WO2018177026A1

WO2018177026A1 - Device and method for determining road edge

Info

Publication number: WO2018177026A1
Application number: PCT/CN2018/075130
Authority: WO
Inventors: 胡传远; 付晶玮
Original assignee: 蔚来汽车有限公司
Priority date: 2017-03-29
Filing date: 2018-02-02
Publication date: 2018-10-04
Also published as: CN106991389B; CN106991389A

Abstract

A device and method for determining a road edge, relating to the technical field of smart cars. The device for determining a road edge comprises: a radar detector (110), which is mounted on a vehicle (100) and can at least detect static targets beside a road edge of a road on which the vehicle (100) is located; and a processing member (130), configured to receive the static targets detected by the radar detector (110) and extract arrangement information of the static targets which are approximately arranged regularly relative to the road, so as to obtain road edge information on the basis of the arrangement information. The device and method are applicable to acquisition of rode edge information on an unstructured road, and can be used for relatively accurately acquiring road edge information at a distance.

Description

Apparatus and method for determining the edge of a road

Technical field

The invention belongs to the technical field of intelligent automobiles, and relates to an apparatus and a method for determining a road edge based on a stationary target beside a road edge.

Background technique

Autonomous driving (including assisted driving) is an important direction for the development of smart cars, and more and more vehicles are beginning to apply automatic driving systems to achieve the automatic driving function of the vehicle. Generally, an automatic driving system can determine the travelable area of the vehicle at any time. In determining the travelable area, an important aspect is the need to determine the road edge of the current traveled road.

At present, an automatic driving system usually determines an image edge by an image including a lane line collected by an image sensor (for example, a camera mounted on a vehicle), wherein the road edge is an image based on a lane line in an image acquired in real time. Processing to determine. This technique of determining the edge of a road has at least one of the following problems:

In the first aspect, it is necessary to rely on the lane line of the lane. It is difficult to determine the edge of the road for the lane line blur, the lane line portion is missing or the lane line is completely absent, or the determined road edge is largely deviated from the real road edge;

In the second aspect, the technique for determining the edge of the road is implemented based on an image sensor. However, in practical applications, the amount of information carried by the image sensor on the close-range image and the long-distance image is different. Generally, in terms of the actual physical distance represented by two pixel points on the image, the distance near the center point of the image sensor lens is smaller than the boundary area of the lens, so that it is easy to bring a problem of poor recognition of the lane line at a long distance. That is, the determination or detection of a road edge at a long distance (relative to the vehicle) is inaccurate.

Summary of the invention

At least one aspect or other technical problem of the technical problem to be solved by the present invention provides the following technical solutions.

According to an aspect of the present invention, an apparatus for determining a road edge is provided, comprising:

a radar detector mounted on the vehicle that is capable of detecting at least a stationary target beside the road edge of the road on which the vehicle is located;

And a processing component configured to: receive the stationary target detected by the radar detector and extract arrangement information of the stationary target substantially aligned with the road, thereby obtaining road edge information based on the arrangement information.

According to still another aspect of the present invention, a method for determining a road edge is provided, comprising the steps of:

(a) detecting a stationary target beside the road edge of the road on which the vehicle is located;

(b) extracting the arrangement information of the stationary objects arranged substantially in a regular manner with respect to the road, and obtaining the road edge information based on the arrangement information.

According to still another aspect of the present invention, a vehicle is provided provided with an automatic driving system in which any of the above-described devices for determining a road edge is provided.

The above features and operations of the present invention will become more apparent from the following description and drawings.

DRAWINGS

The above and other objects and advantages of the present invention will be more fully understood from the aspects of the appended claims.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a block diagram showing the construction of a device for determining a road edge in accordance with an embodiment of the present invention.

FIG. 2 is a schematic diagram of an application scenario of the apparatus of the embodiment shown in FIG. 1 when determining a road edge.

