KR20150068633A - Method for estimating location of mobile device, and apparatus thereof - Google Patents

Abstract

Disclosed is a method for predicting the location of a mobile device. According to an embodiment of the present invention, the invention comprises the steps of: grouping particles, which are distributed around a feature point, into a predetermined number of areas based on a particle center point; saving, in a memory, at least one particle having a weight value within a predetermined range among particles positioned within the grouped areas; replicating, except for the stored particles among the particles within the grouped area, the remaining particles by n number, respectively; and predicting a position of the mobile device based on at least one particle which is saved, and the replicated particles.

Description

Field of the Invention [0001] The present invention relates to a method and apparatus for predicting a location of a mobile device,

The following embodiments relate to a method and apparatus for predicting the position of a mobile device, and more particularly, to a method and apparatus for more accurately predicting a position of a mobile device through re-extraction of particles in an unknown environment map .

The mobile device can acquire the location information of the feature point by searching for the landmark through the sensor during the movement. However, the mobile device can not acquire the exact position of the feature point, and can obtain the probability information that the feature point is in a certain position, and the probability information is called a particle.

Simultaneous Localization and Mapping (SLAM) is a technique for making a map for an unknown environment while the robot is moving and automatically performing position estimation. SLAM is indispensable in the robot autonomous navigation field. The SLAM is performed in a similar manner to the localization, but the location information of the feature point, which is sensor specific information for identifying the detected position and attitude information, and the identification information are not given in advance. The location information and the identification information of the minutiae are different from the localization in that the robot is generated while traveling and is estimated.

There are motion control noise and environmental sensor noise in the course of robot movement, and the estimation error of the predicted robot becomes large, and the efficiency of SLAM becomes low. In order to solve the problem of SLAM, the position of the robot is estimated by using extended Kalman filter-SLAM (EKF-SLAM) and particle filter based on Kalman filter and Kalman filter And SLAM (Factored Solution to the SLAM, FastSLAM) for estimating the location of the minutiae. In particular, FastSLAM increases the computational complexity in proportion to the number of minutiae points, and it can store different estimates for each particle and compensate between estimates. In addition, since FastSLAM uses particle filter and Kalman filter at the same time, it has robust characteristics for an inaccurate robot model. However, FastSLAM can lose the diversity of particles because the particles having a low probability weight are deleted and the position information of the deleted particles is also deleted in the process of re-extracting the particles. That is, in FastSLAM, a problem of particle depletion occurs.

Even if the weight of the particle is low, the particle with low weight can not be removed in the re-extraction process, and the problem of particle depletion occurring in the conventional particle re-extraction technique can be solved.

Particles with low weights can survive the re-extraction process, thus maintaining the variety of particles. In addition, various particles are retained, and the number of surviving particles in the reextraction process increases, so that the position of the mobile device can be more accurately predicted through the surviving particles.

A method for predicting a position of a mobile device according to one side includes: grouping a plurality of particles distributed around a minutiae point into a predetermined number of regions based on a center point of the particle; Storing in the memory at least one particle having a weight within a predetermined range among the particles located within the grouped region; Duplicating the remaining particles except for the stored particles among the particles in the grouped region by n; And predicting a location of the mobile device based on the stored at least one particle and the replicated particles.

Also, n is calculated based on the number of remaining particles and the weight of the remaining particles.

The method of claim 1, further comprising: determining whether the remaining particles are located in a protected area among the particles located within the grouped area, and when it is determined that the remaining particles are located outside the protected area, And weights are given to replicate the remaining particles.

A method for predicting a position of a mobile device according to another side includes: recording a plurality of particles observed at a feature point and weighting the particles; Grouping the particles into a predetermined number of groups; Setting a protected area within the group; Constructing a set of particles based on at least one particle in the group, the weight being within a predetermined range, at least one particle located within the protected area, and at least one particle located outside the protected area; And predicting a position of the mobile device based on the set of particles.

In addition, the step of setting the protected area in the group sets the protected area based on the area center point and a preset protective radius, and the area center point is calculated based on the coordinates of the particles in the group .

An apparatus for predicting a position of a mobile device according to one side groups a plurality of particles distributed around a minutiae point into a predetermined number of regions with respect to a center point of a particle and calculates a weight within a predetermined range among the particles in the grouped region A processor for duplicating the remaining particles except for the at least one particle having n; A memory for storing at least one particle having a weight within the predetermined range; And a predictor for predicting a position of the mobile device based on the stored at least one particle and the copied particles.

In addition, when the processor determines that the remaining particles are outside the protected area, the processor replaces the remaining particles by assigning an addition weight to the n, and when determining that the remaining particles are within the protected area, When the number of particles is less than the predetermined number, n is multiplied by a predetermined number to replicate particles located in the protected area.

