KR20230095662A - Method for automating road data calibration through map data smoothing with precision - Google Patents

Abstract

A method for automatically correcting errors generated geometrically by layers having geometric shapes such as lanes of a map and center lines of lanes with precision through data smoothing is provided. The method for automating road data correction according to an embodiment of the present invention includes: extracting, by a road data correction automation system, data for each layer of a high-precision map from a pre-stored high-precision map database; Generating, by the road data correction automation system, a directed graph including a driving route link in which a plurality of points in the road data are connected and expressed according to the traveling direction of the object based on the extracted data for each layer; Extracting, by an automated road data correction system, starting data for data smoothing from a directed graph; and smoothing, by the road data correction automation system, the geometric road data in the map with precision. As a result, the smoothing quality of the map is performed more accurately based on existing data than when a person manually smoothes the geometric information of the map with precision or smoothes the map through information on the actual behavior of the vehicle. It is possible to increase productivity by automating the process of the existing precision map building system without change.

Description

Method for automating road data calibration through map data smoothing with precision}

The present invention relates to a method for correcting road data of a map with precision, and more particularly, to smooth data errors generated by layers having geometric shapes such as lanes and center lines of lanes of a map with precision. It relates to a method of automatically correcting through.

For complete autonomous driving, not for driving assistance, cognitive and judgment control algorithms are essential, and in particular, there are cognitive-based and map-based methods for path generation included in judgment.

Map-based path creation methods use geometric information of a map to create a path and drive along it, but there are cases where the map is not accurate or the geometric shape of the map is not suitable for driving due to the dynamic characteristics of the vehicle.

In order to generate geometric information of a map with precision including geometric information such as lanes, center lines of lanes, road surface markings, traffic signs, and attribute information for each layer, road environment information is obtained using MMS vehicles and road environment information is obtained and mapped in detail. It goes through some automation, but most of the information is done manually by humans.

It is impossible to draw straight lines, curves, polygons, etc. with a certain accuracy depending on the skill level of the person using the tool that creates the map with precision or even if the person performing the same task.

Self-driving cars create dangerous situations by adjusting various environmental variables as well as in the real environment, and it is essential to use a simulator that can verify functions.

However, in order to utilize the precision map in the coordinate system defined by ASAM, a standard organization for autonomous driving simulators, it is necessary to convert the geometric curve of the precision map into a formula rather than a method of expressing it as a set of points.

The present invention has been made to solve the above problems, and an object of the present invention is not to simply algebraically formulate the geometric road data of a map with precision, but rather to algebraically formulate a set of points by an autonomous vehicle. Therefore, it is an object of the present invention to provide a road data correction automation method through map data smoothing with precision that optimizes the lateral acceleration and acceleration change to be minimized during driving.

According to an embodiment of the present invention for achieving the above object, a method for automating road data correction includes: extracting, by a road data correction automation system, data for each layer of a high-precision map from a pre-stored high-precision map database; Generating, by the road data correction automation system, a directed graph including a driving route link in which a plurality of points in the road data are connected and expressed according to the traveling direction of the object based on the extracted data for each layer; Extracting, by an automated road data correction system, starting data for data smoothing from a directed graph; and smoothing, by the road data correction automation system, the geometric road data in the map with precision.

In addition, in the step of extracting starting data for data smoothing, when a specific point in the graphed road data has a next point but no previous point according to the traveling direction of the object, a specific point is to be set as the starting data. can

In addition, according to an embodiment of the present invention, the method for automating road data correction may include, after starting data is extracted, determining whether an entry/exit virtual point between road data is generated by an automation system for road data correction; and generating, by the road data correction automation system, an entry/exit virtual point according to whether the decision is made.

The step of determining whether to generate an entry/exit virtual point may include determining whether an object next to the current object exists among objects of the road driving route data; And if the next object exists, determining whether the starting point of the next object is connected to a plurality of points; and generating an entry/exit virtual point is to connect the starting point of the next object to a plurality of points. If it is determined, generating entry/exit virtual points so that a plurality of objects entering the starting point of the next object can have differentiable points within a driving path link expressed in connection with the starting point of the next object; and connecting the last point of the plurality of objects entering the starting point and the entry/exit virtual point, and connecting the entry/exit virtual point to the start point of the next object.

In addition, two objects that are connected to each other are connected to each other by a polynomial when the objects are connected, and the entry and exit virtual points may be set such that the differential values at the starting point of the next object to be entered are the same.