3 is a flow chart of a method of determining a road edge in accordance with an embodiment of the present invention.

detailed description

The invention will now be described more fully hereinafter with reference to the accompanying drawings in which FIG. However, the invention may be embodied in many different forms and should not be construed as being limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete. In the drawings, the same reference numerals are used to refer to the same elements or components, and the description thereof will be omitted.

Some of the block diagrams shown in the figures are functional entities and do not necessarily have to correspond to physically or logically separate entities. These functional entities may be implemented in software, or implemented in one or more hardware modules or integrated circuits, or implemented in different network and/or processor devices and/or microcontroller devices.

1 is a schematic view showing the structure of a device for determining a road edge according to an embodiment of the present invention, and FIG. 2 is a schematic view showing an application scenario of the device of the embodiment shown in FIG. 1 when determining a road edge. The apparatus of the embodiment of the present invention and its working principle are exemplified below with reference to FIG. 1 and FIG.

As shown in FIG. 1, a device for determining a road edge (hereinafter simply referred to as "determining device") is mounted on the vehicle 100, and the specific type of the vehicle 100 is not limited, and the vehicle 100 is the determining device with respect to the determining device. Host vehicle. The determining device can be applied to an automatic driving system in which the vehicle 100 is installed.

2, the vehicle 100 is traveling on a road 900 having

corresponding road edges

901a and 901b, wherein 901a is the left road edge and 901b is the right road edge. In this application scenario, the road edge 901a And 901b are not clearly identified by lane lines, or there is no corresponding lane line in a section of road 900 of the example to identify the road rim. There are various stationary (relative road stationary) objects on both sides of the road 900, which are targets for determining device detection, and therefore, also referred to as "stationary targets"; by way of example, stationary targets beside the left road edge 901a of the road 900 are shown For example, tree 801, utility pole 802, isolated 803 (where three isolated 803a, 803b, and 803c are shown), etc., it should be understood that the stationary target beside the road edge is not limited to the object type of the above embodiment, for example, It can be a fence, a sign pole, etc.

The determining device primarily includes a radar detector 110 mounted on the vehicle 100 that is capable of detecting at least a stationary target on at least one side of the road of the road 900 on which the vehicle 100 is located. In one embodiment, the radar detector 110 is a millimeter wave radar that is mounted at the front end of the vehicle 100 and is capable of detecting various objects in front of the road plane at a 90[deg.] detection angle range, including, for example, the road shown in FIG. A stationary target beside the edge 901a. The radar detector 110 emits electromagnetic waves of a certain wavelength during detection and receives reflections from objects in front, so that the position of various objects can be detected, especially for objects at a long distance (for example, 40 meters or more), as with close objects. It can be detected relatively accurately (relative to image sensor 120) and, therefore, it has better remote sensing characteristics relative to image sensor 120.

It should be noted that the vehicle coordinate system, that is, the XY coordinate system, may be defined in the determining device, wherein the center of mass of the vehicle 100 is the circle O, the X axis is defined as the front vertical direction of the vehicle 100, and the X coordinate is defined as the relative vehicle. The deviation of the distance of the centroid in the vertical direction, the Y-axis is defined as the horizontal direction of the vehicle 100, and the Y-coordinate is defined as the deviation in the horizontal direction from the distance from the center of mass of the vehicle. When the radar detector 110 detects various objects (including stationary objects), the coordinates (X, Y) of an object are substantially determined, wherein the X coordinate represents the distance of the object from the centroid of the vehicle 100 in the vehicle coordinate system. The deviation in the vertical direction (i.e., the deviation on the X-axis), the Y-coordinate represents the deviation of the distance of the object from the centroid of the vehicle 100 in the horizontal direction in the vehicle coordinate system (i.e., the deviation on the Y-axis).

Among them, the millimeter wave radar is configured to determine a stationary target from among various detected objects based on the Doppler effect and the vehicle speed of the host vehicle, that is, an object that is stationary relative to the road edge 901. Therefore, the millimeter wave radar can output related information of a stationary target (for example, coordinates in a vehicle coordinate system) in substantially real time.