An apparatus for predicting a position of a mobile device according to another aspect of the present invention includes an update unit for recording a plurality of particles observed at a feature point and assigning weights to the particles; Grouping the particles into a predetermined number of groups and setting a protected area within the group, wherein at least one particle within the group has a weight within a predetermined range, at least one particle located within the protected area, and A re-extractor configured to form a particle set based on at least one particle located outside the protected area; And a predictor for predicting a position of the mobile device based on the set of particles.

Compared with SIR and PSR, which are re-extracting techniques used in the existing FastSLAM, the re-extraction technique used in the present invention can maintain the diversity of particles because the number of surviving particles is large.

Also, the re-extraction technique used in the present invention can reduce the error of the position of the mobile device according to the number of particles as compared with the existing particle re-extraction technique, and can improve the accuracy of the position prediction of the mobile device.

Fig. 1 is a diagram for explaining how, in a map, a mobile device detects a feature point and predicts its position.
FIG. 2 is a diagram for explaining a repetitive importance re-extraction technique generally used in FastSLAM.
3 is a view for explaining a technique of re-extracting particles using information of locally distributed particles.
4 is a flowchart illustrating a method of predicting a location of a mobile device according to an embodiment of the present invention.
5 is a view for explaining storage of particles having a weight within a predetermined range in a method of predicting the position of a mobile device according to an embodiment of the present invention.
6 is a diagram for explaining a method of predicting a position of a mobile device according to another embodiment of the present invention.
7 is a block diagram for explaining a configuration of an apparatus for predicting a position of a mobile device according to an embodiment of the present invention.
8 is a block diagram for explaining a configuration of an apparatus for predicting a position of a mobile device according to another embodiment of the present invention.
9 is a view for explaining performance comparison between a mobile device according to the present invention and a mobile device to which the existing re-extraction technique is applied.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to or limited by the embodiments. In addition, the same reference numerals shown in the drawings denote the same members.

In one embodiment, the device for predicting the location of a mobile device according to the present invention may be embodied in a standalone device or embedded in a mobile device. Here, the mobile device is a device that receives power through a power source and moves itself, and may include a mobile robot.

Fig. 1 is a diagram for explaining how a mobile device searches for a feature point in a map and predicts its own position.

Referring to FIG. 1, the mobile device 110 may detect a plurality of minutiae 120 through 127 through a sensor during movement. Here, the sensor may include a laser sensor, a visual sensor, and the like. The mobile device 110 may store the location information of the minutiae 120-127 every time it detects the minutiae 120-127. For example, the mobile device 110 may measure a distance to a feature point through a sensor.

In the simulation environment, a plurality of particles may be distributed around the feature points 120 to 127 sensed by the sensor. Through the particles distributed around the feature points 120 to 127, the position of the feature points 120 to 127 can be predicted.

Also, in a simulated environment, the mobile device 110 may distribute a plurality of particles around itself. Through the distributed particles around the self, the mobile device 110 can predict its position.

The mobile device 110 may sense the minutiae 120-127, move the particles around the minutiae 120-127, and distribute the particles around themselves, moving along the circle path. At the initial stage of the movement, the mobile device 120 does not have many location information, so that there may be a large error from a given route. However, through the re-extraction process of the particles, the mobile device 120 can move while predicting its position more accurately.

The mobile device 110 may obtain encoder information during movement. The encoder information may include the number of revolutions of the mobile device 110, and the mobile device 110 may know the distance that the mobile device 110 has moved through the number of revolutions of the wheel. The mobile device 110 may estimate its location using encoder information and particles.

FIG. 2 is a diagram for explaining a repetitive importance re-extraction technique generally used in FastSLAM.

와 맵 정보를 나타내는

를 기록한다. 맵 정보는

, 및

를 포함하고,

은 시간 t에서 관찰된 n개의 특징점들을 나타내고,

는 특징점의 수를 나타낸다. 기록된 이동 로봇의 위치, 및 맵 정보를 통해, 이동 로봇의 위치와 지도의 위치는 베이시안(Bayesian) 법칙에 의해

으로 나타낼 수 있다. 여기서,

는 관측(또는 센서정보),

는 로봇 제어 입력,

는 맵 정보를 나타내는

와

사이의 관계를 나타낸다.

은 맵 정보와 이동 로봇의 위치의 경험적 확률(posterior) 밀도를 추정하기 위해 필요하다.

