KR101280392B1 - Apparatus for managing map of mobile robot based on slam and method thereof - Google Patents

Abstract

An apparatus and method for managing a map of a SLAM technology-based mobile robot according to the present invention are disclosed.
An apparatus for managing a map of a mobile robot according to the present invention includes processing means for extracting a landmark from an image obtained from a location where a mobile robot is located; Calculating means for calculating a probability value of the extracted landmark; And determining means for determining whether to use the landmark according to the calculated probability value of the landmark.
Through this, the present invention can not only accurately determine the presence of measurable landmarks, but also reduce the amount of computation of the system to the minimum and efficiently use the limited system resources in the application.

Description

Apparatus and method for managing map of mobile robot based on SLAM technology {APPARATUS FOR MANAGING MAP OF MOBILE ROBOT BASED ON SLAM AND METHOD THEREOF}

The present invention relates to a map management method, which calculates a probability of existence of a landmark with respect to existing landmarks added to a map and moves based on SLAM technology to determine whether to use a landmark based on the calculated probability. An apparatus and method for managing a map of a robot.

As demand for various applications of mobile robots increases, Simulaneous Localization And Mapping (SLAM), an essential technology for autonomous driving of mobile robots, is drawing attention. Slam technology refers to a technology that complements each other by locating and constructing a map, at the same time as a technology of recognizing a location or knowing a location and making a map with an existing map. This is a very similar way of exploring the unknown environment in our daily lives.

In such a slam technique, it is very important to properly manage the number of landmarks constituting the map. The problem of determining whether the measured value of the landmark extracted from the external sensor data is associated with any of the landmarks in the map is called a data association problem.

If the landmarks are too dense, the data associations between measurements and specific landmarks will be blurry, resulting in poor performance of the slam due to incorrect data associations. On the other hand, if the landmarks are too scattered, the robot will often calculate the position only as it appears on the odometer, without a marker to correct its position, increasing the uncertainty of position recognition.

In addition, as the number of landmarks increases, the amount of computation required in both parts of the slam increases.

First, in the data association, the number of landmarks to be compared with the measured values input through the sensor increases, so that the amount of computation required for the data association increases in proportion to the number of landmarks. Secondly, when the state of robots and maps are updated using the extended Kalman filter, the size of the state vector increases according to the number of landmarks, so that the amount of computation increases in proportion to the square of the number of landmarks.

As a result, the amount of computation increases greatly with the number of landmarks, which adversely affects the real-time performance of the system.

Accordingly, an object of the present invention is to calculate the probability that a landmark exists with respect to existing landmarks added to a map and determine whether to use the landmark based on the calculated probability. The present invention provides an apparatus and method for managing a map of a SLAM technology-based mobile robot.

However, the objects of the present invention are not limited to those mentioned above, and other objects not mentioned can be clearly understood by those skilled in the art from the following description.

In order to achieve the above objects, an apparatus for managing a map of a SLAM technology-based mobile robot according to an aspect of the present invention includes processing means for extracting a landmark from an image obtained from a location where the mobile robot is located; Calculating means for calculating a probability value of the extracted landmark; And determining means for determining whether to use the landmark according to the calculated probability value of the landmark.

Preferably, the processing means may binarize the acquired image to extract a landmark from the binarized image.

Preferably, the processing means may extract a boundary between the ceiling and the wall surface as a landmark from the obtained image.

Preferably, the calculation means may calculate a probability value that the extracted landmark exists by using a binary bayes filter.

Preferably, the calculation means may obtain a log-odd value of the probability that the extracted landmark exists using a binary base filter, and calculate a probability value of the extracted landmark based on the obtained log-odd value. .

