other than the lane on which the vehicle is traveling.

[0012] In order to achieve the above object, the present invention provides: Used by the vehicle's computer to calculate the target driving path Lane markings Data point cloud A vehicle lane marking generation device for generating a lane marking, The lane marking generation device includes a computer and an on-board camera that provides image data of the area around the vehicle or a navigation system that provides map information, Lane marking generation device Computer teeth, Based on the image data or the map information Multiple data points indicating the position information of multiple discrete points on the lane markings are collected. Get At least three points located within a window that specifies a predetermined distance range of the lot line Multiplea data point identification step for identifying data points within the window; At least 3 points or more A centroid generation process for calculating a geometric centroid of a plurality of data points, and a process for calculating a centroid of a specified area within the window. At least 3 points or more a virtual circle generating step of calculating a virtual circle that approximates a plurality of data points; a modified data point generating step of setting, as a modified data point, one of two intersections between the virtual circle and a line connecting the center point of the virtual circle and the center of gravity, which is closer to the center of gravity; a window moving along the lane, repeating the data point specifying step, center of gravity generating step, virtual circle generating step, and modified data point generating step to generate a point cloud of the plurality of modified data points, and aligning the point cloud of the plurality of modified data points with the lane markings Data point cloud and a division line generating step of setting the division line to the above.

[0013] In the present invention, preferably, a lane marking generating device Computer In the virtual circle generating step, the virtual circle is calculated by the least squares method. In the present invention, preferably, a lane marking generating device Navigation system is further configured to perform the step of receiving map information from a map information source. In the present invention, the lane preferably includes a lane on which a vehicle is traveling, and a lane adjacent to or merging with the lane on which the vehicle is traveling. [Effects of the Invention]

[0014] According to the present invention, smoothed lane lines can be generated from point cloud data of lane lines without using complex polynomial approximation. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a configuration diagram of a vehicle control system according to an embodiment of the present invention. [Figure 2] FIG. 10 is an explanatory diagram showing the flow of a demarcation line generation process according to an embodiment of the present invention. [Figure 3] FIG. 10 is an explanatory diagram of a demarcation line generation process according to an embodiment of the present invention. [Figure 4]10 is a flowchart of a demarcation line generation process according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0016] A vehicle control system according to an embodiment of the present invention will be described below with reference to the accompanying drawings. First, the configuration of the vehicle control system will be described with reference to Fig. 1. Fig. 1 is a configuration diagram of the vehicle control system.

[0017] The vehicle control system 100 of this embodiment is configured to calculate and generate roadway markings when the vehicle 1 (see FIG. 3) travels along a roadway (or lane). The markings are, for example, white lines separating two adjacent lanes, or lane edge lines separating lanes from areas other than the lanes (such as sidewalks).

[0018] In this embodiment, after generating lane markings on both sides of the lane, a target driving path can be further calculated and generated in the center of the lane markings on both sides. When performing autonomous driving, the vehicle 1 is controlled to drive on this target driving path. Furthermore, when a preceding vehicle is driving ahead of the vehicle 1, the vehicle control system 100 can determine the lateral movement (such as a lane change) of the preceding vehicle based on the generated lane markings.

[0019] In this embodiment, the lane markings to be generated are not limited to those of the lane on which vehicle 1 is traveling. For example, lane markings of adjacent lanes (including oncoming lanes) adjacent to the lane on which vehicle 1 is traveling may also be generated. Furthermore, at an intersection on a highway, lane markings of a merging road that merges with the main road on which vehicle 1 is traveling, or lane markings of a main road that merges with the merging road when vehicle 1 is traveling on the merging road may also be generated.

[0020] 1, a vehicle control system 100 is mounted on a vehicle 1 and includes a vehicle control unit (ECU) 10, which is a lane marking generation device, multiple sensors, and multiple control systems. The multiple sensors include an on-board camera 21, a millimeter-wave radar 22, a wheel speed sensor 23, an acceleration sensor 24, a gyro sensor 25, a steering angle sensor 26, an accelerator sensor 27, a brake sensor 28, a positioning sensor 29, and a navigation system 30. The multiple control systems also include a steering control system 31, an engine control system 32, and a brake control system 33.