은 베이시안 필터링(Basyesian filtering)을 통해 수학식 1로 표현될 수 있다.FastSLAM indicates the position of the mobile robot at time t

And map information

Lt; / RTI > The map information

, And

Lt; / RTI >

Represents the n feature points observed at time t,

Represents the number of feature points. The position of the mobile robot and the position of the map are recorded by the Bayesian rule through the recorded position of the mobile robot and the map information

. here,

(Or sensor information),

A robot control input,

Indicates map information

Wow

Lt; / RTI >

Is necessary to estimate the empirical posterior density of the map information and the position of the mobile robot.

Can be expressed by Equation (1) through Basyesian filtering.

[Equation 1]

When the posterior of the position of the robot is known, Equation (1) can be factorized as Equation (2).

&Quot; (2) "

에 대해 분리된 추정함수를 사용한다. 인수분해로 인하여, FastSLAM은 각 특징점마다 분리된 EKF를 유지할 수 있고, EKF-SLAM보다 더 효율적으로 갱신된 추정값을 얻을 수 있다. 시간 t에서, m번째 파티클 정보

은 수학식 3과 같이 나타낼 수 있다.For each feature point in the map, FastSLAM uses the location of the particle

Using a separate estimation function. Due to the factorization, FastSLAM can maintain a separate EKF for each feature point, and can obtain updated estimates more efficiently than EKF-SLAM. At time t, the mth particle information

Can be expressed by Equation (3).

&Quot; (3) "

은 [m]번째 파티클의 경로 추정치,

, 및

는 각각 파티클

와 관련된 특징점 위치를 나타내는 가우시안(gaussian) 평균과 공분산을 나타낸다.Here, [m] is the index of the particle,

Is the path estimate of the [m] th particle,

, And

Respectively,

And a gaussian mean and a covariance representing the location of the minutiae associated with the target.

이 기록된다. 측정데이터 갱신 단계에서, 특징점에서 관찰된 파티클들이 새로운 파티클들이 아닌 경우, 관찰된 파티클들 각각에 대한 평균

, 및 공분산

을 수행한다. 가중치부여 단계에서, 각각의 파티클들에 대한 가중치

가 계산되며,

는 수학식 4와 같이 나타낼 수 있다.The FastSLAM repeatedly performs four steps of sampling, updating measurement data, weighting, and re-sampling. In the extraction step, when the mobile robot moves, the particles observed at the feature point measured by the sensor

Is recorded. In the measurement data update step, when the particles observed at the feature point are not new particles, the average

, And covariance

. In the weighting step, the weight for each particle

Lt; / RTI >

Can be expressed by Equation (4).

&Quot; (4) "

Here, the target distribution is the actual probability distribution, and the proposal distribution is the estimated probability distribution.

When each particle is weighted, particles with probabilities proportional to the weight are extracted, and the extracted particles are duplicated.

은 재추출 단계에서 3배가 복제되고, 낮은 가중치를 부여받은 [m+m]번째 파티클

은 재추출 단계에서 자신과 동일한 개수가 복제된다. 또한, 일부 파티클들은 가중치가 낮게 부여되어 완전히 제거된다. 파티클들이 복제되고, 제거되는 과정이 반복되어, 파티클들이 기록하는 이동 로봇 경로, 및 특징점의 추정 값이 상실될 수 있다. 따라서, 시간이 지남에 따라 독특한 특성을 가진 파티클들의 수가 줄어드는 파티클 고갈(particle depletion) 문제가 발생하여, 파티클 다양성(particle diversity)이 상실될 수 있다.
Referring to FIG. 2, particles having a target distribution larger than the proposed distribution are given high weights, and particles having a target distribution smaller than the proposed distribution are given low weights. In Fig. 2, when the step is k-1, the [m] th particle

Is multiplied three times in the re-extraction step, and the [m + m] -th particles

The same number as that of itself is copied in the re-extraction step. In addition, some particles are given a low weight and are completely removed. The processes of copying and removing the particles are repeated so that the estimated values of the mobile robot path and the feature point recorded by the particles may be lost. Thus, over time, particle depletion problems may occur that reduce the number of particles with unique characteristics, which may result in loss of particle diversity.

3 is a view for explaining a technique of re-extracting particles using information of locally distributed particles.

In order to solve the problem of particle diversity loss occurring in FastSLAM, the present invention uses information of locally distributed particles.

Referring to FIG. 3 (a), through the center coordinates of the particles, the region where the particles are distributed can be divided into K regions (for example, four regions). Within each region, there are particles with different weights. Particles having a weight within a predetermined range among the weights of the particles may be excluded from the re-extraction process (310). Particles having weights within a predetermined range are used to predict the position of the mobile robot.

Referring to FIG. 3 (b), particles having a weight within a predetermined range are excluded. Particles with weights within a predetermined range can be stored in memory.

