KR20130142336A - Method for localization and mapping using 3d line, and mobile body thereof - Google Patents

Abstract

The present invention relates to a method for localization and mapping using three-dimensional lines, and a moving body using the same. The method for generating a map of space by the moving body comprises: obtaining each of images for space around the moving body using a plurality of cameras varying fields of view, respectively; extracting lines from the obtained images, respectively; matching the extracted lines to generate three-dimensional spatial coordinates; and generating a map, based on the three-dimensional spatial coordinates, using a Plucker matrix. [Reference numerals] (310) Obtain each of images for space around a moving body using a plurality of cameras varying fields of view, respectively;(320) Extracti lines from the obtained images, respectively;(330) Match the extracted lines to generate three-dimensional spatial coordinates;(340) Generate a map, based on the three-dimensional spatial coordinates, using a Plucker matrix;(AA) Start;(BB) End

Description

Technical Field [0001] The present invention relates to a method of recognizing a position using a three-dimensional straight line and generating a map, and a moving object according to the method,

The present invention relates to a technique of recognizing the position of a moving object and generating a map, and more particularly, to a method of recognizing the position of the moving object itself using a three-dimensional straight line and simultaneously generating a map of the surrounding environment based on the position, And a moving body according to the method.

Efforts to solve the problem of locating objects have been made for a long time. The problem of locating such a location has become increasingly more useful in modern times. Particularly, among the basic technologies that become the driving force of this era, the location recognition technology will occupy an important position to find the fields that are not used such as robot control, disaster, military, hospital, traffic.

Since artificial intelligence and robotic technology have been developed with the help of miniaturized microprocessors, it has become more and more widely used in various fields to grasp the position of the mobile body itself . Particularly, moving objects such as robots are characterized in that their positions are not fixed, unlike conventional conventional position estimating devices. Therefore, various technical means for position estimation are required.

In this regard, Simultaneous Localization and Mapping (SLAM) has been proposed as a technique for controlling the traveling of the moving object and grasping the surrounding environment where the moving object itself is located. SLAM is a technology that recognizes its position by creating a map of the surrounding environment such as an autonomous mobile robot. In the non-patent literature presented below, the technology to perform SLAM by extracting features from distance and image information is introduced .

 SLAM of mobile robot using infrared sensor and vision sensor, Yong - Joo Lim, Byung - Doo Lim, Song Jae - Bok, Korea Institute of Machinery, 2006.

SUMMARY OF THE INVENTION The object of the present invention is to overcome the problem that the performance of position recognition and map generation is deteriorated when a point is used as a feature point in a conventional mobile vision recognition technology based on vision, It is necessary not only to create a space map for the straight line extending in the vertical direction but also to overcome the limitation of the error in the position and the map construction of the moving object.

According to an aspect of the present invention, there is provided a method of generating a map of a moving object according to an embodiment of the present invention includes acquiring images of a surrounding space of the moving object using a plurality of cameras having different viewing angles step; Extracting a line from the acquired image; Generating three-dimensional spatial coordinates by matching the extracted straight lines; And generating a map based on the generated three-dimensional space coordinates using a Plucker matrix.

In the method of generating a map for the space according to an exemplary embodiment, the step of generating the three-dimensional space coordinate may include extracting depth information from a disparity generated through matching of the extracted straight lines; And defining a three-dimensional straight line using the extracted depth information and the duality between dual plucker coordinates and plucker line coordinates. The dual flutterer coordinates define the position of the camera based on pose information composed of a quaternion value considering the current position of the moving body and the rotation of the moving body, And matching the end points of the straight line extracted from the acquired image.

In the method of generating a map for the space according to an exemplary embodiment, the step of generating a map based on the three-dimensional space coordinate may include the step of generating an inverse depth And performs SLAM (Simultaneous Localization And Mapping) using the property.

In the method of generating a map for the space according to an exemplary embodiment, when the extracted straight line information is less than a threshold value, the step of generating the three-dimensional space coordinate may further include using odometry of the moving object .