In the smoothing step, the smoothing operation is performed from the starting data to perform the smoothing of the extracted data until there is no next data, and when the smoothing operation is performed, if a specific object represented by a formula exceeds the preset maximum length of the formula, a specific You can split an object into multiple formulas.

와

의 각 3차 도함수의 제곱값을 합한 목적함수이며, 하기 수식 4 및 수식 5의 제약조건은, 분할된 다항식 간의 값이 2차 도함수까지 같아야하는 제약조건이며, 평활화하는 단계는, 목적함수가 최소화 되도록, 최적화된 다항식의 계수를 활용하여 평활화된 점의 집합을 제공할 수 있다. In addition, in the smoothing step, when the equations consisting of the set of points defined in the Cartesian coordinate system are defined as polynomials of Equations 1 and 2 below, the objective function of Equation 3 and Equation 4 are calculated by using the coefficients of the polynomials as variables. And with the constraints of Equation 5, the objective function is minimized, and Equation 3 below

and

It is an objective function that is the sum of the squares of each third derivative of , and the constraints of Equations 4 and 5 below are constraints that the values between the divided polynomials must be the same up to the second derivative, and the smoothing step is to minimize the objective function If possible, a set of smoothed points may be provided by utilizing the coefficients of the optimized polynomial.

(Equation 1)

(Equation 2)

(Formula 3)

(Formula 4)

(Formula 5)

On the other hand, according to another embodiment of the present invention, the road data correction automation system includes a storage unit including a precision map database; And extracting the data for each layer of the precision map from the database of the map database stored in the storage unit, generating a directed graph in which a plurality of points in the road data are connected according to the moving direction of the object based on the extracted data for each layer, and a processor for extracting starting data to perform data smoothing from the graph and smoothing the geometric road data within the map with precision.

As described above, according to the embodiments of the present invention, the geometric information of the map with precision is more accurately compared to the manual smoothing of the map by a person or the smoothing through information on the actual behavior of the vehicle. It is possible to increase the quality of smoothing by performing smoothing work based on data, and to increase productivity by automating the process of the map building system with existing precision without change.

In addition, according to the embodiments of the present invention, the geometric information representation method of the map with precision is a set of points, but it can be expressed as equations through data smoothing automation, ASAM's OpenDRIVE, which is a standard in the field of autonomous driving or driving simulators. It can be used to convert to the coordinate system of

1 is a diagram provided for explanation of an automated system for correcting road data through smoothing map data with precision according to an embodiment of the present invention;
2 is a diagram illustrating a process of generating a virtual point according to an embodiment of the present invention;
3 is a diagram illustrating a process of connecting virtual points according to an embodiment of the present invention;
4 is a diagram provided in the detailed description of a processor according to an embodiment of the present invention; and
5 is a diagram provided to explain a method for automating correction of road data through smoothing map data with precision according to an embodiment of the present invention.

Hereinafter, the present invention will be described in more detail with reference to the drawings.

1 is a diagram provided to explain an automated system for correcting road data through smoothing map data with precision according to an embodiment of the present invention, and FIG. 2 illustrates a process of generating virtual points according to an embodiment of the present invention. 3 is a diagram illustrating a process of connecting virtual points according to an embodiment of the present invention.

MMS (Mobile Mapping System), which is used to construct a high-precision road map, is a system that generates 3D spatial information by combining a camera, a 3D laser system, a satellite navigation device, and a mileage sensor.

MMS can perform on-site survey, control point survey, acquisition of 3D spatial data through MMS vehicles, and produce maps with precision through detailed drawing and structured editing based on spatial data.

High-precision maps are often used by self-driving vehicles to locate themselves, create optimal routes, and follow them. In addition, it is used to increase the recognition accuracy or improve the recognition speed through the extraction of the sensing region of interest of the recognition algorithm through the sensor using the precision map.

Most of the detailed drawing and structured editing processes of precision road maps are performed by human hands, and many parts are being automated with the recent development of artificial intelligence technology. However, among the road data, a driving path, which means the center line of a lane, is manually created by a person through a drawing tool. In this case, different types of lines are drawn depending on the person, and the quality of the map may deteriorate due to precision.

In addition, when an autonomous vehicle creates an optimal route based on a precision road map and follows it, the quality of the road map is very important. If the quality of the log map is maintained at a certain level or higher, or if a process of correcting the map with precision so that the driving texture is similar even if the driving is based on the map with any precision, this process is automated to provide additional resources in the technology development process. It is important to ensure that allocation is not required.