에 의해 구하고, 여기서, 상기

는 랜드마크의 존재 여부를 나타내고, 상기 Z_1:t는 랜드마크에 대한 측정값을 나타내며, 상기 x_1:t는 로봇의 위치값을 나타내며, 상기 c_1:t는 대응 변수(correspondence variable)를 나타내며, 상기

는 i번째 랜드마크가 존재할 확률의 log-odd값을 나타낼 수 있다.Preferably, the probability that the landmark exists is the following equation

Obtained by the above, where

Denotes the presence or absence of a landmark, Z _{1: t} denotes a measurement value for the landmark, x _{1: t} denotes a position value of the robot, and c _{1: t} denotes a correspondence variable. Indicated above

May represent a log-odd value of the probability that the i-th landmark exists.

Preferably, the determining means may determine the use of the landmark if the calculated probability value is greater than a preset threshold.

Preferably, the determining means may determine to remove the landmark when the calculated probability value is smaller than a preset threshold.

According to another aspect of the present invention, a method for managing a map of a SLAM technology-based mobile robot includes: extracting a landmark from an image obtained from a location where a mobile robot is located; Calculating a probability value of the extracted landmark; And determining whether to use the landmark according to the calculated probability value of the landmark.

Preferably, the extracting may binarize the obtained image to extract a landmark from the binarized image.

Preferably, the extracting may extract the boundary line between the ceiling and the wall from the acquired image as a landmark.

Preferably, the calculating may calculate a probability value of the extracted landmark using a binary bayes filter.

Preferably, the calculating may include obtaining a log-odd value of the probability that the extracted landmark exists using a binary base filter, and calculating a probability value of the extracted landmark based on the obtained log-odd value. have.

에 의해 구하고, 여기서, 상기

는 랜드마크의 존재 여부를 나타내고, 상기 Z_{1 :t}는 랜드마크에 대한 측정값을 나타내며, 상기 x_1:t는 로봇의 위치값을 나타내며, 상기 c_1:t는 대응 변수(correspondence variable)를 나타내며, 상기

는 i번째 랜드마크가 존재할 확률의 log-odd값을 나타낼 수 있다.Preferably, the probability that the landmark exists is the following equation

Obtained by the above, where

Denotes the presence or absence of a landmark, Z _{1 : t} denotes a measured value for the landmark, x _{1: t} denotes a position value of the robot, and c _{1: t} denotes a corresponding variable. Indicated above

May represent a log-odd value of the probability that the i-th landmark exists.

Preferably, the determining may determine the use of the landmark if the calculated probability value is greater than a preset threshold.

Preferably, the determining may determine removal of the landmark when the calculated probability value is smaller than a preset threshold.

In this way, the present invention calculates the probability that the landmark exists for the existing landmarks added to the map, and determine whether or not to use the landmark based on the calculated probability, the presence of a measurable landmark accurately There is an effect that can be determined.

In addition, the present invention is to calculate the probability of the landmark exists for the existing landmarks added to the map and determine whether to use the landmark based on the calculated probability, thereby reducing the amount of computation of the system to a minimum There is.

In addition, the present invention calculates the probability that the landmark exists for the existing landmarks added to the map and determine whether to use the landmark based on the calculated probability, thereby efficiently using the limited system resources in the application. It has an effect.

1 is an exemplary view showing an apparatus for managing a map according to an embodiment of the present invention.
2 is an exemplary view for explaining a principle of extracting a landmark according to the present invention.
3 is an exemplary view showing a method for managing a map of a mobile robot according to an embodiment of the present invention.
4 is an exemplary diagram illustrating a detailed method of calculating a probability value illustrated in FIG. 3.
5 is an exemplary view showing a simulation result of an algorithm according to an embodiment of the present invention.

Hereinafter, an apparatus and method for managing a map of a SLAM technology-based mobile robot according to an embodiment of the present invention will be described with reference to FIGS. 1 to 5. The present invention will be described in detail with reference to the portions necessary for understanding the operation and operation according to the present invention. Like reference numerals in the drawings denote like elements throughout the specification.

In particular, the present invention proposes a method of calculating the probability of the presence of a landmark with respect to existing landmarks added to the map and determining whether to use the landmark based on the calculated probability. Herein, the present invention may mathematically calculate the probability of a landmark through a binary bayes filter.