According to an exemplary embodiment of the present invention, there is provided a method of generating a map for a space, the method comprising: comparing the generated three-dimensional space coordinate with a previously stored space coordinate; And selectively updating feature information on the previously stored spatial coordinates according to the comparison result. When the feature information of the generated three-dimensional spatial coordinates is present in the pre-stored spatial coordinates, the feature information is calculated using the straight line predicted using the plurality of cameras and the end point of the extracted straight line, Calculating an update value for the current time; And updating the feature information of the previously stored space coordinates using the coordinates of the moving object, the float line coordinates, and the calculated update value.

Further, the following description provides a computer-readable recording medium on which a program for causing a computer to execute a method of generating a map of a space according to the present invention described above.

According to an aspect of the present invention, there is provided an apparatus for generating a map for a space using a moving object, the apparatus comprising: a driving unit for moving the moving object in the space; A plurality of cameras provided on the mobile body to acquire images of a peripheral space of the mobile body and to have different viewing angles; And a processor for generating a three-dimensional map of the space using the obtained image, wherein the processing unit extracts a straight line from the obtained image, and matches the extracted straight line to obtain a three-dimensional space coordinate And generates a map based on the generated three-dimensional space coordinates using a floater matrix.

In an apparatus for generating a map for the space according to an exemplary embodiment, the processing unit may extract depth information from a mismatch generated through matching of the extracted straight lines, and extract the extracted depth information and dual- The three-dimensional space coordinates are generated by defining a three-dimensional straight line using the duality between the rucker linear coordinates.

In an apparatus for generating a map for the space according to an embodiment, the processing unit performs SLAM using properties of an inverse depth on a three-dimensional straight line defined through the three-dimensional space coordinates.

According to an embodiment of the present invention, an apparatus for generating a map for the space further includes an odometer for measuring travel information of the moving object, and when the extracted straight line information is less than a threshold value, And generates the three-dimensional space coordinates by additionally using the travel information measured using the travel recorder.

The apparatus for generating a map for the space according to an embodiment of the present invention further includes a storage unit for storing coordinates and feature information for the space in advance, And if the feature information of the generated three-dimensional space coordinates is present in the pre-stored space coordinates, the feature information on the pre-stored space coordinates is updated.

The embodiments of the present invention can improve the position recognition and map generation performance by generating three-dimensional spatial information by using a straight line extracted from an image obtained through a plurality of cameras. In defining a straight line in a three-dimensional space, By using the Lucker Patrix, it is possible to extract all the straight lines existing in the space other than the floor and the vertical direction, thereby reducing errors related to the position of the moving object and the map configuration, thereby enabling more accurate position recognition and map generation .

1 is a diagram illustrating a result of recognizing a space using a mono camera in the technical field to which the present invention belongs.
FIG. 2 is a diagram for explaining a method of recognizing a three-dimensional space using a stereo camera adopted in embodiments of the present invention.
3 is a flowchart illustrating a method of generating a map for a space according to an embodiment of the present invention.
4 is a diagram for explaining a method of recognizing three-dimensional space coordinates adopted by embodiments of the present invention.
5 is a diagram for explaining a method of updating a three-dimensional map according to an observation model adopted by embodiments of the present invention.
FIG. 6 is a flowchart illustrating an entire process of applying a method for generating a map to a space according to an embodiment of the present invention to SLAM (Simultaneous Localization And Mapping) using a mobile object.
FIG. 7 is a block diagram illustrating an apparatus for generating a map of a space using a moving object according to an embodiment of the present invention. Referring to FIG.
FIGS. 8 and 9 are views illustrating experimental results of generating a three-dimensional map of a specific space using a moving object according to an embodiment of the present invention.

Before describing the embodiments of the present invention in detail, after briefly introducing the techniques utilized in the technical field related to the position recognition and the space mapping in which the embodiments of the present invention are utilized, the embodiments of the present invention Suggest basic ideas.

1 is a diagram illustrating a result of recognizing a space using a mono camera in the technical field to which the present invention belongs.