Referring to FIG. 3 (c), some of the particles in the region are within the protected area 320. Through the position coordinates of the particles present in each region, the position of the local center point can be calculated. Since the region where the particles are distributed is divided into K regions, the positions of the K regional center points can be calculated. Within each region, particles that are within a certain distance from the local center point are protected during re-extraction. That is, the protected area 320 may be set within a certain range from the area center point. Since the particles present in the protected area 320 are not removed in the re-extraction process, various kinds of particles can survive.

Referring to FIG. 3 (d), the particles located in the protected area 320 are not removed but are copied. Some of the particles located outside the protected area 320 are replicated and some are removed. Particles located outside the protected area 320 may be given an addition weight. With particle re-extraction techniques, particles can be distributed over a large area and particle diversity can be maintained.

4 is a flowchart illustrating a method of predicting a location of a mobile device according to an embodiment of the present invention.

The method for predicting the position of the mobile device according to an exemplary embodiment may use the particle re-extraction technique described in FIG.

Referring to FIG. 4, in a method of predicting a position of a mobile device according to an exemplary embodiment, a plurality of particles distributed around a feature point may be grouped into a predetermined number of regions based on a center point of a particle (410). The principle of the K-Means algorithm, which is a non-hierarchical clustering method, can be used to divide the particles distributed in various regions into a certain group. The K-Means algorithm pre-assigns the number of clusters and assigns the objects to the appropriate clusters. If n objects are given, the K-Means algorithm clusters the objects until the given object is included in each of the specified clusters. The K-means algorithm can be applied to various types of data and can cluster data at high speed.

를 계산하기 위해, 이동 기기가 이동 중에 파티클들의 위치를 측정하는 경우

가 양수인지 음수인지 고려해야 한다. 보다 구체적으로, 측정된 파티클들의

가 180도와 -180도 일 때, 평균을 구하면 180도가 되지만 단순히 더하면 0도가 된다. 따라서, 각도에 따라 좌표 값을 별개로 저장하여, +영역의 값과 - 영역의 값의 차를 더한 평균좌표가 파티클 중심점으로 계산될 수 있으며, 수학식 5와 같다.In order to cluster a plurality of particles distributed around a feature point into K regions, the particle center point of a plurality of particles distributed around the feature point should be calculated. Particle center point

, The mobile device measures the position of the particles during movement

Is positive or negative. More specifically, the measured particles

Is 180 degrees and -180 degrees, the average is 180 degrees but merely 0 degrees. Accordingly, the coordinate values are separately stored according to the angle, and the average coordinate obtained by adding the difference between the value of the + area and the value of the area can be calculated as the center point of the particle.

&Quot; (5) "

는 파티클 중심점의 좌표,

, 및

은 각각

가 0보다 크거나 같을 때의 x좌표, 및 y좌표,

, 및

는 각각

가 0보다 작을 때의 x좌표, 및 y좌표를 나타낸다.

은

가 0보다 크거나 같을 때의 가중치, 및

는

가 0보다 작을 때의 가중치를 나타낸다. 파티클들의 좌표 값의 합과 가중치의 합의 평균을 이용하여 파티클 중심점이 계산될 수 있다. 또한, K지역의 지역 중심점

는 K지역 내에 위치한 파티클들의 좌표 값의 합과 가중치 합의 평균을 이용하여 계산될 수 있다. 또한, 파티클 중심점과 지역 중심점간의 거리

가 계산될 수 있으며, 이는 수학식 6과 같다.here,

The coordinates of the particle center point,

, And

Respectively

Is greater than or equal to zero, and the y coordinate,

, And

Respectively

The x coordinate, and the y coordinate when x is less than zero.

silver

Is greater than or equal to zero, and

The

Is less than zero. The particle center point can be calculated using the average of the sum of the coordinates of the particles and the sum of the weights. Also, the regional center point of K region

Can be calculated using the average of the sum of the coordinates of the particles located in the K region and the sum of the weights. Further, the distance between the center point of the particle and the center point of the region

Can be calculated, which is shown in Equation (6).

&Quot; (6) "

는 파티클 중심점,

는 K지역에 위치한 파티클들의 좌표이다.

는 파티클 중심점과 지역 중심점 사이의 거리로, 유클라디안 거리(Euclidean distance)이며, 수학식 7을 통해 구할 수 있다.here,

The particle center point,

Is the coordinates of the particles located in the K region.

Is the distance between the particle center point and the local center point, and is the Euclidean distance, which can be obtained from equation (7).

&Quot; (7) "

The method of predicting the position of a mobile device according to an exemplary embodiment may store 420 at least one particle having a weight within a predetermined range among the particles located within a grouped region. Will be described in detail with reference to FIG.

5 is a view for explaining storage of particles having a weight within a predetermined range in a method of predicting the position of a mobile device according to an embodiment of the present invention.