1 is an exemplary view showing an apparatus for managing a map according to an embodiment of the present invention.

As shown in FIG. 1, an apparatus for creating a map of a mobile robot according to the present invention includes an acquisition means 110, a processing means 120, a calculation means 130, a determination means 140, and a storage means ( 150) and the like.

Acquisition means 110 is an apparatus for acquiring an image or an image of the surrounding environment in which the mobile robot is located, for example, may include an image sensor such as a charge coupled device (CCD), a complementary metal oxide semiconductor (CMOS) and the like. . In particular, since the present invention intends to use the boundary line between the ceiling and the wall as a line landmark or landmark in an indoor space, the acquiring means 110 may be installed to acquire an image of the ceiling. .

The processor 120 may extract a landmark from the acquired image. This will be described with reference to FIG. 2.

2 is an exemplary view for explaining a principle of extracting a landmark according to the present invention.

As shown in Fig. 2, the image (a) shows the acquired image. In Figure (b), the obtained image is binarized to generate a binary image. Binarization involves comparing the original value of each pixel of the image with a certain threshold. If the threshold is exceeded, the binary value for this pixel may be set to the first value, for example black, otherwise it may be set to the second value, for example white.

In Figure (c), the boundary line between the ceiling and the wall surface is extracted from the binarized image as a landmark. That is, one value of pixels, eg all black pixels, may represent a wall, while another value of pixels, eg all white pixels, may represent a ceiling.

The calculation unit 130 may determine whether the extracted landmark is an existing landmark or a new landmark previously added to the map, and calculate a probability value that the extracted landmark exists. This will be described in more detail as follows.

The probability value of such an i-th landmark may be expressed as shown in Equation 1 below.

[Equation 1]

는 랜드마크의 존재 여부를 나타낼 수 있다. 예컨대,

가 1이면 랜드마크가 존재하는 것이고,

가 0이면 랜드마크가 존재하지 않는 것이다.here,

May indicate whether a landmark exists. for example,

Is 1, the landmark exists,

Is 0, no landmark exists.

Z _{1: t} is a measurement for the landmark and may represent a measurement at time 1 to t. x _{1: t} may represent a position value of the robot from time 1 to t in terms of a position value, for example, coordinates and directions. c _{1: t} may represent a correspondence variable from time 1 to t. If c _t = i, it means that z _t measured the i-th landmark on the map.

At this time, the meaning of the landmark should be considered from the robot's point of view. In other words, even though the landmark is actually present, if the boundary line, i.e., the line segment is not extracted from the image, it does not exist from the robot's point of view. Therefore, the probability value of the presence of the landmark in [Equation 1] described above means the probability value of the landmark to be found by the robot.

으로 변환하면 변환된 log-odd 값

은 다음의 [수학식 2]와 같이 나타낼 수 있다.log-odd value for the probability that the i-th landmark exists using a binary base filter

Log-odd value when converted to

Can be expressed as Equation 2 below.

&Quot; (2) "

는 이전 시간까지 축적한 log-odd 값을 나타내고,

는 아무런 측정없이 추측한 초기 log-odd 값을 나타내며,

는 시간 상 가장 최근의 log-odd 값을 나타낼 수 있다. 이때,

는 특별한 사전 지식이 없는 한 보통 0이 된다.here,

Represents the log-odd value accumulated up to the previous time,

Represents the initial log-odd value estimated without any measurement,

May represent the most recent log-odd value in time. At this time,

Is usually 0 unless there is special prior knowledge.

를 역변환하면 다음의 [수학식 3]과 같이 나타낼 수 있다.Back of Equation 2

When inversely transformed, it can be expressed as Equation 3 below.

&Quot; (3) "

만 알면

를 알 수 있게 되고, 다시 [수학식 3]에서

를 알면 i번째 랜드마크가 존재할 확률값을 알 수 있게 된다.According to these log-odd values, in [Equation 2]

If you know

Can be known, and again in [Equation 3]

By knowing the probability value that the i-th landmark exists.