[0021] The ECU 10 is configured as a computer equipped with a CPU, a memory for storing various programs, an input / output device, etc. The ECU 10 is configured to be able to output request signals to a steering control system 31, an engine control system 32, and a brake control system 33 based on signals received from a plurality of sensors, for appropriately operating the steering system, engine system, and brake system, respectively.

[0022] The vehicle-mounted camera 21 captures an image of the surroundings of the vehicle 1 and outputs image data of the image. The ECU 10 identifies dividing lines (lane boundaries, white lines, yellow lines, etc.) based on the image data.

[0023] The millimeter-wave radar 22 is a measuring device that measures the position and speed of an object (particularly, a preceding vehicle, a parked vehicle, a pedestrian, an obstacle, etc.), and transmits radio waves (transmission waves) ahead of the vehicle 1 and receives reflected waves generated when the transmission waves are reflected by the object. Then, the millimeter-wave radar 22 measures the distance between the vehicle 1 and the object (for example, the inter-vehicle distance) and the relative speed of the object with respect to the vehicle 1 based on the transmission waves and the received waves. Note that in this embodiment, instead of the millimeter-wave radar 22, a laser radar, an ultrasonic sensor, etc. may be used to measure the distance to the object and the relative speed. Furthermore, a position and speed measuring device may be configured using multiple sensors.

[0024] The wheel speed sensor 23 detects the absolute speed of the vehicle 1 . The acceleration sensor 24 detects the acceleration of the vehicle 1 (longitudinal acceleration / deceleration in the longitudinal direction, and lateral acceleration in the lateral direction). The gyro sensor 25 detects the angular velocity of the vehicle 1. The steering angle sensor 26 detects the rotation angle (steering angle) of the steering wheel of the vehicle 1. The accelerator sensor 27 detects the amount of depression of the accelerator pedal. The brake sensor 28 detects the amount of depression of the brake pedal. The positioning sensor 29 is a GPS system, and detects the position of the vehicle 1 (current vehicle position information).

[0025] The navigation system 30 can provide map information to the ECU 10. The ECU 10 identifies roads, intersections, traffic signals, buildings, etc. that exist around the vehicle 1 (particularly ahead in the direction of travel) based on the map information and current vehicle position information. The navigation system 30 has a communication device for receiving map information from an external map information source, receives high-precision map information around the current vehicle position from the external map information source, and stores and updates the information in its internal memory as needed. Alternatively, the navigation system 30 may store the map information in its internal memory.

[0026] The steering control system 31 is a controller that controls the steering device of the vehicle 1. When it is necessary to change the traveling direction of the vehicle 1, the ECU 10 outputs a steering direction change request signal to the steering control system 31, requesting a change in the steering direction.

[0027] The engine control system 32 is a controller that controls the engine of the vehicle 1. When it is necessary to accelerate or decelerate the vehicle 1, the ECU 10 outputs an engine output change request signal to the engine control system 32, requesting a change in engine output.

[0028] The brake control system 33 is a controller for controlling the brake device of the vehicle 1. When it is necessary to decelerate the vehicle 1, the ECU 10 outputs a brake request signal to the brake control system 33, requesting the generation of a braking force on the vehicle 1.

[0029] Next, the flow of the lane marking generation process according to this embodiment will be described with reference to Figures 2 and 3. When the vehicle control device 10 acquires map information (point cloud data of lane marks), it executes lane marking generation processing to generate lane marks to be used for control (such as generating a target driving route for autonomous driving). These lane marks are generated by modifying the point cloud data of the map information through the lane marking generation processing. In this embodiment, smoothed lane marks can be calculated with a smaller calculation load without using complex m-th degree polynomial approximation (m>2) as in the past.