As described briefly above, SLAM (Simultaneous Localization And Mapping) means that a moving body such as a mobile robot or a car recognizes its own position and simultaneously creates a map of the surrounding environment. Conventional vision-based SLAM technology uses points as feature information extracted from acquired vision information (which can be image information obtained through a camera). However, in the indoor environment, it has been found that SLAM using line information rather than point information can provide more useful information to humans because of its spatial characteristics. Based on the results of these studies, the results of SLAM implementation using linear information are shown in FIG.

In [A] of FIG. 1, a mono camera is used to acquire an image of a corridor in a building, and the SLAM is applied to extract linear information from the acquired image. However, in the case of the SLAM using the monocamera, there is a weak point that it is possible to create a map only for the bottom straight line and the straight line in the vertical direction. Therefore, it has been found that space recognition and mapping are not performed accurately as shown in [B] of FIG. 1 due to a lack of extraction information.

Also, when a mono camera is used, different modeling techniques are applied to the vertical straight line and the bottom straight line, so that it is difficult to apply the uniform rule to all the straight lines. Furthermore, since there is a limit to use only the odometry of the robot without using the linear information projected on the image at the position where the straight line is extracted as the feature information, the error in the position and the map configuration of the mobile robot is large Respectively. Therefore, in order to solve the above problems, embodiments of the present invention propose technical means as described below.

FIG. 2 is a diagram for explaining a method of recognizing a three-dimensional space using a stereo camera adopted in embodiments of the present invention.

Embodiments of the present invention proposed below propose a method of acquiring three-dimensional spatial information by capturing an image having a different viewing angle using a plurality of cameras and comparing feature information extracted from the images with each other. From the viewpoint of implementation, such a plurality of cameras utilize left and right stereo cameras, but it is also possible to use stereo cameras or three or more cameras simultaneously, which are configured up and down according to the implementation environment and necessity .

In addition, the embodiments of the present invention use feature information existing in an image photographed by each camera using a stereo camera (as described above, embodiments of the present invention use straight lines instead of points as feature information) And acquires spatial information using a disparity between mutually corresponding feature information. That is, the mobile robot according to the embodiments of the present invention obtains stereo image information about its own surroundings while moving by itself, and determines the position of the camera based on the positions of both ends of the straight line projected on each stereo image, The left / right plane can be calculated, and when two planes meet, three-dimensional space coordinates can be expressed.

As shown in FIG. 2, a straight line is extracted from an image photographed through the left camera and an image photographed through the right camera, and the position of the moving object itself (which can be the position of the camera) Dimensional space information through matching of straight lines corresponding to each other, thereby finally generating a three-dimensional map of the space that the moving object travels. In the experiment result of FIG. 2, the identification numbers of the straight lines corresponding to each image and the three-dimensional map are given, so that mutual matching relation can be confirmed.

Hereinafter, embodiments of the present invention for solving the above-mentioned technical problems will be described in detail with reference to the drawings. In the following description and the accompanying drawings, detailed description of well-known functions or constructions that may obscure the subject matter of the present invention will be omitted. It is to be noted that the same components are denoted by the same names and reference numerals as possible throughout the drawings.

FIG. 3 is a flowchart showing a method of generating a map for a space according to an embodiment of the present invention. Here, only a general outline is first described.

In step 310, the moving object obtains images of the surrounding space of the moving object using a plurality of cameras having different viewing angles. Such a plurality of cameras can be realized by a stereo camera composed of left and right sides as described above.

In step 320, the moving object extracts a line from the image obtained in step 310. In step 320, The embodiments of the present invention extract and use straight line information rather than a dot in order to extract more accurate and more information in the indoor space. Of course, since the straight line extracted from the image has both end points, it is possible to utilize a fixed point from the extracted straight line.

In step 330, the moving object generates three-dimensional spatial coordinates by matching the straight line extracted in step 320. [ For this, embodiments of the present invention extract depth information from a disparity generated through matching of straight lines extracted in step 320, and extract the depth information and dual plucker coordinates A three-dimensional straight line is defined using the duality between the plucker line coordinates. Dualplurer coordinates and plaster linear coordinates are the means used to extract the depth information corresponding to the three-dimensional information, rather than recognizing the straight line as only plane information through the stereo image. This coordinate system is utilized as a means of processing straight lines in all directions existing in the indoor environment in addition to the straight lines in the bottom and vertical directions. The process of linear matching and generation of three-dimensional space coordinates will be described later in more detail with reference to FIG.