In the road driving route data of the map with precision, a set of points becomes one object, and when these objects are gathered, it becomes one layer. Data smoothing refers to fitting a curve through a plurality of point data, and through this, a plurality of point data is re-extracted as point data within a curve to be smoothed and corrected. A plurality of point data is classified into a set of points with a certain length, and each set of points is fitted with a single curve. The road driving route data of the precision road map has a branching point at the part where the characteristics of the road change. At these branch points, in particular, in many cases, lines having a large gradient change are connected between objects of the driving route data. If the line segment with a large gradient change is connected, there is a problem that the error of curve fitting increases.

Therefore, the road data correction automation system through precision map data smoothing according to the present embodiment (hereinafter, collectively referred to as 'road data correction automation system') generates virtual points between objects for data smoothing automation. By removing the non-differentiable points and minimizing the above-mentioned errors through an optimization method that includes an objective function and constraints reflecting the dynamic characteristics of the vehicle and reflecting the characteristics of the road, the lanes of the map and the center line of the lane with precision Errors generated geometrically by layers having the same geometric shape of the same line are automatically corrected through data smoothing. When the vehicle drives along the road data, the lateral acceleration and the change in acceleration may be optimized to be minimized.

To this end, the road data correction automation system may include a communication unit 100, a processor 200, and a storage unit 300.

The communication unit 100 is a communication means for transmitting and receiving data necessary for the processor 200 to operate, and may perform communication in a wireless communication method or a wired communication method.

The storage unit 300 is a storage medium that stores programs and data necessary for the processor 200 to operate.

Specifically, the storage unit 300 may include a precision map database for storing data on a precision map.

The processor 200 may automate data correction of the map with precision including lane-unit information used for autonomous vehicles, robotics, road management, and the like.

Specifically, the processor 200 automatically corrects errors generated geometrically by layers having a line-shaped geometric shape, such as the lanes of the map and the center lines of the lanes, through data smoothing, and allows the autonomous vehicle to drive on the road. When driving according to the data, the lateral acceleration and the change in acceleration can be optimized to be minimized.

To this end, the processor 200 extracts precision map layer data from the precision map database stored in the storage unit 300, and based on the extracted data for each layer, a plurality of points in the road data indicate the moving direction of the object. It is possible to create a directed graph connected according to the direction graph, extract starting data for data smoothing from the directed graph, and smooth the geometric road data in the map with precision.

Specifically, the processor 200, when extracting start data to perform data smoothing, when a specific point in the graphed road data has a next point but no previous point according to the traveling direction of the object, the specific point can be set as starting data.

In this case, the precision road map may be composed of a plurality of layers including driving route points representing roads, driving route nodes, road surface line markings (lanes), traffic lights, road surface markings, and speed bumps.

That is, the processor 200 extracts data such as a driving route link, a driving route node, and a road surface line display (lane) from among a plurality of layers of the precision road map, and uses the extracted driving route points, driving route nodes, and lanes. Thus, a directed graph in which driving route points are connected to edges and driving route nodes are connected to nodes can be created.

At this time, in the directed graph, a graph having a structure in which several points can be connected to one point due to the characteristics of the road and has all information about this is generated.

Further, after the start data is extracted, the processor 200 may determine whether the road data correction automation system generates an entry/exit virtual point between road data, and may generate an entry/exit virtual point according to the determination.

Specifically, the processor 200 may determine whether there is an object next to the current object among objects of the road driving path data when determining whether to generate an entry/exit virtual point.

In addition, if the next object exists, the processor 200 may determine whether the starting point of the next object is connected to a plurality of points in order to determine whether the next object enters or exits.

In addition, when the processor 200 determines that the starting point of the next object is connected to a plurality of points, as illustrated in FIG. Create an entry/exit virtual point so that it can have a differentiable point within the driving path link expressed in connection with the start point of, and connect the entry/exit virtual point with the last point of a plurality of objects entering the start point as illustrated in FIG. and connect the entry/exit virtual point with the starting point of the next object.

Specifically, if a specific object includes an entry/exit point, since a plurality of points are connected to the entry/exit point, the connecting portion of each connected point is an indifferentiable point.

Therefore, in order to separate point objects and objects and fit and connect them to polynomials with different coefficients, the connection point of each connected point is the same point as the start point and end point of the object, respectively, and the first and second derivatives at this point are must have the same constraints.