The radar detector 110 has the advantage of being relatively low cost and capable of accurately detecting a stationary target at a long distance (for example, 40 meters or more) when using a millimeter wave radar, but it should be understood that the radar detector 110 is not limited to a millimeter wave radar, for example, It is also possible for Lidar, which is relatively more accurate in detecting various stationary targets (including distant stationary targets), but is relatively expensive and requires more data processing capability for subsequent processing components 130.

Continuing, as shown in FIG. 1, the determining device further includes a processing component 130 that is also disposed on the vehicle 100, which may be implemented by a processing device in an automated driving system on the vehicle 100, or may be independently set up with respect to the automated driving system. To achieve. The processing component 130 can process the algorithm code stored therein and execute instructions from the automated driving system or vehicle, and the specific hardware implementation of the processing component 130 is known and will not be described in detail herein.

The processing component 130 can mainly perform data processing on the stationary target related information transmitted by the radar detector 110 to obtain a road edge curve, which is configured to: receive the stationary target detected by the radar detector 110 and extract a substantially regular arrangement of the relative roads. Arrangement information of the stationary target, thereby obtaining road edge information based on the arrangement information. Among them, the road edge information can be expressed as road edge curve information. The specific working principle of the processing unit 130 will be exemplified below by taking the road edge curve of the left road edge 901a shown in FIG.

In an embodiment, the number of stationary targets transmitted by the radar detector 110 after one scan detection may be up to several tens of orders of magnitude. Therefore, a corresponding screening unit 131 is provided in the processing component 130, which is capable of being able to Among the plurality of stationary targets transmitted by the detector 110, at least three still targets are selected as reference targets. As shown in FIG. 2, the tree 901 having the substantially regular arrangement of the opposite roads 900, the utility pole 902, and the isolation 903 as the reference target of the left road edge 901a can be selected from a plurality of stationary targets, and not for the opposite left road edge 901a. Static objects such as other trees and electric poles that are regularly arranged may not be selected as reference targets or filtered.

The Applicant has noted that since both sides of the road 900 will generally have objects that are generally regularly aligned with respect to the road 900, such as trees 901 and utility poles 902, etc., it may be based on the vehicle 100 when determining whether a stationary target is substantially regularly aligned with respect to the road 900. The current yaw rate (eg, can be acquired from components such as the steering system of the vehicle 100) to obtain a predicted travel trajectory that roughly corresponds to the current road curve, and thus can be based substantially on whether the stationary target is relative The predicted travel trajectories are generally regularly arranged to determine whether their stationary targets are substantially regularly aligned with respect to the road 900, so that the corresponding stationary target can be screened as a reference target. It should be understood that "substantially" in the "roughly regular arrangement" reflects that the stationary targets on both sides of the road 900 are not necessarily aligned neatly with respect to the road according to a certain rule, for example, on the order of several meters on the alignment of the opposite roads. Upper tolerances, etc.

As shown in FIG. 2, the regularly arranged trees 801, the utility poles 802, the isolation piers 803, and the like beside the left road edge 901a are determined by the screening unit 131 to be substantially regularly arranged relative to the predicted traveling trajectory, and therefore, at least Three are used as reference targets, for example, three or more trees 801 are selected as reference targets, or

isolation piers

803a, 803b, and 803c are selected as reference targets, or a plurality of trees 801 and one utility pole 802 and one isolation pier 803a are selected as reference targets. . The more the number of reference targets, the more favorable it is to get the road edge curve accurately.