Finally, in step 340, the moving object generates a map based on the three-dimensional spatial coordinates generated through step 330 using a Plucker matrix.

If the extracted straight line information is smaller than the predetermined threshold value, the step of generating the three-dimensional space coordinates in step 330 is preferably performed using the odometry of the moving object . In the case of the image acquisition using the stereo camera utilized in the embodiments of the present invention, it is possible to extract more linear information than the image acquisition using the conventional monochrome camera, but in some cases, It may be scarce. Therefore, in this case, it is possible to acquire and utilize the travel information through the travel recorder provided in the mobile body.

FIG. 4 is a view for explaining a method of recognizing three-dimensional space coordinates adopted by embodiments of the present invention. FIG. 4 shows a method of acquiring spatial information including depth information from left and right images photographed by a stereo camera .

As described briefly above, the embodiments of the present invention define a three-dimensional straight line by using the depth information extracted from the stereo image and the duality between the dual and global planar coordinates. More specifically, the dual plane coordinates define the position of the camera based on pose information composed of a quaternion value considering the current position of the moving body and the rotation of the moving body, And matching the end points of the straight line extracted from the image obtained through the matching process.

(One) motion  Model( motion model )

이라고 표현하고,

는 로봇의 주행기록계(odometer)를 통해 측정된 주행 정보(odometry)라고 하자. 이동 후의 로봇의 포즈는

이며, 로봇의 포즈에 대한 표현은

이라고 정의된다. 여기서,

는 로봇의 현재 위치이고,

는 로봇의 쿼터니온(quaternion)이다. 또한, 로봇의 주행 정보를 나타내는 주행 정보 벡터(odometer vector)는

와 같이 표현될 수 있다. 여기서,

는 공간을 구성하는 3개의 축(x, y, z)을 기준으로 한 회전 벡터(rotation vector)를 나타낸다.The robot's current pose

And,

(Odometry) measured through the odometer of the robot. The pose of the robot after the movement

And the expression of the pose of the robot is

. here,

Is the current position of the robot,

Is the quaternion of the robot. Also, the odometer vector representing the traveling information of the robot

Can be expressed as here,

Represents a rotation vector with respect to three axes (x, y, z) constituting the space.

(2) Line initialization

과 오른쪽

으로 정의하고, 이미지 상에 투영된 직선의 끝점을

이라고 하자.

가 트래킹 되었을 때, 왼쪽 카메라를 통해 촬영된 이미지 상에서 추출된 직선의 양 끝점

와 왼쪽 카메라의 위치

을 이용하여 하나의 평면

을 생성할 수 있다. 또한, 오른쪽 카메라를 통해 촬영된 이미지 상에서 추출된 직선의 양 끝점

와 오른쪽 카메라의 위치

를 이용하여 다른 하나의 평면

를 생성할 수 있다. 이제, 도 4와 같이 각각 왼쪽과 오른쪽에서 결정된 3개의 점을 연결하여 양 평면이 만나는 지점을 통해 듀얼 플뤼커 좌표(dual Plucker coordinates)를 생성할 수 있다. 듀얼 플뤼커 좌표는 다음의 수학식 1과 같은 플뤼커 매트릭스(Plucker matrix)를 이용하여 산출된다.Position the stereo camera left

And right

And the end point of the straight line projected on the image is defined as

.

The left and right ends of the straight line extracted on the image taken through the left camera,

And the position of the left camera

Lt; RTI ID = 0.0 > plane

Can be generated. Further, both end points of a straight line extracted on the image photographed through the right camera

And the location of the right camera

Using one plane

Lt; / RTI > Now, as shown in Fig. 4, it is possible to create dual plucker coordinates through the points where the two planes meet by connecting the three points determined on the left and right sides, respectively. The dualplurality coordinates are calculated using a Plucker matrix as shown in the following equation (1).

In addition, the dual flutterer coordinates and theplurier linear coordinates calculated through the above process can generate the flicker linear coordinates using the duality between the elements as shown in the following Equation (2).