외 모든 객체)의 끝 점을 가상 포인트와 연결하고 가상 포인트를 진입 대상 객체(그림 2 및 도 3의

)의 시작점과 연결할 수 있다. At this time, as illustrated in FIG. 2 , the processor 200 may have a differentiable point in the driving route link expressed by connecting the entry target object and the points of the two driving route objects when entering one object as an entry target, as illustrated in FIG. 2 . Each object entering by creating a virtual point (Figs. 2 and 3)

All other objects) are connected with the virtual point, and the virtual point is connected to the entry target object (Fig. 2 and Fig. 3

) can be connected to the starting point of

In summary, two objects connected to each other are connected to each other by a polynomial when the objects are connected, and the entry and exit virtual points may be set such that the differential values at the starting point of the next object, which is the entry target object, are the same.

4 is a diagram provided for a detailed description of a processor 200 according to an embodiment of the present invention;

Referring to FIG. 4 , the processor 200 may include a road data object division unit 210, a data smoothing unit 220, and a data re-extraction unit 230.

The road data object division unit 210 may divide a specific road data object into several equations when a specific road data object represented by a formula exceeds a preset maximum length of the equation.

In this case, it may be defined as a polynomial of Equations 1 and 2 consisting of a set of points defined in a Cartesian coordinate system.

(Equation 1)

(Equation 2)

The data smoothing unit 220 may utilize an optimization technique so that the objective function is minimized with the following objective function and constraints using the coefficient of the polynomial as a variable.

Specifically, the data smoothing unit 220 calculates the objective function of Equation 3 and Equation 3 by using the coefficients of the polynomials as variables when equations consisting of sets of points defined in the Cartesian coordinate system are defined as polynomials of Equations 1 and 2. With the constraints of 4 and Equation 5, the objective function can be minimized.

(Formula 3)

와

의 각 3차 도함수의 제곱값을 합한 목적함수이며, 수식 4 및 수식 5의 제약조건은, 분할된 다항식 간의 값이 2차 도함수까지 같아야하는 제약조건일 수 있다. At this time, Equation 3 is,

and

It is an objective function that is the sum of the squares of each third derivative of , and the constraints of Equations 4 and 5 may be constraints that the values between the divided polynomials must be the same up to the second derivative.

(Formula 4)

(Formula 5)

Here, the physical meaning of the 3rd derivative is the change in acceleration, and when the vehicle travels along the points of the formula resulting from the optimization through the objective function that minimizes it, the physical meaning is minimized to enable smooth driving by minimizing the change in acceleration. contains

The data re-extraction unit 230 may provide a set of smoothed points by utilizing the coefficients of the optimized polynomial.

Through this, the processor 200 performs a smoothing operation from the starting data to be smoothed on the extracted data until there is no next data, and provides a set of smoothed points by utilizing the coefficients of the optimized polynomial expression to obtain an objective function can be minimized.

5 is a diagram provided to explain a method for automating correction of road data through smoothing map data with precision according to an embodiment of the present invention.

The method for automating road data correction through smoothing of map data with precision according to the present embodiment (hereinafter, collectively referred to as 'automatic method for road data correction') automatically calibrates road data described above with reference to FIGS. 1 to 4 It can run through the system.

Referring to FIG. 5 , the road data correction automation method extracts data for each layer of a precision map from a pre-stored precision map database using a road data correction automation system (S510), and based on the extracted data for each layer As a result, a directed graph including a driving route link in which a plurality of points in the road data are connected and expressed according to the traveling direction of the object can be generated (S520).

In addition, the road data correction automation method extracts starting data for data smoothing from the directed graph (S530), determines whether to create virtual points for entry and exit between road data, and generates virtual points for entry and exit according to whether or not the decision is made. .

Specifically, in the road data correction automation method, when determining whether to generate an entry/exit virtual point, it is determined whether there is an object next to the current object among objects of the road driving route data (S540), and if the next object exists (S540- Y), it is determined whether the starting point of the next object is connected to a plurality of points (S550), and if it is determined that the starting point of the next object is connected to a plurality of points (S550-Y), to enter the starting point of the next object An entry/exit virtual point is created so that a plurality of objects can have a differentiable point within a driving route link expressed by being connected to the start point of the next object (S560), and the last point and entry/exit virtual point of the plurality of objects entering the start point. can be connected, and the entry/exit virtual point can be connected with the start point of the next object.

In addition, the road data correction automation method smoothes the geometric road data in the map with precision (S570), and repeatedly performs the smoothing operation of the geometric road data until the smoothing operation of all objects is completed (S580-Y). there is.