[0030] In the following embodiments, map information acquired from an external map information source is used as the map information. However, the present invention is not limited to this. Map information may also be map information stored in the memory of the navigation system 30 or position information of lane markings calculated from image data acquired by the camera 21 of the vehicle 1. In these cases, the position information of lane markings is provided as point cloud data representing the position coordinates of multiple discrete points.

[0031] First, in the data point identification process, the current vehicle position acquired by the positioning sensor 29 and map information acquired from the navigation system 30 are prepared. The map information is point cloud data (waypoints) representing lane markings. Each point data is position information for each discrete point. As shown in FIG. 3, the EUC 10 refers to the current vehicle position from the acquired point cloud data, and selects point cloud data consisting of multiple points WPn (white circles and shaded circles; n=1, . . . ) that meet a threshold value as a selective process. This selects point cloud data within a predetermined distance range (e.g., 100 m) from the vehicle 1. Note that, for ease of understanding, FIG. 3 shows point cloud data WPn for lane markings on the right side of a lane. However, in this embodiment, similar processing is performed for lane markings on the left side of a lane, etc.

[0032] Furthermore, in the data point identification process, as shown in FIG. 3, among the multiple points WPn, multiple points Pm (shaded circles; m=1, . . . ) located within a window Wk (k=1, . . . , kmax) along the traveling direction of the vehicle 1 are identified. Each window Wk moving along the lane identifies a partial section of the sequence of the multiple points WPn. The length of the window Wk is set to a length such that at least three points Pm exist within the window. In this embodiment, the length of the window Wk is set to, for example, 5 m, and if there are no missing points within this window Wk, 10 or more points Pm exist. Furthermore, in the data point identification process, a process is executed to exclude points having abnormal position information from the multiple points Pm.

[0033] The window Wk is set within a partial range within the set range for generating lane markings, and moves sequentially along the lane. The set range is, for example, from the proximal end position closest to the vehicle 1 (to the side of the vehicle 1) to the distal end position farthest from the vehicle 1 (100 m ahead of the vehicle 1). The window Wk is set to move sequentially a predetermined movement distance, for example, from the proximal end position (k=1) on the side of the vehicle 1 to a distal end position (k=kmax) a predetermined distance ahead. As will be described later, a corrected data point Qk (shaded rectangle) is generated at the set position of each window Wk.

[0034] Next, in the centroid generating step, a centroid Gk (triangle) of the multiple points Pm identified in the data point identifying step is calculated for the window Wk, as shown in Fig. 3. The centroid or centroid position is a geometric centroid on the xy plane when the multiple points Pm are viewed in a planar view.

[0035] Simultaneously with the centroid point generating step, or before or after, a virtual circle generating step is executed. In the virtual circle generating step, as shown in FIG. 3, a virtual circle VCk that best approximates the multiple points Pm identified in the data point identifying step is calculated for the window Wk. That is, a virtual circle VCk that best arranges the multiple points Pm in the window Wk on a circle is calculated by the least squares method. In FIG. 3, a virtual circle VCk ((x-x)) with a center point Ck(x0, y0) and a radius r is calculated. 2+(y-y0) 2 =r 2 ) is calculated. Since the multiple points Pm include at least three points, a unique virtual circle is calculated by the least squares method. In FIG. 3, the longitudinal direction of the vehicle 1 is the y-axis, and the direction perpendicular to the y-axis is the x-axis.

[0036] In this embodiment, in the least squares method when calculating the virtual circle VCk, for example, a process of solving a minimization problem of the evaluation function J is executed. J=1 / 2Σ(xi 2 +yi 2 +αxi+βyi+γ) 2 where α=-2x0, β=-2y0, γ=x0 2 +y0 2 -r 2 is.

[0037] Normal equation A{α,β,γ} T x0, y0, r are derived from α, β, γ obtained by solving =b. Normal equation A is a simultaneous equation obtained by transforming a matrix with zero differentials in α, β, γ and collecting αβγ only on the left side.