In an embodiment, the processing component 130 is provided with a target curve fitting unit 132 configured to curve at least three or more reference targets in a vehicle coordinate system to obtain a corresponding reference target alignment curve. . Specifically, the reference target alignment curve that needs to be obtained is defined in advance by a quadratic function, that is, the following functional relationship (1):

Y=C ₂ ×X ² +C ₁ ×X+C ₀ ' (1)

Where X is an independent variable, which corresponds to the X coordinate in the vehicle coordinate system, and the X coordinate is defined as the deviation in the vertical direction from the centroid of the vehicle; Y is the dependent variable, which corresponds to the vehicle coordinate The Y coordinate under the system, which is defined as the deviation in the horizontal direction from the centroid of the vehicle; C ₂ is the quadratic coefficient, C ₁ is the primary term coefficient, and C ₀ ' is the constant term.

Further, the coordinates of the plurality of reference targets are substituted into the quadratic function relation (1), and the quadratic coefficient C ₂ , the term coefficient C _{1 ,} and the constant term C ₀ ' in the quadratic function relation (1) are calculated. The value, thus obtaining the relation (1), that is, determining the reference target arrangement curve.

In still another embodiment, the turning radius of the current vehicle may also be calculated based on the current yaw rate of the vehicle, thereby calculating the quadratic coefficient C ₂ in the relation (1), at this time. The sub-function relation (1) is reduced to a linear function relation. Based on the coordinates of a plurality of reference targets, the values of the primary term coefficient C ₁ and the constant term C ₀ ' can be further calculated, thereby obtaining the relation (1). That is, the reference target alignment curve is determined.

In an embodiment, the processing component 130 is provided with a road edge estimation unit 133 for estimating a road edge curve based on the reference target alignment curve. In an embodiment, the road edge estimation unit 133 obtains a corresponding road edge curve by obtaining the following quadratic function relationship (2) from the quadratic function relation (1) as a road edge curve:

Y=C ₂ ×X ² +C ₁ ×X+C ₀ (2)

Where C ₀ is a constant term and C ₀ = C ₀ '+D, where D is the distance constant of the road target relative to the stationary target that is substantially regularly arranged next to it, for example, the corresponding stationary target next to the left road edge 901a is estimated in advance ( The distance constant D of the tree 801, the utility pole 802, the isolated 803, etc. relative to the left road edge 901a, generally, the distance between the tree 801, the utility pole 802, the isolated 803 and the road edge 901 respectively has corresponding specifications, which may be based on These specified values are used to estimate the distance constant D (specifically, for example, 0.5 m)

Thus, the quadratic function relation (2) is determined, that is, the road edge curve is determined.

It should be noted that although the above example determines the reference target alignment curve and the road edge curve in a quadratic function relationship, it should be understood that it can also be based on a higher-order functional relationship (for example, a cubic function relationship, a quadratic function relationship). Equation) to determine the reference target alignment curve and the road edge curve, of course, the higher the number of functional relations, the more the number of reference targets required.

The determining device of the above embodiment can determine the road edge information based on the stationary targets on both sides of the road, is completely independent of the lane line, and is therefore very suitable for application to unstructured roads (for example, lane line blur, lane line disappearance or missing) The road edge information is obtained in the road; and it is not dependent on the image sensor. Therefore, the problem that the long-distance road cannot accurately acquire the corresponding road edge information can be obtained, and the road edge information can be obtained relatively accurately at a long distance.

The determining device of the above embodiment can be applied to the vehicle 100 having an automatic driving system based on the road edge curve provided by the determining device, which can not only give the near-end travelable area, but also can give a relatively accurate far end. The travelable area, wherein the algorithm for determining the drivable area based on the road edge curve is not limiting.

As shown in FIG. 1 , in another embodiment, an image sensor 120 may be further disposed in the determining device, which may be installed, for example, at a substantially rear view mirror position inside the vehicle. The image sensor 120 may specifically be a camera or the like, which may The lane line image information of the lane line 901 of the road 900 (if the lane line 901 is present) is acquired in real time. Of course, in practical applications, the image information acquired by the image sensor 120 is not limited to lane line image information, for example, Includes image information such as vehicles ahead, pedestrians, obstacles, etc.