On the other hand, in the case of matching a straight line in an indoor space such as a corridor environment, if the disparity does not appear between matched straight lines, if the calculation is performed according to the proposed linear initialization method, appear. Because of this problem, it is desirable to apply inverse depth characteristics to the flare line coordinates for all straight lines. That is, an infinite straight line error does not occur even in a mismatch between straight lines that are matched by performing SLAM (Simultaneous Localization And Mapping) using the property of the inverse depth on the three-dimensional straight line defined by the three-dimensional spatial coordinates.

FIG. 5 is a diagram for explaining a method for updating a three-dimensional map according to an observation model adopted by embodiments of the present invention. In FIG. 5, To update the location information by using the location information.

Embodiments of the present invention create and record a spatial map while a moving object moves within a space. Therefore, it is necessary to re-measure the same space according to the movement of the moving object, or update the space map stored in advance by itself. Accordingly, it is preferable that the moving object according to the embodiments of the present invention compares the previously generated three-dimensional space coordinates with previously stored space coordinates, and selectively updates the feature information on the previously stored space coordinates according to the comparison result .

In particular, when the feature information of the generated three-dimensional spatial coordinates is present in the pre-stored spatial coordinates as a result of the comparison, both the straight line predicted using the plurality of cameras and both end points of the extracted straight line are used for the feature information And update the feature information of the previously stored spatial coordinates of the coordinates of the moving object, the float line coordinates, and the calculated updated value.

(3) Update update )

이고, 추정된 직선으로부터 추출된 직선의 양 끝점은

이다. 이로부터 이노베이션을 구하는 방법은 다음의 수학식 3과 같다.From the viewpoint of implementation, in FIG. 5, after calculating the innovation process and EKF-update (left Kalman filter update) for the left camera, the same calculation is performed for the right camera. For convenience of explanation, only the calculation process for the left camera will be described here. The predicted straight line of the left camera

, And both end points of the straight line extracted from the estimated straight line are

to be. The method of obtaining the innovation from this is shown in Equation (3).

The updating process according to the EKF is expressed by the following equation (4).

는 이노베이션 공분산(covariance)이고,

는 관측 모델에 대해 상태 변수(로봇 좌표 및 플뤼커 라인 좌표에 해당한다.)로 자코비안(Jacobian)을 산출한 값이고,

는 관측 모델에 대해 측정값으로 자코비안을 산출한 값이고,

는 갱신 전 상태 변수의 공분산이고,

은 이미지 픽셀 노이즈이고,

는 칼만 이득(Kalman gain)이고,

는 갱신된 상태 변수이고,

는 이노베이션이며,

는 갱신된 상태 변수의 공분산을 나타낸다.In Equation (4)

Is the innovation covariance,

Is a value obtained by calculating the Jacobian with the state variable (corresponding to the robot coordinate and the blur line coordinate) for the observation model,

Is a value obtained by calculating a Jacobian as a measurement value with respect to an observation model,

Is the covariance of the state variable before the update,

Is the image pixel noise,

Is the Kalman gain,

Is an updated state variable,

Is innovation,

Represents the covariance of the updated state variable.

FIG. 6 is a flowchart illustrating an entire process of applying a method for generating a map to a space according to an embodiment of the present invention to SLAM (Simultaneous Localization And Mapping) using a mobile object.

In step 610, the mobile unit attempts to recognize the position. For this, the left and right images are photographed through steps 621 and 622, respectively, using a stereo camera. The moving object extracts a straight line from the photographed image through steps 631 and 632, respectively. Then, the moving object generates three-dimensional spatial coordinates by utilizing both end points of the projected straight line on the image extracted in operation 640 and the position of the camera.

If the moving object continues to acquire an image while moving, it is determined whether or not the measured straight line projected on the image through the step 650 is a straight line that has already been obtained. As a result of the determination, if the straight line corresponds to the straight line that has already been obtained, the flow advances to step 670 to perform an update process in the EKF-SLAM process. On the other hand, if the straight line is not a straight line that has already been acquired, the 3D space coordinate generation process of step 660 is performed. This process corresponds to the linear initialization process of the EKF-SLAM process described above. In step 680, the moving object controls its own motion based on the estimated position and the generated map information.