Once the local center point of each region is determined, particles located at a predetermined distance from the particle center point can be grouped, and each group is assigned a group number. Each group may have at least one particle with a weight within a predetermined range.

이며, 하한은

이다. 미리 정해진 범위 내의 가중치를 갖는 적어도 하나의 파티클은 재추출 과정에서 제외될 수 있다.

, 및

을 구하는 방법은 수학식 8과 같다.Referring to FIG. 5, the weights of the particles in the group may be sorted in descending order. According to one embodiment, the weights w3 to w6 are weights within a predetermined range, and the upper limit of the predetermined range is

And the lower limit is

to be. At least one particle having a weight within a predetermined range may be excluded from the re-extraction process.

, And

Is expressed by Equation (8).

&Quot; (8) "

는 파티클의 개수를 나타내고,

는 사용자의 입력값을 나타내며,

는 0.1 내지 0.9의 값일 수 있다. 미리 정해진 범위 내의 가중치를 갖는 적어도 하나의 파티클은 복제과정을 거친 후,

에 저장될 수 있다.here,

Represents the number of particles,

Represents the input value of the user,

May be a value of 0.1 to 0.9. At least one particle having a weight within a predetermined range is subjected to a replication process,

Lt; / RTI >

The weights w1, w2, w7, and w8 outside the predetermined range are the remaining particle groups (M).

Referring again to FIG. 4, a method for predicting the location of a mobile device according to one embodiment may replicate (420) as many as n remaining particles except those stored in the grouped regions. Here, n can be calculated based on the number of remaining particles and the weight of the remaining particles, as shown in Equation (9).

&Quot; (9) "

은 n을 나타내고, M은 나머지 파티클들의 개수를 나타내며,

는 가중치를 나타낸다.

Represents n, M represents the number of remaining particles,

Represents a weight.

A method for predicting the location of a mobile device according to an embodiment may predict the location of the mobile device based on the stored at least one particle and the replicated particles (440).

를 계산할 수 있다. 상기 보호 지역은 수학식 6으로부터 계산된

, 및

를 기초로 설정될 수 있다. 보호 지역에 존재하는 파티클들 중 가중치가 작게 부여된 파티클은 파티클 다양성 유지를 위해 제거되지 않는다. 파티클이 보호 지역에 존재하는지는 수학식 10을 통해 알 수 있다.According to one embodiment, a method for predicting a location of a mobile device may determine whether the remaining particles among the particles located within the grouped region are in a protected area. To record the distribution of particles among the remaining particles except those with a predetermined range of weights

Can be calculated. The protection area is calculated from Equation (6)

, And

As shown in FIG. Particles that are assigned a small weight in the protected area are not removed to maintain particle diversity. It can be seen from equation (10) whether the particle exists in the protected area.

&Quot; (10) "

와 K지역에 위치하는 파티클의 좌표인

간의 거리가

와

를 곱한 값보다 작은 거리에 존재하는 파티클은 보호 지역에 존재하는 것으로 판단된다.Area center point

And the coordinates of the particle located in the K region

Distance between

Wow

It is judged that the particles existing at a distance smaller than the value obtained by multiplying the number of particles are present in the protected area.

According to one embodiment, if the remaining particles are determined to be in the protected area and n is less than the predetermined number, then the predetermined number of n may be added to thereby replicate the patacles located within the protected area. For example, if n of the particles in the protected area is greater than 1, then that particle is replicated by n, and if n in the protected area is less than 1, that particle may be replicated by n + 1. This is expressed by Equation (11).

&Quot; (11) "

은 수학식 12를 통해 계산될 수 있다.According to one embodiment, when it is determined that the remaining particles are out of the protected area, an addition weight is given to n so that the remaining particles can be replicated. The addition weight is the regional equalization weight described in FIG. Weighted

Can be calculated through Equation (12).

&Quot; (12) "

는 K지역에 위치한 파티클들의 개수를 나타내고, M은 K지역에 위치한 파티클들 중에서 미리 정해진 범위의 가중치를 갖는 파티클들을 제외한 나머지 파티클들의 개수를 나타내며, K는 분할된 지역의 개수이다. 보호지역 외에 위치한 파티클들 중 파티클의 수가 적은 지역에 위치한 파티클의 n은

, 및 K를 이용하여 높은 가산가중치가 부여된다.

은 미리 정해진 범위의 가중치를 갖는 파티클을 제외한 나머지 파티클의 개수에서 각 지역에 분포해야할 파티클의 평균 수치를 구하는 것을 나타낸다. 예를 들어, 100개의 파티클들을 4개의 지역으로 나누었을 경우 각 지역마다 존재 해야 할 파티클의 수는 (100/4)/100으로 0.25가 된다.