In still another embodiment, the processing component 130 in the determining device further receives the lane line image information, and further calculates a road edge curve of the road 900 in real time based on the lane line image information; performs image processing based on the lane line image information and The calculation of the road edge curve is well known in the art and will not be described in detail herein. Therefore, the processing component 130 may obtain two road edge curves obtained by the two mechanisms respectively, and the processing component 130 may determine the road edge curve of the road 900 based on the two road edge curves in different scenarios.

By way of example, in one scenario, lane lines of road 900 exist and there are stationary targets along road 900 that are generally regularly aligned with respect to road 900, based on the above two mechanisms or two road edge curves, for close-range roads, road edges The curve can use the road edge curve calculated based on the lane line image information. For the long-distance road, the road edge curve is calculated based on the stationary target to obtain the road edge curve, thus overcoming the road edge calculated based on the lane line image information. The problem of curves being inaccurate in the long distance segment.

By way of example, in yet another scenario, the lane line of the road 900 is missing or unclear, and there are stationary targets along the road 900 that are substantially regularly aligned with respect to the road 900, for missing or unclear sections of the lane line of the road The road edge curve can be calculated based on the stationary target. Thereby, the problem that the image sensor 120 cannot obtain or cannot accurately obtain the road edge curve on some road sections is overcome.

3 is a flow chart showing a method of determining a road edge in accordance with an embodiment of the present invention. A method of determining a road edge in accordance with an embodiment of the present invention is illustrated in conjunction with FIGS. 1 through 3.

First, in step S310, a stationary target beside the road edge of the road on which the vehicle is located is detected.

This step S310 can be implemented in a radar detector 110 such as a millimeter wave radar. A stationary target of at least one of the sides of the road of the road 900 where the vehicle 100 is located (for example, the side of the left road edge 901a) is detected by the radar detector 110, particularly for objects at a long distance (for example, 40 meters or more), and close objects. The same can be detected relatively accurately. The millimeter wave radar is configured to be able to determine a stationary target from among various detected objects based on the Doppler effect, that is, an object that is stationary relative to the road edge 901, for example, including a tree 801, a utility pole 802, and an isolation 803. Therefore, the millimeter wave radar can output related information of a stationary target (for example, coordinates in a vehicle coordinate system) in substantially real time.

Further, in step S320, the reference target is selected from the stationary target.

This step S320 is implemented in the screening unit 131 of the main processing unit 130. In this step, at least three still objects are selected from the plurality of stationary targets transmitted from the radar detector 110 as reference targets. The principle of screening is based on whether the stationary targets are roughly arranged relative to the road 900. In particular, the roads 900 may correspond to predicted driving trajectories that may be based on the current yaw angular velocity of the vehicle 100 (eg, may be from the vehicle 100) Obtained from the components such as the steering system, the trees 801, the utility poles 802, the isolation piers 803, and the like arranged based on the rules of the left road edge 901a are determined to be substantially regularly arranged relative to the predicted traveling trajectory, thereby screening them. For reference purposes. The more the number of reference targets, the more favorable it is to get the road edge curve accurately.

Further, in step S330, curve fitting is performed on the reference target to obtain a reference target alignment curve.

This step S330 is mainly implemented in the target curve fitting unit 132 of the processing unit 130. In an embodiment, the reference target alignment curve that needs to be obtained is defined in advance by a quadratic function, that is, the following functional relationship (1):

Y=C ₂ ×X ² +C ₁ ×X+C ₀ ' (1)

Further, the coordinates of the plurality of reference targets (for example, the coordinates of the tree 801, the utility pole 802, and the isolation pier 803 in the vehicle coordinate embodiment) are substituted into the quadratic function relation (1), and the quadratic function relation (1) is calculated. The value of the quadratic coefficient C ₂ , the term coefficient C ₁ and the constant term C ₀ ', thereby obtaining the relation (1), that is, determining the reference target arrangement curve.

Further, in step S340, the road edge curve is estimated based on the reference target arrangement curve.