는 업데이트되는 값이기 때문에 정확하게 계산하는 일은 매우 어렵다. 그래서 본 발명에서는 다음의 세 가지 경우 즉, 랜드마크가 발견되는 경우, 랜드마크가 FoV(Field of View) 내에 있으나 발견되지 않는 경우, 랜드마크가 FoV 밖에 있어 발견되지 않는 경우로 나누어 각 경우에 해당하는 고정값을 할당하고자 한다.At this time,

Is an updated value, so it is very difficult to calculate exactly. Therefore, in the present invention, the following three cases, that is, when the landmark is found, the landmark is in the field of view (FoV), but not found, the landmark is out of the FoV and is found in each case We want to assign a fixed value.

는 다음의 [수학식 4]와 같이 정의할 수 있다.1) Update value when a landmark is found

Can be defined as in Equation 4 below.

&Quot; (4) "

는 다음의 [수학식 5]와 같이 정의할 수 있다.2) Update value when the landmark is in the FoV but is not found.

May be defined as in Equation 5 below.

&Quot; (5) "

은 0이다.3) Update value when the landmark is not found because it is outside the FoV.

Is zero.

를 제외한

와

는 상황에 따라 각 값이 다르게 할당될 수 있다.At this time, it has a value of 0

excluding

Wow

Each value can be assigned differently depending on the situation.

=

가 되어 앞의 [수학식 3]에 의해 i번째 랜드마크가 존재할 확률값

은 0.5가 된다.For example, if it is determined that the extracted landmark is a new landmark,

=

And the probability value that the i th landmark exists by [Equation 3]

Becomes 0.5.

The determination unit 140 may compare the probability value of the calculated landmark with the preset threshold and determine whether to use the landmark according to the comparison result. That is, the determination means 140 removes only the landmarks having a high probability value and the landmarks having a low probability.

At this time, the determination means 140 adds a new landmark on the map. The determination unit 140 determines that the landmark is a well-found landmark when the probability value calculated in the case of the existing landmark is larger than the preset threshold value and leaves the landmark on the map, and the calculated probability value is larger than the preset threshold value. If it is small, it is determined that the landmark is not easily found, and the landmark is not removed from the map.

The storage means 150 may store maps and landmarks created by the mobile robot. Landmarks stored in the storage means 150 may be added or deleted through an update.

3 is an exemplary view showing a method for managing a map of a mobile robot according to an embodiment of the present invention.

As shown in FIG. 3, the map management apparatus according to the present invention may acquire an image of a ceiling on which a mobile robot is located (S310) and extract a landmark from the acquired image (S320).

At this time, the map management apparatus may binarize the acquired image and extract the boundary line or line segment of the ceiling and the wall from the binarized image as a landmark.

Next, the map management apparatus may calculate a probability value that the extracted landmark exists. That is, the map management apparatus determines whether the extracted landmark is an existing landmark or a new landmark added to the map, and based on the result of the determination, the map management apparatus determines a probability value of the extracted landmark using the binary base filter. You can calculate.

Next, the map management apparatus may determine whether to use the landmark according to the calculated probability value (S340). That is, the map management apparatus may check whether the calculated probability value is greater than a preset threshold. The map management device determines that the landmark is found if the probability value calculated according to the result is greater than the preset threshold value, and determines the use of the landmark. If the calculated probability value is less than the preset threshold value, the map management device is not found. It is determined that the landmark is removed to determine the landmark.

As such, the present invention manages the map by confirming whether the line segment landmarks or landmarks constituting the map constructed through the SLAM technology are strongly and repeatedly found, leaving only the highly repeatable landmarks and removing the landmarks that are not. Through this, the present invention can reduce the amount of computation to a minimum while maintaining the performance of the SLAM.

4 is an exemplary diagram illustrating a detailed method of calculating a probability value illustrated in FIG. 3.

As illustrated in FIG. 4, the map management apparatus according to the present invention may determine whether the extracted landmark is an existing landmark or a new landmark added on the map (S331).