은 각 지역에 존재하는 파티클의 실제 수를 구하기 위함이다.
here,

Represents the number of particles located in the K region, M represents the number of particles excluding the particles having a predetermined range weight among the particles located in the K region, and K is the number of the divided regions. Particles located outside the protected area with a small number of particles

, ≪ / RTI > and K are given.

Represents an average value of particles to be distributed in each region from the number of remaining particles excluding particles having a predetermined range of weights. For example, if 100 particles are divided into 4 regions, the number of particles to be present in each region is (100/4) / 100, which is 0.25.

Is to find the actual number of particles in each region.

6 is a diagram for explaining a method of predicting a position of a mobile device according to another embodiment of the present invention.

The method of predicting the position of the mobile device according to another embodiment may be the particle re-extraction technique described in FIG.

A method of predicting the position of a mobile device according to another embodiment may record a plurality of particles observed at a feature point and weight particles 610.

A method of predicting the position of a mobile device according to another embodiment may group 620 particles into a predetermined number of groups.

A method of predicting the location of a mobile device according to another embodiment may set a protected area within a group (630). Here, the protected area is set based on the area center point, and the preset protective radius, and the area center point can be calculated based on the coordinates of the particles in the group.

A method for predicting a position of a mobile device according to another embodiment is based on at least one particle within a group whose weight is within a predetermined range, at least one particle located within the protected area, and at least one particle located outside the protected area A particle set may be constructed 640.

According to another embodiment, the distance between the at least one particle located within the protected area and the local center point may be less than the center distance, and a distance established based on a predetermined protective radius. Here, the center distance is the distance between the particle center point position and the local center point position, and the particle center point position can be calculated based on the coordinates of the particles and the weight of the particles.

According to another embodiment, in step 640, at least one particle located within the protected area, and at least one particle located outside the protected area may be replicated by n. In the case of at least one particle located outside the protected area, an addition weight is given to n, and at least one particle outside the protected area can be replicated by n given the addition weight. In the case of at least one particle located within the protected area, if n is less than the predetermined number, at least one particle in the protected area may be replicated by adding a predetermined number to n.

7 is a block diagram for explaining a configuration of an apparatus for predicting a position of a mobile device according to an embodiment of the present invention.

An apparatus for predicting the position of a mobile device according to an embodiment may be implemented through the particle re-extraction technique described with reference to FIG.

Referring to FIG. 7, an apparatus 700 for predicting the position of a mobile device according to an embodiment includes a memory 710, a processor 720, and a predictor 730.

The memory 710 may store at least one particle having a weight within a predetermined range.

The processor 720 may group a plurality of particles distributed around the feature points into a predetermined number of regions based on the particle center point. In addition, the processor 720 may replicate n remaining particles except for at least one particle having a weight within a predetermined range among the particles in the grouped region. Where n can be calculated based on the number of remaining particles and the weight of the remaining particles.

According to one embodiment, the processor 720 may determine whether the remaining particles are in a protected area. More specifically, the processor 720 may determine that the remaining particles are in the protected area if the distance between the location of the local center point and the location of the remaining particles is less than the center distance, and the distance established based on the predetermined guard radius . Here, the position of the local center point can be calculated based on the coordinates of the particles positioned in the grouped region, and the center distance can be the distance between the position of the particle center point and the position of the local center point. The center point of the particle can be calculated based on the coordinates of a plurality of particles distributed around the feature point, and the weight of the particles.

According to one embodiment, when processor 720 determines that the remaining particles are outside the protected area, n may be given an addition weight to replicate the remaining particles. In addition, when the processor 720 determines that the remaining particles are within the protection area, when n is less than the predetermined number, the processor 720 may add the predetermined number to the n to replicate the particles located in the protection area.

The predicting unit 730 can predict the location of the mobile device based on the stored at least one or more particles and the copied particles.

8 is a block diagram for explaining a configuration of an apparatus for predicting a position of a mobile device according to another embodiment of the present invention.

The apparatus for predicting the position of the mobile device according to another embodiment may be implemented through the particle re-extraction technique described in FIG.

Referring to FIG. 8, an apparatus 800 for predicting a position of a mobile device according to another embodiment includes an updating unit 810, a re-extracting unit 820, and a predicting unit 830.

The update unit 810 may record a plurality of particles observed at the feature point, and may assign a weight to the particles.

The re-extracting unit 820 groups the particles into a predetermined number of groups, sets a protected area within the group, and includes at least one particle within the group whose weight is within a predetermined range, at least one particle , And a particle set based on at least one particle located outside the protected area.

The prediction unit 830 can predict the position of the mobile device based on the set of particles.

According to another embodiment, the re-extracting unit 820 can set the protected area based on the area center point and the preset protection radius. Here, the local center point can be calculated based on the coordinates of the particles in the group.

