KR101934297B1 - METHOD FOR DEVELOPMENT OF INTERSECTION RECOGNITION USING LINE EXTRACTION BY 3D LiDAR - Google Patents

Abstract

An intersection recognition method based on line segment extraction using 3D LiDAR is disclosed. Embodiments of the present invention include: recognizing environmental information around a vehicle using a LiDAR sensor; Removing an obstacle within a reference radius range based on environmental information; And recognizing the intersection by comparing the extracted line segment with neighboring line segments by using a line segment extraction technique.

Description

TECHNICAL FIELD [0001] The present invention relates to an intersection recognition method based on segment extraction using 3D LiDAR,

The present invention relates to an intersection recognition method based on line segment extraction using a 3D laser scanner.

Research and development of self-propelled vehicles, which are attracting attention as a next-generation automobile technology, are being actively carried out. In order to enable autonomous driving of the vehicle, it is necessary to recognize the driving environment, route planning, route tracking, and vehicle control technology. In particular, the recognition of the driving environment is the basis for operating all systems of autonomous driving. This is because path planning, path tracing, and vehicle control are possible with the current position of the vehicle being accurately recognized.

The information for determining the current position of such a vehicle may be a building, a crosswalk, an intersection, the number of lanes, and the like. The intersection is an important node for matching the current position of the vehicle with the travel route.

A camera, a RADAR, a LiDAR, a GPS, or the like is used as a sensor for recognizing the position information of the vehicle. Among them, LiDAR, which is less influenced by illumination, has advantages of being able to use in dark night, and also can measure distance and height information, so it is mounted on autonomous vehicles. Therefore, methods for recognizing intersection recognition, which is crucial position information providing route generation and present position of a vehicle, through 3D LiDAR are actively researched and developed.

However, existing methods recognize only objects having a specific size such as automobiles or people, so that the points other than the intersections are often recognized as intersections.

It is therefore an object of the present invention to provide an intersection recognition method based on segment extraction using three-dimensional LiDAR that can improve the recognition rate of an intersection in determining the current position of a vehicle using LiDAR.

To this end, the intersection recognition method based on line segment extraction using 3D LiDAR according to an embodiment of the present invention includes: recognizing environmental information around a vehicle using LiDAR sensor; Removing an obstacle within a reference radius range based on environmental information; And recognizing an intersection by comparing the extracted line segment with neighboring line segments using a line segment extraction technique.

In one embodiment, the step of removing the obstacle includes the steps of: determining a reference radius range in consideration of the width of the road and the size of the vehicle; And removing all obstacles present within the determined reference radius range.

In one embodiment, the step of removing the obstacle is characterized by removing all obstacles existing within a radius of 4 m around the current position of the vehicle.

In one embodiment, the step of recognizing the environment information of the surroundings of the vehicle includes the steps of generating a two-dimensional grid map based on the three-dimensional scan data input through the LiDAR sensor, And displaying the current position.

In one embodiment, the step of comparing the extracted line segments with neighboring line segments to recognize an intersection comprises: displaying a predetermined number of rays on the generated grid map; Calculating a distance to a point at which each radiation meets the first obstacle; Extracting a plurality of lines by using the maximum value and average value information of the calculated distance, and applying a splitting & merge to the extracted lines; And recognizing the intersection by comparing the difference in length between the line segments that are finally extracted and neighboring line segments.

In one embodiment, the step of comparing the extracted line segments with neighboring line segments to recognize an intersection includes extracting line segments whose difference in length from neighboring line segments is less than 2/3 times the maximum value, Thereby recognizing the intersection.

In one embodiment, the method further includes updating the intersection recognition by repeating the obstacle control process and the segment extraction process based on the information on the running direction, the running speed, and the approaching obstacle of the vehicle while the vehicle is running .