[0038] For example, if multiple points Pm are located exactly on a circle that follows the curve of the lane, a virtual circle VCk having the radius of curvature of the curve is calculated.Also, if multiple points Pm are located exactly on a straight line, a virtual circle VCk having a center at an infinitely far side and a radius of infinite length is calculated.

[0039] Next, in the modified data point generating step, a modified data point Qk in the window Wk is calculated. In the modified data point generating step, a straight line Lk is calculated connecting the center of gravity Gk calculated in the center of gravity generating step and the center point Ck of the virtual circle VCk calculated in the virtual circle generating step, and the intersection of the straight line Lk and the virtual circle VCk is calculated as the modified data point Qk (shaded square). Of the two intersections, the intersection closer to the center of gravity Gk is selected as the modified data point Qk.

[0040] The window Wk is moved sequentially within the set range, and for each window Wk (k=1 to kmax), a data point specifying step, a centroid point generating step, a virtual circle generating step, and a corrected data point generating step are executed.

[0041] Next, in the lane marking generation process, a modified lane marking is calculated based on the modified data points Qk (k = 1 to kmax) calculated for each of the multiple windows Wk. The modified lane marking may be a collection of multiple modified data points Qk. Alternatively, the modified lane marking may be a collection of adjacent discrete points spaced at a predetermined interval within a set range. When using discrete points other than the modified data points, additional discrete points can be added between adjacent modified data points by interpolation. For example, the additional discrete points can be placed on a straight line connecting adjacent modified data points.

[0042] In this embodiment, when smoothing the acquired point cloud data WPn, there is no need for approximation calculations using an mth-order polynomial (m>2) as in the past. That is, in this embodiment, only a simple approximation calculation such as the virtual circle VCk, which is a second-order polynomial, is required, and this approximation calculation does not require pre-processing or post-processing such as coordinate transformation. Therefore, in this embodiment, it is possible to perform the smoothing process when calculating the lane markings with a low calculation load.

[0043] Next, the processing flow of the lane marking generation processing according to this embodiment will be described with reference to Fig. 4. Fig. 4 is a flowchart of the lane marking generation processing. First, the ECU 10 acquires the current vehicle position of the vehicle 1 and the position coordinates of data points WPn of a data point group within a predetermined range ahead of the vehicle 1 (S11). This information is acquired from the positioning sensor 29 and the navigation system 30. In this processing, a window Wk is set, and multiple data points Pm located within the window Wk are selected.

[0044] Next, the ECU 10 executes a process of excluding points having abnormal position information from the selected plurality of data points Pm (S12). In this embodiment, data points Pm located at a distance from the imaginary circle VCk-1 derived in the preprocessing cycle for the window Wk-1 that is equal to or greater than a predetermined threshold (threshold distance) are excluded. In this embodiment, for example, regardless of whether the data point Pm is located inside or outside the imaginary circle VCk-1, if the shortest distance from the data point Pm to the imaginary circle VCk-1 is equal to or greater than a predetermined threshold, the data point Pm is excluded. Note that this process does not need to be executed if the imaginary circle VCk-1 does not exist, such as in the first processing cycle (k=1).

[0045] Next, the ECU 10 determines whether the number of remaining data points Pm is three or more (S13). If the number of data points is two or less (S13; No), the corrected data point Qk cannot be calculated in this processing cycle, so the processing for the window Wk ends and the process moves to the next processing cycle for the next window Wk+1. On the other hand, if the number of remaining data points is three or more (S13; Yes), the process moves to the center of gravity point generation process. Note that steps S11 to S13 correspond to the data point identification process.

[0046] Next, the ECU 10 executes a center of gravity generating step (S14) to calculate a center of gravity Gk for the plurality of data points Pm in the window Wk. Furthermore, the ECU 10 executes a virtual circle generating step (S15) to calculate a virtual circle VCk for the window Wk by the least squares method.