FIG. 7 is a block diagram illustrating an apparatus for generating a map for a space using a moving object according to an embodiment of the present invention, wherein each element corresponds to each step described above with reference to FIG. 3, Focus on the enemy characteristics, but only outline the outline to avoid duplication of explanation.

The driving unit 10 serves to move the moving body 700 in the space. From the viewpoint of implementation, the driving unit 10 can be implemented by various moving means including a wheel or a joint based on the power transmitting means.

A plurality of cameras 20 are provided in the moving body 700 so as to acquire images having different viewing angles, and images of the surrounding space of the moving body are acquired. The camera 20 is preferably a stereo camera.

The processing unit 30 generates a three-dimensional map of the space using the acquired image, and may be implemented by at least one processor. The processing unit 30 extracts a straight line from the image obtained through the camera 20, generates three-dimensional spatial coordinates by matching the extracted straight lines, and outputs the generated three-dimensional spatial coordinates Based map. Thus, in addition to the processor described above, program code including a memory and a set of instructions necessary in an arithmetic operation can be utilized to process such arithmetic operations.

In particular, the processing unit 30 extracts depth information from the discrepancy generated through the matching of the extracted straight lines, and calculates a three-dimensional straight line by using the extracted depth information and the duality between the dual and global Furuker coordinates Dimensional spatial coordinates of the three-dimensional space. Here, the dual plane coordinates define the position of the camera based on pose information composed of quaternion values considering the current position of the moving body and the rotation of the moving body, and the position of the camera, which is extracted from the image obtained through the defined camera, And matching the end points of the straight lines.

On the other hand, in order to prevent the error when the duality is not detected from the acquired image, the processing unit 30 preferably performs the SLAM using the property of the inverse depth on the three-dimensional straight line defined by the three-dimensional space coordinates Do.

7 may further include an odometer 15 for measuring the traveling information of the moving object. If the straight line information extracted through the processing unit 30 is less than a preset threshold value The three-dimensional space coordinates can be generated by additionally using the travel information measured by using the traveling recorder 15.

The moving object 700 of FIG. 7 may further include a storage unit 40 for storing coordinates and characteristic information of the space in advance. The three-dimensional space coordinates generated by the processing unit 30 and the three- And the feature information on the previously stored spatial coordinates can be updated if the feature information of the generated three-dimensional spatial coordinates exists in the pre-stored spatial coordinates as a result of the comparison. Here, the processing unit 30 calculates an update value for the feature information by using the straight line predicted using the plurality of cameras 20 and the end point of the extracted straight line, and calculates the coordinates of the moving object, It is preferable to update the feature information of the previously stored space coordinates using the line coordinates and the calculated update value.

FIG. 8 and FIG. 9 illustrate experimental results of generating a three-dimensional map of a specific space using a moving object according to an exemplary embodiment of the present invention. In particular, FIG. 8 and FIG. 9 illustrate experimental results of a stereo float SLAM .

The experimental environment is as follows. The Pioneer P3-AT robot was used as a moving object, and a stereo camera was installed in the robot. The data received from the robot included an odometer of the robot and images obtained from the left and right cameras. The image is 320 × 240 pixels and the pixel noise 0.2 is added to the Matlab toolbox.

Experiments for performance evaluation were performed based on image data acquired from a corridor of a specific building as shown in [A] of FIG. 8. There were reflection phenomena caused by light and mirrors coming from a window in front, There were things such as fire extinguishers and bicycles. EKF-SLAM was performed using 3372 sheets of left and right images. When a single camera is used, a map can be generated only for the bottom straight line and the vertical straight line in generating the spatial map, whereas referring to [B] of FIG. 8 according to the present embodiment, since the stereo camera is applied in the map generation process, You can see that it is possible to map the straight line.

9 is an experiment for confirming the accuracy when applying the Furuker linear coordinate and the dual flutter matrix to a stereo camera. The actual distance value and the estimated distance value (Z) for a straight line end point in the three-dimensional space are compared Respectively. Referring to the comparison table of FIG. 9, as a result of applying the embodiment of the present invention, it is shown that the distance estimation is very similar to the actual value, and an accurate spatial map is generated.