According to the intersection recognition method based on line segment extraction using 3D LiDAR according to the embodiment of the present invention, it is possible to solve the difficulty of recognizing an intersection that may occur due to an obstacle approaching an unmanned vehicle. Also, it is possible to increase the recognition rate of the intersection recognition by comparing the lengths of the lines extracted from the grid map with the neighboring lines and verifying whether the lines intersect each other. Furthermore, by improving the intersection recognition rate, which is important position information in the running route of the unmanned vehicle, the autonomous vehicle can provide a clear basis for route planning, vehicle tracking, and vehicle control.

1 is a flowchart illustrating an intersection recognizing method according to an embodiment of the present invention.
2 is a view showing a state in which obstacles existing within a radius of 4 m from a vehicle are controlled according to an embodiment of the present invention.
FIGS. 3A and 3B are diagrams showing the comparison between the case where the line segment is extracted and the case where the line segment is extracted when there is an obstacle close to the vehicle according to the embodiment of the present invention.
FIGS. 4A and 4B are diagrams illustrating accuracy of an algorithm for determining whether an extracted line segment is an intersection according to an embodiment of the present invention. FIG.

First, the intersection recognition method according to the embodiment of the present invention can be applied to all systems requiring an unmanned surveillance reconnaissance in addition to an unmanned vehicle.

In addition, the present invention is capable of various modifications and various embodiments, and specific embodiments are illustrated in the drawings and described in detail in the detailed description. It should be understood, however, that the invention is not intended to be limited to the particular embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

Terms including ordinals such as first, second, etc. described herein can be used to describe various elements, but the elements are not limited to these terms. That is, the terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as a second component, and similarly, the second provisional component may also be referred to as a first component. The term " and / or " includes any combination of a plurality of related listed items or any of a plurality of related listed yields.

Also, when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, but other elements may be present in between have. On the other hand, when an element is referred to as being "directly connected" or "directly connected" to another element, it should be understood that there are no other elements in between.

Also, the terms used in the present application are used only to describe certain embodiments and are not intended to limit the present invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having" and the like are used to specify that there is a feature, a number, a step, an operation, an element, a component or a combination thereof described in the specification, Should not be construed to preclude the presence or addition of one or more other features, integers, steps, operations, elements, parts, or combinations thereof.

Also, unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the contextual meaning of the related art and are to be interpreted as either ideal or overly formal in the sense of the present application Do not.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the like elements throughout. The description will be omitted.

In the present invention, an intersection recognition method using three-dimensional LiDAR is proposed for recognizing an intersection, which is important information for determining the current position of an unmanned vehicle. For this purpose, in the present invention, an intersection recognition method using an obstacle removal technique and a line segment extraction approaching an unmanned vehicle is applied. LiDAR (LiDAR) is a sensor used for magnetic location recognition, which means a 3D laser scanner. LiDAR (LiDAR) is a technology capable of image modeling of space including distance information on the direction of the laser beam. Specifically, the laser beam reflected and reflected by the object is collected through a multi-array receiving element and implemented as a three-dimensional image, which is mainly installed in an outer loop of an unmanned vehicle. As a product specification to be applied to the present invention, for example, Velodyne HDL-32L, LMS511-PRO, LMS291-S05 and the like can be applied.

1 is a flowchart illustrating an intersection recognizing method according to an embodiment of the present invention.

First, a step of recognizing environmental information around the unmanned vehicle is performed using a LiDAR sensor (S10). Specifically, a three-dimensional point entered through LiDAR can be input to generate a two-dimensional grid-map. At this time, the height information of the three-dimensional point is included in each cell of the generated grid map.

In addition, in one embodiment, a GPS, an encoder, a camera (front and rear, positioning, etc.), a CAN receiver, a gyroscope, and the like may be further used for collecting environmental information in addition to the LiDAR sensor. Environmental information collected through these sensors may be displayed in the form of coordinate values, for example, in the generated grid map. For example, it is possible to recognize the stop line of the intersection, the left and right lanes and the like through the camera and the Lidar sensor, and recognize the heading direction of the unmanned vehicle at the intersection through the gyroscope.

Next, based on the collected environment information, a step of removing all obstacles within a reference radius range around the unmanned vehicle is performed (S20). Specifically, a method of removing all the obstacles in the running route of the unmanned vehicle during traveling is used. In particular, the present invention is characterized by removing not only an obstacle satisfying a predetermined size, but also all obstacles that are smaller in size, have an unspecified size, or are not fixed in position.