[0047] Next, the ECU 10 calculates a straight line Lk connecting the center point Ck of the imaginary circle VCk and the center of gravity Gk, and derives the intersection point between the straight line Lk and the imaginary circle VCk (S16). Since there are two intersection points, the ECU 10 sets the intersection point that is closer to the center of gravity Gk of the two intersection points as the correction data point Qk for the window Wk (S17). Note that steps S16 to S17 correspond to a correction data point generating step.

[0048] Next, the ECU 10 moves the window Wk (k=1 to kmax) along the lane line within a predetermined range, and repeats steps S11 to S17 for each window Wk. As a result, the ECU 10 generates a set of corrected data points Qk within the predetermined range ahead of the vehicle 1. Finally, the ECU 10 executes a lane marking generation step (S18). That is, the ECU 10 outputs the set or point cloud of the generated corrected data points Qk as a lane marking.

[0049] Next, the operation of the vehicle control system of this embodiment will be described. This embodiment is a lane marking generation method for vehicles for generating lane markings, and includes a data point specification step (S11 to S13) for specifying at least three or more data points Pm located within a window Wk that specifies a predetermined distance range of the lane marking based on map information including a plurality of data points WPn that indicate position information of a plurality of discrete points on the lane marking; a gravity point generation step (S14) for calculating a geometric gravity point Gk of the specified plurality of data points Pm within the window Wk; a virtual circle generation step (S15) for calculating a virtual circle VCk that approximates the specified plurality of data points Pm within the window Wk; and a method for generating a lane marking within the virtual circle VCk. This method for generating lane lines is characterized by comprising: a modified data point generation process (S16 to S17) for setting the intersection point of the two intersection points between the virtual circle VCk and the straight line Lk connecting the center point Ck and the center of gravity Gk, which is closer to the center of gravity Gk, as the modified data point Qk; and a lane line generation process (S18) for moving the window Wk along the lane and repeating the data point identification process (S11 to S13), the center of gravity generation process (S14), the virtual circle generation process (S15), and the modified data point generation process (S16 to S17), generating a point cloud of multiple modified data points Qk, and setting the point cloud of multiple modified data points Qk as a lane line.

[0050] In this embodiment, the lane marking data (multiple data points) from the map information is divided into sections by windows, approximated by virtual circles, and corrected data points for each window are calculated. Then, in this embodiment, by moving the window Wk along the lane, corrected lane marks consisting of multiple corrected data points can be generated over a predetermined range of the lane. Therefore, in this embodiment, even if the lane marking data points in the map information are missing or vary, smooth lane marking data can be generated with a low computational load while avoiding processes that require high computational loads, such as complex polynomial approximation and coordinate transformation processing.

[0051] In the present embodiment, the virtual circle VCk is calculated by the least squares method in the virtual circle generation step (S15). According to the present embodiment configured in this manner, the virtual circle VCk can be calculated with a low calculation load using a quadratic polynomial that represents the circle.

[0052] In addition, this embodiment further includes a step of receiving map information from a map information source. According to this embodiment configured as described above, since the map information is received from an external map information source, it is not necessary to perform calculations to newly generate map information from image data captured by a camera or the like in the vehicle.

[0053] In this embodiment, the lane includes the lane on which the vehicle 1 is traveling and lanes adjacent to or merging with the lane on which the vehicle 1 is traveling. According to this embodiment configured as described above, it is possible to easily generate dividing lines for lanes other than the lane on which the vehicle 1 is traveling.

[0054] Furthermore, this embodiment is an ECU (lane marking generation device) 10 for a vehicle for generating lane markings, and the lane marking generation device 10 includes a data point specification step (S11 to S13) for specifying at least three or more data points Pm located within a window Wk that specifies a predetermined distance range of the lane marking, based on map information including a plurality of data points WPn that indicate position information of a plurality of discrete points on the lane marking; a gravity point generation step (S14) for calculating a geometric gravity point Gk of the specified plurality of data points Pm within the window Wk; and a virtual circle generation step (S15) for calculating a virtual circle VCk that approximates the specified plurality of data points Pm within the window Wk. and a lane line generation process (S18) in which the window Wk is moved along the lane, and the data point identification process (S11 to S13), the center of gravity point generation process (S14), the virtual circle generation process (S15), and the modified data point generation process (S16 to S17) are repeated to generate a point cloud of multiple modified data points Qk, and to set the point cloud of multiple modified data points Qk as a lane line. [Explanation of symbols]