The embodiments of the present invention described above generate three planes corresponding to the left and right viewing angles by using three reference points such as an end point of a straight line extracted from a Stereo image and an optical center of a camera, We used the fact that we can express it as one straight line. In order to apply a straight line with no disparity, the inverse depth property was applied to the straight line to perform the SLAM.

According to the embodiments described above, it is possible to improve the position recognition and the map generation performance by generating the three-dimensional spatial information using the straight line extracted from the image obtained through the plurality of cameras. In addition, By using the Fuller Patrix, it is possible to extract all the straight lines existing in the space other than the floor and the vertical direction, thereby reducing the errors related to the position and the map configuration of the moving object, It is possible.

Meanwhile, the present invention can be embodied in computer readable code on a computer readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored.

Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device and the like, and also a carrier wave (for example, transmission via the Internet) . In addition, the computer-readable recording medium may be distributed over network-connected computer systems so that computer readable codes can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present invention can be easily deduced by programmers skilled in the art to which the present invention belongs.

The present invention has been described above with reference to various embodiments. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the disclosed embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is defined by the appended claims rather than by the foregoing description, and all differences within the scope of equivalents thereof should be construed as being included in the present invention.

700: mobile body
10: Driving unit 15: Odometer
20: camera 30:
40: storage unit

Claims (10)

In a method for a mobile body to generate a map for a space,
Acquiring images of a peripheral space of the moving object using a plurality of cameras having different viewing angles;
Extracting a line from the acquired image;
Generating three-dimensional spatial coordinates by matching the extracted straight lines; And
And generating a map based on the generated three-dimensional spatial coordinates using a Plucker matrix. The method of claim 1,
Wherein the generating the three-dimensional spatial coordinates comprises:
Extracting depth information from a disparity generated through matching of the extracted straight lines; And
And defining a three-dimensional straight line using the extracted depth information and the duality between dual plucker coordinates and plucker line coordinates. 3. The method of claim 2,
The dual-
The position of the camera is defined based on pose information including a current position of the moving object and quaternion values considering the rotation of the moving object, and the straight line extracted from the image obtained through the defined camera. Generated by matching the endpoints together. The method of claim 1,
Wherein the step of generating a map based on the three-
Wherein SLAM (Simultaneous Localization And Mapping) is performed using a property of an inverse depth on a three-dimensional straight line defined through the three-dimensional spatial coordinates. The method of claim 1,
When the extracted linear information is less than a threshold, generating the three-dimensional spatial coordinates,
And using the driving information (odometry) of the moving object. The method of claim 1,
Comparing the generated three-dimensional space coordinates with previously stored space coordinates; And
And selectively updating feature information on the previously stored spatial coordinates according to the comparison result. The method according to claim 6,
If the feature information of the generated three-dimensional space coordinates is present in the pre-stored space coordinates as a result of the comparison,
Calculating an update value for the feature information by using a straight line predicted using the plurality of cameras and an end point of the extracted straight line; And
And updating feature information of the pre-stored spatial coordinates using the coordinates of the moving object, the flicker line coordinates and the calculated update value. The method of claim 1,
Wherein the plurality of cameras is a stereo camera consisting of left and right cameras. An apparatus for generating a map of a space using a moving object,
A driving unit for moving the moving object in the space;
A plurality of cameras provided on the mobile body to acquire images of a peripheral space of the mobile body and to have different viewing angles; And
And a processor for generating a three-dimensional map of the space using the acquired image,
Wherein,
And extracting a straight line from the obtained image, matching the extracted straight lines to generate three-dimensional spatial coordinates, and generating a map based on the generated three-dimensional spatial coordinates using a flicker matrix. The method of claim 9,
Wherein,
Depth information is extracted from the mismatch generated through the matching of the extracted straight lines, and the three-dimensional space coordinates are defined by defining a three-dimensional straight line by using the extracted depth information and the duality between the dual flicker coordinates and the flicker linear coordinates. Device for producing.

Images