In the past, only objects with specific sizes such as automobiles, people, etc. were recognized. On the road, however, there are many obstacles of unspecified size such as a traffic light, a road sign, and branches spread out on the road. Therefore, the obstacles that can not be removed by the conventional method are removed in consideration of the width of the travel route and the size of the unmanned vehicle. In this connection, FIG. 2 shows a state in which all obstacles existing within a radius of 4 m from the vehicle are controlled.

For example, since the width of the road is approximately 3.3 m to 3.6 m, it is possible to increase the recognition rate of intersection recognition by removing all obstacles having a radius of 4 m from the vehicle as shown in Fig.

Then, the step of recognizing the intersection is performed by comparing lengths of the plurality of line segments extracted using the line segment extraction technique with neighboring line segments (S30).

Specifically, the scan data obtained from the LiDAR sensor (LiDAR), for example, 36 rays in the order of 1 to 36 are drawn on the grid map, and the distance to the point where the first obstacle meets is measured. Next, a plurality of line segments are extracted using the maximum value and the average value information of the measured distances. At this time, the split and merge are performed using the difference between the extracted line segment and the line segment number. The merging and separation of segments was developed by Pavlidis and Horowitz (Segmentation of plane curves, IEEE Transaction on Computers, Vol. C23, no. 8, pp. 860-870, 1974) Aldon (Line extraction in 2D range images for mobile robots, Journal of Intelligent and Robotic Systems, vol. 40, pp. 267-297, 2004). Therefore, a detailed description will be omitted here.

In one embodiment, line segments are extracted in the same direction (forward direction) as the input sequence of scan data in order to reduce the computational complexity of line segment merging and segmentation, and line segments in the opposite direction (reverse direction) It is possible to reduce the computation complexity to be less than the computational complexity proportional to the number N of scan data by adopting a method of determining a line segment to include the scan data at the end and updating the non-linear data.

Next, by using the difference in length between the line segments extracted through merging and separation, the line segment recognized as an intersection is filtered. Specifically, in order to accurately discriminate whether the extracted line segment is extracted from an intersection or an adjacent obstacle, extraction is performed if the difference in length from the neighboring line segment is smaller than 2/3 times the maximum value, . Then, the problem that the line segment other than the intersection is mistaken as the intersection due to the adjacent obstacle can be solved.

In this regard, FIGS. 3A and 3B are diagrams showing the comparison between the case where the line segment is extracted and the case where the line segment is not extracted when there is an obstacle close to the vehicle.

FIG. 3A is a method proposed by the present invention, and the diagram shown in FIG. 3B is an intersection recognition using an existing method. As shown in FIGS. 3A and 3B, when there is an obstacle approaching the unmanned vehicle, segment extraction (for example, a yellow line segment at 7 o'clock direction) is performed properly in FIG. 3A and line segment extraction is not performed in FIG. 3B , No line segment is detected at 7 o'clock direction).

Also, in one embodiment, the obstacle removing process and the segment extracting process described above can be updated in real time while considering the relative speed and direction of the moving obstacle approaching on the basis of the traveling direction and the traveling speed of the vehicle during running of the unmanned vehicle . For this, feedback of the above-described processes can be performed. Further, a new route can be searched for and displayed on the grid map or the grid map can be updated accordingly.

Further, in one embodiment, after recognizing the intersection, the relative speed of the vehicle approaching the unmanned vehicle until the unmanned vehicle escapes the intersection and the direction of travel of the unmanned vehicle (e.g., east, west, south, North, etc.), and based on the set vehicle speed and user set time, it is possible to predict whether or not the vehicle will collide within an intersection. In such a case, the collision estimation time (TTC) may be calculated and alerted based on the traveling direction of other nearby vehicles, the vehicle speed, and the time taken to the intersection center area.

Also, in one embodiment, a variable signal at an intersection may be anticipated and recognized to update whether or not there is a conflict with a moving obstacle or another vehicle and a collision estimated time (TTC).