[0055] 1 vehicle 10 Vehicle control device (ECU, lane marking generation device) 100 Vehicle Control System Ck center point Gk center of gravity Lk straight line Pm data points Qk corrected data points VCk Virtual Circle WPn data points Wk window r radius

Claims

1. A method for generating lane markings for a vehicle, for generating a data point cloud of lane markings used by a vehicle computer to calculate a target driving path, wherein the vehicle is equipped with an on-board camera that provides image data of the surroundings of the vehicle or a navigation system that provides map information, a data point specifying step in which the computer acquires a plurality of data points indicating position information of a plurality of discrete points on the lane markings based on the image data or the map information, and specifies at least three or more data points located within a window that specifies a predetermined distance range of the lane markings; a centroid generating step in which the computer calculates a geometric centroid of the at least three or more data points identified within the window; a virtual circle generation step in which the computer calculates a virtual circle that approximates the at least three or more data points identified within the window; a correction data point generating step in which the computer sets, as a correction data point, one of two intersection points between the virtual circle and a straight line connecting the center point of the virtual circle and the center of gravity, which is closer to the center of gravity; a lane marking generation process in which the computer moves the window along the lane, repeats the data point identification process, the center of gravity point generation process, the virtual circle generation process, and the corrected data point generation process, generates a point cloud of a plurality of corrected data points, and sets the point cloud of the plurality of corrected data points as the data point cloud of the lane marking.

2. The lane marking generating method according to claim 1 , wherein in the virtual circle generating step, the computer calculates the virtual circle by a least squares method.

3. The method for generating demarcation lines as described in claim 1, further comprising a step in which the navigation system receives the map information from a map information source.

4. The lane marking generation method according to claim 1 , wherein the lanes include a lane on which the vehicle is traveling and lanes adjacent to or merging with the lane on which the vehicle is traveling.

5. A lane marking generation device for a vehicle for generating a data point cloud of lane markings used by a computer of the vehicle to calculate a target driving path, comprising: The lane marking generation device includes the computer, and an on-board camera that provides image data of an area around the vehicle or a navigation system that provides map information, The computer of the lane marking generation device a data point identifying step of acquiring a plurality of data points indicating position information of a plurality of discrete points on the lane markings based on the image data or the map information, and identifying at least three or more data points located within a window that identifies a predetermined distance range of the lane markings; a centroid generating step of calculating a geometric centroid of the at least three or more data points identified within the window; a virtual circle generation step of calculating a virtual circle that approximates the at least three or more data points identified within the window; a correction data point generating step of setting, as a correction data point, one of two intersections between the virtual circle and a straight line connecting the center point of the virtual circle and the center of gravity, which is closer to the center of gravity; a lane marking generation process for generating a point cloud of a plurality of corrected data points by moving the window along the lane and repeating the data point identification process, the center of gravity point generation process, the virtual circle generation process, and the corrected data point generation process, and for setting the point cloud of the plurality of corrected data points as the data point cloud of the lane marking.

6. The lane marking generation device according to claim 5 , wherein the computer of the lane marking generation device calculates the virtual circle by a least squares method in the virtual circle generating step.

7. The lane marking generation device according to claim 5 , wherein the navigation system of the lane marking generation device is further configured to receive the map information from a map information source.

8. The lane marking generation device according to claim 5 , wherein the lanes include a lane on which the vehicle is traveling and lanes adjacent to or merging with the lane on which the vehicle is traveling.

Citation Information

Patent Citations

  • Lane detector

    JP1997027097A

  • Image processor

    JP2015153319A

  • Vehicle travel control device

    JP2018154304A

  • Route generation device and route generation method

    JP2021004734A

  • Route generation device and route generation method

    JP6884173B2