Next, the map management apparatus may acquire a log-odd value of the probability that the extracted landmark exists using the binary base filter according to the determined result (S332). In this case, the log-odd value may have different values depending on 1) when the landmark is found, 2) when the landmark is in the FoV but not found, and 3) when the landmark is outside the FoV.

This log-odd value may be obtained by the above Equation 4 to Equation 5.

Next, the map management apparatus may calculate a probability value of the extracted landmark based on the log-odd value of the probability value of the acquired landmark. That is, the probability value of the extracted landmark exists may be calculated by Equation 3.

5 is an exemplary view showing a simulation result of an algorithm according to an embodiment of the present invention.

As shown in FIG. 5, the simulation result using the algorithm according to the present invention shows the test environment and the trajectory of the mobile robot. Here, it can be seen that the number of final landmarks is 26 and the number of removed landmarks is 232.

The remaining landmarks on the final map were detected 91 times in 2890 frames each.

Various modifications and variations may be made by those skilled in the art to manage the map of the SLAM technology-based mobile robot according to the present invention, and those skilled in the art. Therefore, the embodiments disclosed in the present invention are intended to illustrate rather than limit the scope of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of the present invention.

110: acquisition means
120: processing means
130: calculation means
140: means of determination
150: storage means

Claims (16)

Processing means for extracting a landmark from an image acquired from a location where the mobile robot is located;
Calculating means for calculating a probability value of the extracted landmark; And
Determining means for determining whether to use the landmark according to the calculated probability value of the landmark;
It includes, wherein the calculation means is to obtain a log-odd value of the probability that the extracted landmark exists using a binary base filter, and calculates the probability of existence of the extracted landmark based on the obtained log-odd value Apparatus for managing the map of the SLAM technology-based mobile robot, characterized in that. The method according to claim 1,
The processing means,
And binarizing the acquired image to extract a landmark from the binarized image. The method of claim 2,
The processing means,
Apparatus for managing the map of the SLAM technology-based mobile robot, characterized in that for extracting the boundary line between the ceiling and the wall from the obtained image as a landmark. delete delete The method according to claim 1,
The probability value that the landmark exists,
The following equation

Obtained by the above, where

Denotes the presence or absence of a landmark, Z _{1: t} denotes a measurement value for the landmark, x _{1: t} denotes a position value of the robot, and c _{1: t} denotes a correspondence variable. Indicated above

Is a log-odd value of the probability that the i-th landmark exists. The method according to claim 1,
The determining means,
And determining the use of the landmark when the calculated probability value is greater than a preset threshold. The method according to claim 1,
The determining means,
And determining the removal of the landmark when the calculated probability value is smaller than a predetermined threshold value. Extracting a landmark from an image acquired from a location where the mobile robot is located;
Calculating a probability value of the extracted landmark; And
Determining whether to use the landmark according to the calculated probability of the landmark being present;
The calculation may include calculating a log-odd value of a probability that the extracted landmark exists using a binary base filter, and calculating a probability value of the extracted landmark based on the obtained log-odd value. Method for managing the map of the SLAM technology-based mobile robot, characterized in that. 10. The method of claim 9,
Wherein the extracting comprises:
And binarizing the acquired image to extract a landmark from the binarized image. The method of claim 10,
Wherein the extracting comprises:
And a boundary line between a ceiling and a wall surface is extracted from the acquired image as a landmark. delete delete 10. The method of claim 9,
The probability value that the landmark exists,
The following equation

Obtained by the above, where

Denotes the presence or absence of a landmark, Z _{1: t} denotes a measurement value for the landmark, x _{1: t} denotes a position value of the robot, and c _{1: t} denotes a correspondence variable. Indicated above

Is a log-odd value of the probability that the i-th landmark exists. 10. The method of claim 9,
Wherein the determining comprises:
And determining the use of the landmark if the calculated probability is greater than a preset threshold. 10. The method of claim 9,
Wherein the determining comprises:
And if the calculated probability value is less than a predetermined threshold value, the removal of the landmark is determined.

Images