As described above, in the present invention, when there is an obstacle approaching the unmanned vehicle, it can be accurately extracted as a line segment. Accordingly, the recognition rate of intersection recognition can be improved.

FIGS. 4A and 4B are diagrams illustrating the accuracy of the algorithm for determining whether an extracted line segment is an intersection according to an embodiment of the present invention. FIG.

Comparing FIGS. 4A and 4B, it can be seen that an advantageous effect is obtained when an algorithm for determining whether or not intersections are extracted between obstacle gaps is applied using the method proposed by the present invention. The diagram shown in FIG. 4A is a diagram to which the algorithm of the present invention is applied, and the diagram shown in FIG. 4B shows a case where an intersection is erroneously recognized by applying an existing method (for example, City, red line segments at 6 o'clock) are recognized as an intersection). That is, in the diagram shown in FIG. 4A, it can be seen that the extracted line segments due to the neighboring obstacles are filtered through the application of the proposed algorithm (for example, the yellow line segments in the 4: 00 direction are not recognized as intersections).

As described above, according to the intersection recognition method based on line segment extraction using 3D LiDAR according to the embodiment of the present invention, it is possible to solve the difficulty of recognizing an intersection that may occur due to an obstacle approaching an unmanned vehicle. Also, it is possible to increase the recognition rate of the intersection recognition by comparing the lengths of the lines extracted from the grid map with the neighboring lines and verifying whether the lines intersect each other. Furthermore, by improving the intersection recognition rate, which is important position information in the running route of the unmanned vehicle, the autonomous vehicle can provide a clear basis for route planning, vehicle tracking, and vehicle control.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation, And may be modified, changed, or improved in various forms. Further, the method according to the present invention described herein can be implemented in software, hardware, or a combination thereof. For example, a method according to the present invention may be stored in a software program that can be stored in a storage medium (e.g., terminal internal memory, flash memory, hard disk, etc.) and executed by a processor May be implemented with embedded codes or instructions.

Claims (7)

Recognizing environmental information around the vehicle using LiDAR sensor;
Removing an obstacle within a reference radius range based on environmental information; And
And recognizing the intersection by comparing the extracted line segment with neighboring line segments by using a line segment extraction technique,
Comparing the extracted line segments with neighboring line segments to recognize an intersection,
Displaying a predetermined number of radiations on the grid map;
Calculating a distance to a point at which each radiation meets the first obstacle;
Extracting a plurality of lines using the calculated maximum value and the average value information, and applying merging and separation to the extracted lines; And
And recognizing the intersection by comparing the lengths of the extracted line segments with neighboring line segments. The intersection recognition method is based on line segment extraction using 3D LiDAR. The method according to claim 1,
The step of removing the obstacle includes:
Determining a reference radius range in consideration of the width of the road and the size of the vehicle; And
And removing all the obstacles existing within the determined reference radius range based on the extracted intersection. 3. The method of claim 2,
Wherein the step of removing the obstacle is a step of removing all obstacles existing within a radius of 4m around the current position of the vehicle, wherein the intersection recognition method is based on line segment extraction using 3D LiDAR. The method according to claim 1,
The step of recognizing the environmental information of the surroundings of the vehicle includes:
Generating a two-dimensional grid map based on the three-dimensional scan data input through the LiDAR sensor, and displaying the current position of the vehicle on the generated grid map. An Intersection Recognition Method Based on Segment Extraction Using. delete The method according to claim 1,
Comparing the extracted line segments with neighboring line segments to recognize an intersection,
Extracting a line segment whose length difference from the neighboring line segment is smaller than 2/3 times the maximum value and recognizing the intersection line by eliminating the line segment that does not satisfy the line segment. Recognition method. The method according to claim 1,
Further comprising the step of updating the intersection recognition by repeating the obstacle control process and the segment extraction process based on the information about the running direction, the traveling speed, and the approaching obstacle of the vehicle while the vehicle is traveling A Cross - Section Recognition Method Based on Segment Extraction Using 3D LiDAR.

Images