Through this, the geometric information of the map with precision is more accurately smoothed based on the existing data than when a person manually smoothes the map or smoothes through information on the vehicle's actual behavior, thereby improving the smoothing quality. can improve

Meanwhile, it goes without saying that the technical spirit of the present invention can also be applied to a computer-readable recording medium containing a computer program for performing the functions of the apparatus and method according to the present embodiment. In addition, technical ideas according to various embodiments of the present invention may be implemented in the form of computer readable codes recorded on a computer readable recording medium. The computer-readable recording medium may be any data storage device that can be read by a computer and store data. For example, the computer-readable recording medium may be ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, and the like. In addition, computer readable codes or programs stored on a computer readable recording medium may be transmitted through a network connected between computers.

In addition, although the preferred embodiments of the present invention have been shown and described above, the present invention is not limited to the specific embodiments described above, and the technical field to which the present invention belongs without departing from the gist of the present invention claimed in the claims. Of course, various modifications are possible by those skilled in the art, and these modifications should not be individually understood from the technical spirit or perspective of the present invention.

100: Ministry of Communication
200: processor
210: road data object segmentation
220: data smoothing unit
230: data re-extraction unit
300: storage unit

Claims (8)

Extracting, by the road data correction automation system, data for each map layer with precision from the map database with pre-stored precision;
Generating, by the road data correction automation system, a directed graph including a driving route link in which a plurality of points in the road data are connected and expressed according to the traveling direction of the object based on the extracted data for each layer;
Extracting, by an automated road data correction system, starting data for data smoothing from a directed graph; and
A method for automating road data correction, comprising smoothing, by an automated road data correction system, geometric road data within a map with precision.
The method of claim 1,
The step of extracting the starting data for data smoothing,
Road data correction automation method, characterized in that when a specific point in the graphed road data exists as the starting data, when the next point exists but the previous point does not exist according to the traveling direction of the object.
The method of claim 1,
After the start data is extracted, the road data correction automation system determines whether to create an entry/exit virtual point between the road data; and
The method for automating road data correction, further comprising: generating, by the road data correction automation system, virtual entry and exit points according to whether or not the decision is made.
The method of claim 3,
The step of determining whether to generate an entry/exit virtual point,
determining whether there is an object next to the current object among objects of the road driving route data; and
Including; if the next object exists, determining whether the starting point of the next object is connected to a plurality of points,
The step of creating an entry/exit virtual point,
If it is determined that the start point of the next object is connected to a plurality of points, multiple objects entering the start point of the next object can have differentiable points within the driving path link expressed by being connected to the start point of the next object. generating virtual points; and
Connecting the final point of a plurality of objects entering the starting point and the entry/exit virtual point, and connecting the entry/exit virtual point with the start point of the next object; characterized in that it comprises a road data correction automation method.
The method of claim 4,
Two objects connected to each other are
When connecting between objects, they are connected by polynomials,
Entry and exit virtual points,
Road data correction automation method, characterized in that the differential value at the starting point of the next object, which is the entry target object, is set to be the same.
The method of claim 4,
The smoothing step is
Smoothing is performed from the starting data to be smoothed to the extracted data until there is no next data,
When performing a smoothing operation, when a specific object represented by a formula exceeds a preset maximum length of a formula, a method for automating road data correction, characterized in that by dividing a specific object into several formulas.
The method of claim 6,
The smoothing step is
When the equations consisting of the set of points defined in the Cartesian coordinate system are defined as the polynomials of Equations 1 and 2 below, the objective function of Equation 3 below and the constraints of Equations 4 and 5 are obtained by using the coefficients of the polynomials as variables. Let the objective function be minimized,
Equation 3 below is

and

The objective function is the sum of the squares of each third derivative of
The constraints of Equations 4 and 5 below are:
It is a constraint that the values between the divided polynomials must be the same up to the second derivative,
The smoothing step is
Road data correction automation method, characterized in that for providing a set of smoothed points by utilizing the coefficients of the polynomial optimized so that the objective function is minimized.
(Equation 1)

(Formula 2)

(Formula 3)

(Formula 4)

(Formula 5)

a storage unit containing a map database with precision; and
From the precision road map database stored in the storage, data for each layer of the map is extracted, based on the extracted data for each layer, a directed graph is created in which a plurality of points in the road data are connected according to the moving direction of the object, and a directed graph is created. The road data correction automation system comprising a; processor for extracting starting data to perform data smoothing and smoothing the geometric road data within the map with precision.

Images