This step S330 is mainly implemented in the road edge estimation unit 133 of the processing unit 130. In an embodiment, the corresponding road edge curve is obtained by obtaining the following quadratic function relation (2) from the quadratic function relation (1) as a road edge curve:

Y=C ₂ ×X ² +C ₁ ×X+C ₀ (2)

The method for determining the road edge of the embodiment shown in FIG. 3 above is not dependent on the lane line or the image sensor, and is very suitable for application to unstructured roads (for example, lane line blur, lane line disappearing or missing roads). The road edge information is obtained in the road, and the road edge information can also be obtained relatively accurately at a long distance.

As used herein, the terms "close distance" and "distance" are respectively based on the effective detection distance of the radar detector and the effective detection distance of the image sensor, respectively. Generally, the effective detection distance of the radar detector is relative to the image. The effective detection range of the sensor is farther; therefore, the range of distance less than or equal to the effective detection distance of the image sensor is defined as "close distance", and the range of distance beyond the effective detection distance of the image sensor is defined as "distance" of the present application. . It should be understood that the "close distance" and "distance" are not based on fixed distance values. For example, the effective detection distance of different types of image sensors may also be different, for example, with image sensor technology. Development, the effective detection distance of newly emerging image sensors after this application date may also be further.

The above examples mainly illustrate the apparatus and method for determining a road edge of the present invention. Although only a few of the embodiments of the present invention have been described, it will be understood by those skilled in the art that the invention may be practiced in many other forms without departing from the spirit and scope of the invention. The present examples and embodiments are to be considered as illustrative and not restrictive, and the invention may cover various modifications without departing from the spirit and scope of the invention as defined by the appended claims With replacement.

Claims

A device for determining a road edge, comprising:

a radar detector mounted on the vehicle that is capable of detecting at least a stationary target beside the road edge of the road on which the vehicle is located;

And a processing component configured to: receive the stationary target detected by the radar detector and extract arrangement information of the stationary target substantially aligned with the road, thereby obtaining road edge information based on the arrangement information.
The apparatus of claim 1 wherein said processing component comprises:

a screening unit for screening out, from the stationary target, three or more of the stationary targets arranged substantially regularly with respect to the road as a reference target;

a target curve fitting unit for performing curve fitting on three or more of the reference targets in a vehicle coordinate system to obtain a corresponding reference target arrangement curve;

A road edge estimation unit is configured to estimate a first road edge curve based on the reference target alignment curve.
The apparatus according to claim 2, wherein said screening unit is configured to: calculate a predicted travel trajectory of the vehicle based on a current yaw rate of said vehicle, further based on whether said stationary target is traveling relative to said predicted target The trajectories are roughly arranged regularly to screen out the corresponding stationary targets as reference targets.
The apparatus according to claim 2, wherein said reference target arrangement curve is the following quadratic function relation (1):

Y=C 2 ×X 2 +C 1 ×X+C 0 ' (1)

Where X is an independent variable, which corresponds to the X coordinate in the vehicle coordinate system, and the X coordinate is defined as a deviation in the vertical direction from the centroid of the vehicle; Y is a dependent variable, and the corresponding a Y coordinate in the vehicle coordinate system, the Y coordinate being defined as a deviation in a horizontal direction from a centroid of the vehicle; C 2 is a quadratic coefficient, C 1 is a primary term coefficient, and C 0 ' is a constant item;

Wherein, based on the coordinate information of the reference target in the vehicle coordinate system, the quadratic coefficient C 2 , the first term coefficient C 1 and the constant term C 0 ' in the quadratic function relation (1) are calculated;

Wherein, the current yaw rate of the vehicle is used to calculate a turning radius of the current vehicle, thereby obtaining the quadratic coefficient C 2 ;

Further, the road edge estimating unit is further configured to: obtain the following quadratic function relation (2) from the quadratic function relation (1) as the first road edge curve:

Y=C 2 ×X 2 +C 1 ×X+C 0 (2)