According to another embodiment, the distance between the at least one particle located within the protected area and the local center point may be less than the center distance, and a distance established based on a predetermined protective radius. Here, the center distance is a distance between the particle center point position and the local center point position, and the particle center point position can be calculated based on the coordinates of the particles and the weight of the particles.

According to another embodiment, at least one particle located in the protected area and at least one particle located outside the protected area can be replicated by n.

According to another embodiment, re-extractor 820 may assign an addition weight to n in the case of at least one particle located outside the protected area. Also, the re-extracting unit 820 may replicate at least one particle located outside the protected area by the added weight n.

According to another embodiment, in the case of at least one particle located in the protected area, the re-extracting unit 820 may add a predetermined number to n if n is less than the predetermined number. Also, the re-extracting unit 820 may replicate at least one particle located in the protected area by a predetermined number n.

FIG. 9 is a view for explaining a performance comparison between the particle re-extraction technique used in the present invention and the existing particle re-extraction technique.

FIG. 9A is a diagram comparing the number of remaining particles in the re-extraction process, FIG. 9B is a diagram comparing RMS position errors, FIG. 9C is a diagram comparing RMS feature point errors, And the average RMS position error according to the number. 9 (a) to 9 (d), a represents SIR (Sequential Importance Resampling), b represents PSR (Partial stratified resampling), and c represents re-extraction technique used in the present invention.

The RMS position error refers to the difference between the given path and the path predicted by the mobile device. That is, the difference between the actual position of the mobile device and the position predicted by the mobile device is called an RMS position error.

RMS feature point error refers to the difference between the feature points and the particles around the feature point. That is, the difference between the position of the actual feature point and the position of the feature point predicted by the mobile device is referred to as an RMS feature point error.

9 (a), as the particle re-extraction process proceeds, the existing re-extraction technique rapidly reduces the number of particles. However, the particle re- Is not greater than the re-extraction technique. Also, as the re-extraction process proceeds, the particle re-extraction technique used in the present invention maintains more particles than the existing re-extraction technique.

9 (b) and 9 (c), the particle re-extraction technique used in the present invention has lower RMS position error and RMS feature point error than the existing re-extraction technique. That is, the particle re-extraction technique used in the present invention can stably correct the position of the mobile device by maintaining the low RMS position error and the RMS feature point error as compared with the existing re-extraction technique.

Referring to FIG. 9 (d), when the number of particles is the same, the particle re-extraction technique used in the present invention has a lower RMS position error than the existing re-extraction technique. Compared with the existing re-extraction technique, the particle re-extraction technique used in the present invention can effectively reduce the RMS position error of the mobile device even though the number of particles is small.

The apparatus described above may be implemented as a hardware component, a software component, and / or a combination of hardware components and software components. For example, the apparatus and components described in the embodiments may be implemented within a computer system, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA) , A programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For ease of understanding, the processing apparatus may be described as being used singly, but those skilled in the art will recognize that the processing apparatus may have a plurality of processing elements and / As shown in FIG. For example, the processing unit may comprise a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as a parallel processor.

The software may include a computer program, code, instructions, or a combination of one or more of the foregoing, and may be configured to configure the processing device to operate as desired or to process it collectively or collectively Device can be commanded. The software and / or data may be in the form of any type of machine, component, physical device, virtual equipment, computer storage media, or device , Or may be permanently or temporarily embodied in a transmitted signal wave. The software may be distributed over a networked computer system and stored or executed in a distributed manner. The software and data may be stored on one or more computer readable recording media.

The method according to an embodiment may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions to be recorded on the medium may be those specially designed and configured for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. For example, it is to be understood that the techniques described may be performed in a different order than the described methods, and / or that components of the described systems, structures, devices, circuits, Lt; / RTI > or equivalents, even if it is replaced or replaced.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

700: Position estimating device of mobile equipment.
710: Memory
720: Processor
730:
800: Position estimating device of mobile equipment.
810:
820:
830:

Claims (21)