Where C 0 is a constant term and C 0 = C 0 '+D, where D is the distance constant of the road edge relative to the stationary target that is regularly arranged next to it.
The apparatus of claim 1 or 2 wherein said radar detector is a millimeter wave radar.
The apparatus according to claim 2, further comprising: an image sensor mounted on said vehicle for acquiring lane line image information of said road;

Wherein the processing component is further configured to: calculate a second road edge curve of the road based on the lane line image information, and determine the road based on the first road edge curve and the second road edge curve The curve of the road edge.
The apparatus according to claim 6, wherein said processing means is further configured to: said road edge curve of a close road adopts said second road edge curve, and a road edge curve of a long distance road adopts said first Road edge curve.
The apparatus of claim 6 wherein said processing component is further configured to use said first road edge curve as its road edge curve for a road segment where the lane line of the road is missing or unclear.
A method for determining a road edge, comprising the steps of:

(a) detecting a stationary target beside the road edge of the road on which the vehicle is located;

(b) extracting the arrangement information of the stationary objects arranged substantially in a regular manner with respect to the road, and obtaining the road edge information based on the arrangement information.
The method of claim 9 wherein said step (b) comprises the substeps of:

(b1) screening, from the stationary target, three or more of the stationary targets arranged substantially regularly with respect to the road as a reference target;

(b2) performing curve fitting on three or more of the reference targets in a vehicle coordinate system to obtain a corresponding reference target arrangement curve;

(b3) estimating the first road edge curve based on the reference target arrangement curve.
The method according to claim 10, wherein in said sub-step (b1), a predicted travel trajectory of the vehicle is calculated based on a current yaw rate of said vehicle, further based on whether said stationary target is relative to said prediction The driving trajectory is roughly arranged regularly to screen out the corresponding stationary target as a reference target.
The method according to claim 10, wherein in the step (b2), the reference target arrangement curve is the following quadratic function relation (1):

Y=C 2 ×X 2 +C 1 ×X+C 0 ' (1)

Where X is an independent variable, which corresponds to the X coordinate in the vehicle coordinate system, and the X coordinate is defined as a deviation in the vertical direction from the centroid of the vehicle; Y is a dependent variable, and the corresponding a Y coordinate in the vehicle coordinate system, the Y coordinate being defined as a deviation in a horizontal direction from a centroid of the vehicle; C 2 is a quadratic coefficient, C 1 is a primary term coefficient, and C 0 ' is a constant item;

Wherein, based on the coordinate information of the reference target in the vehicle coordinate system, the quadratic coefficient C 2 , the first term coefficient C 1 and the constant term C 0 ' in the quadratic function relation (1) are calculated;

Wherein, the turning radius of the current vehicle is calculated based on the current yaw rate of the vehicle, thereby obtaining the quadratic coefficient C 2 ;

In the step (b3), the following quadratic function relation (2) is obtained from the quadratic function relation (1) as the first road edge curve:

Y=C 2 ×X 2 +C 1 ×X+C 0 (2)

Where C 0 is a constant term and C 0 = C 0 '+D, where D is the distance constant of the road edge relative to the stationary target that is regularly arranged next to it.
The method of claim 9 further comprising the step of:

Obtaining lane line image information of the road;

Calculating a second road edge curve of the road based on the lane line image information;

A road edge curve of the road is determined based on the first road edge curve and the second road edge curve.
The method according to claim 13, wherein in the step of determining a road edge curve of the road, the road edge curve of the close road adopts the second road edge curve and the road edge curve of the long distance road The first road edge curve.
The method according to claim 14, wherein in the step of determining a road edge curve of the road, the road segment having a missing or unclear lane line of the road adopts the first road edge curve as its road Edge curve.
An automatic driving system for a vehicle, comprising the apparatus for determining a road edge according to any one of claims 1 to 8.
A vehicle provided with an automatic driving system, characterized in that the automatic driving system is provided with means for determining a road edge according to any one of claims 1-8.