A method for predicting a location of a mobile device,
Grouping a plurality of particles distributed around a feature point into a predetermined number of regions based on a particle center point;
Storing in the memory at least one particle having a weight within a predetermined range among the particles located within the grouped region;
Duplicating the remaining particles except for the stored particles among the particles in the grouped region by n; And
Predicting a location of the mobile device based on the stored at least one particle and the replicated particles
And estimating a position of the mobile device.
The method according to claim 1,
Wherein n is a number
The number of remaining particles, and the weight of the remaining particles,
A method for predicting a location of a mobile device.
The method according to claim 1,
Determining whether the remaining particles among the particles located within the grouped region are in a protected area
Further comprising:
If it is determined that the remaining particles are out of the protected area,
And adding the weight to the n to replicate the remaining particles,
A method for predicting a location of a mobile device.
The method of claim 3,
If it is determined that the remaining particles are in the protected area,
Wherein when n is less than a predetermined number, adding a predetermined number to the n to replicate particles located within the protected area,
A method for predicting a location of a mobile device.
The method of claim 3,
Wherein determining whether the remaining particles are in a protected area comprises:
Determining that the remaining particles are in the protected area when the distance between the position of the local center point and the position of the remaining particles is less than a center distance and a distance set based on a predetermined protective radius,
A method for predicting a location of a mobile device.
6. The method of claim 5,
The position of the regional center point
The coordinates of the particles located within the grouped region,
The center distance,
Which is a distance between a position of the particle center point and a position of the area center point,
A method for predicting a location of a mobile device.
The method according to claim 1,
The particle center point,
The coordinates of the particles, and the weight of the particles,
A method for predicting a location of a mobile device.
A method for predicting a location of a mobile device,
Recording a plurality of particles observed at the feature point, and weighting the particles;
Grouping the particles into a predetermined number of groups;
Setting a protected area within the group;
Constructing a set of particles based on at least one particle in the group, the weight being within a predetermined range, at least one particle located within the protected area, and at least one particle located outside the protected area; And
Predicting a position of the mobile device based on the set of particles
And estimating a position of the mobile device.
9. The method of claim 8,
Wherein setting the protected area within the group comprises:
The area center point, and the preset protection radius,
The area center point
And calculating, based on the coordinates of the particles in the group,
A method for predicting a location of a mobile device.
10. The method of claim 9,
Wherein the distance between the at least one particle located within the protected area and the local center point is less than the set distance based on a center distance and a predetermined protective radius,
The center distance,
The distance between the particle center point position and the local center point position,
The particle center point position
The coordinates of the particles, and the weight of the particles,
A method for predicting a location of a mobile device.
9. The method of claim 8,
Wherein the step of constructing the particle set comprises:
Replicating at least one particle located within said protected area and at least one particle located outside said protected area by n,
And estimating a position of the mobile device.
12. The method of claim 11,
Wherein the copying comprises:
In the case of at least one particle located outside the protected area, assigning an addition weight to the n and replicating at least one particle located outside the protected area by n given the added weight,
A method for predicting a location of a mobile device.
12. The method of claim 11,
Wherein the copying comprises:
For at least one particle located within said protected area, replicating at least one particle located within said protected area by adding a predetermined number to said n if n is less than a predetermined number,
A method for predicting a location of a mobile device.
A computer-readable recording medium having recorded thereon a program for executing the method according to any one of claims 1 to 13.
An apparatus for predicting a location of a mobile device,
Grouping a plurality of particles distributed around a minutiae point into a predetermined number of regions with respect to a center point of the particle and selecting n remaining particles excluding at least one particle having a weight within a predetermined range among the particles in the grouped region Respectively;
A memory for storing at least one particle having a weight within the predetermined range; And
A predictor for predicting a position of the mobile device based on the stored at least one particle and the copied particles,
And estimates the position of the mobile device.
16. The method of claim 15,
Wherein n is a number
The number of remaining particles, and the weight of the remaining particles,
A device for predicting the location of a mobile device.
16. The method of claim 15,
The processor comprising:
And when n is less than a predetermined number, determining that the remaining particles are within the protection area if the remaining particles are determined to be outside the protection area, replicating the remaining particles by assigning an addition weight to n, Adding a predetermined number to said n to replicate particles located within said protected area,
A device for predicting the location of a mobile device.
An apparatus for predicting a location of a mobile device,
An update unit for recording a plurality of particles observed at the feature point and assigning a weight to the particles;
Grouping the particles into a predetermined number of groups and setting a protected area within the group, wherein at least one particle within the group has a weight within a predetermined range, at least one particle located within the protected area, and A re-extractor configured to form a particle set based on at least one particle located outside the protected area; And
A predictor for predicting a position of the mobile device based on the set of particles,
And estimates the position of the mobile device.
19. The method of claim 18,
The re-
The area center point, and the preset protection radius,
The area center point
And calculating, based on the coordinates of the particles in the group,
A device for predicting the location of a mobile device.
20. The method of claim 19,
Wherein the distance between the at least one particle located within the protected area and the local center point is less than the set distance based on a center distance and a predetermined protective radius,
The center distance,
The distance between the particle center point position and the local center point position,
The particle center point position
The coordinates of the particles, and the weight of the particles,
A device for predicting the location of a mobile device.
19. The method of claim 18,
The re-
At least one particle located within said protected area, and at least one particle located outside said protected area,
A device for predicting the location of a mobile device.

Images