KR101716381B1 - Apparatus and method for classifying step movement - Google Patents

Abstract

The present invention relates to an apparatus and method for classifying a walking motion, and a walking motion classifying apparatus according to the present invention includes an inertial sensor unit including an acceleration sensor and a geomagnetic sensor, Calculate the magnitude of the linear acceleration from the acceleration vector, calculate the magnitude value of the geomagnetism vector output from the geomagnetism sensor, and use the magnitude value of the linear acceleration and the magnitude value of the geomagnetism vector, A step operation discriminating section for discriminating the step operation is provided, so that the step operation can be accurately discriminated.

Description

[0001] APPARATUS AND METHOD FOR CLASSIFYING STEP MOVEMENT [0002]

The present invention relates to an apparatus and method for classifying a stepping motion, and more particularly, to an apparatus and method for classifying a stepping motion that can distinguish a stepping motion from a forward moving step and a repeated stepping motion.

As of December 2012, the number of smartphone subscribers reached 33 million in Korea. As most people use smartphones, Location-Based Services (LBS) has become one of the main services. One of the typical LBS is the car navigation service. If the environment in which the LBS is provided is outdoor, the current position can be grasped by using a Global Navigation Satellite System (GNSS) represented by a Global Positioning System (GPS).

However, the GNSS method can not be used in the indoor LBS such as the guidance in a complex indoor environment such as a large indoor amusement park or shopping mall, and the intelligent exhibition service in various exhibition halls, and research has been actively conducted.

However, representative and effective methods and systems such as GNSS have not yet been developed or commercialized. However, in recent years, a variety of methods have been researched and developed to provide location accuracy that is economical and capable of various kinds of LBS by using various sensors and communication functions of a smart phone according to popularization of smart phones.

One of them is researching and developing a dead-reckoning method which estimates the current position using only the inertial sensor of the device without requiring a separate infrastructure. Especially, many inertial sensors have been installed in many mobile devices, and many methods have been developed to detect the movement of the user and track movement.

In these methods, it is very important to accurately detect the actual moving walking movement. Especially, it is very important to develop a method to classify various behaviors similar to walking behavior, for example, in situ walking and actually moving walking, in many daily activities.

In order to solve the above-described problems, the present invention relates to an apparatus and method for classifying a stepping motion that can distinguish an actual forward-moving stepping motion from an in-situ repeated stepping motion.

According to an aspect of the present invention, there is provided an apparatus for classifying a gesture according to an embodiment of the present invention, including an inertia sensor unit including an acceleration sensor and a geomagnetic sensor, Calculating a magnitude of a linear acceleration from the linear acceleration vector, calculating a magnitude value of the geomagnetism vector output from the geomagnetism sensor, and calculating a magnitude value of the geomagnetism vector outputted from the geomagnetism sensor, using a magnitude value of the linear acceleration and a magnitude value of the geomagnetism vector, And a gesture motion determination unit that distinguishes the gesture motion that is repeated in place.

The step motion determiner continuously calculates a magnitude value of the linear acceleration for a predetermined time interval, calculates a data sum of magnitude values of the linear acceleration, continuously obtains the magnitude value of the geomagnetism during the predetermined time interval, And then apply the complementary filter to the computed linear acceleration-related data sum and the calculated geomagnetism-related dispersion values.

The step operation determination unit may store the filter output value applied the complementary filter in a buffer, calculate a variance value for a predetermined number of output values, calculate an average value using the calculated variance values, have.

The step of determining the step of the step of comparing the step of comparing the step-related characteristic value with the predetermined threshold may determine whether the step is a step that is actually moving forward or a step that is repeated in the place.

The inertial sensor unit may further include a gyro sensor and may further include a detection unit for detecting a holding mode of the walking motion classification apparatus using the magnitude of the gravitational acceleration obtained from the magnitude of the angular velocity of the gyro sensor and the acceleration vector of the acceleration sensor And the handwriting detection unit may change the predetermined threshold value, which is compared with the keying operation-related feature value, by the key operation determination unit according to the detected handwriting detection method.

According to another embodiment of the present invention, there is provided a method of classifying a walking motion in an apparatus including an acceleration sensor and a geomagnetic sensor, the method comprising: calculating a magnitude of a linear acceleration from a linear acceleration vector obtained by removing gravitational acceleration from an acceleration vector output from the acceleration sensor; Calculating a magnitude value of the geomagnetism vector outputted from the geomagnetism sensor; calculating a magnitude value of the linear acceleration and a geomagnetism magnitude value calculated in the step of calculating the magnitude value of the linear acceleration; The above-described object can be achieved by providing a step of distinguishing between the stepping motion that actually moves forward and the stepping motion that is repeated in the place, using the magnitude value of the geomagnetism vector calculated in step.

According to the above-described configuration, the present invention can accurately distinguish between a stepping motion that actually moves forward and a stepping motion that repeats in place.

FIG. 1 is a block diagram of a stepping motion classifying apparatus according to an embodiment of the present invention. Referring to FIG.
FIG. 2 is a block diagram of the step-motion classifying apparatus shown in FIG. 1 in more detail.
FIG. 3 is a diagram illustrating the motion characteristic calculation part shown in FIG. 2 in more detail.
4 (a), 4 (b) and 4 (c) are graphs illustrating graphs according to an embodiment of the present invention.
5 is a flowchart illustrating a method of classifying a stepping motion according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of an apparatus and method for classifying a stepping motion according to the present invention will be described with reference to the accompanying drawings. In the following description of the present invention, it is to be understood that the terminology used herein is for the purpose of describing particular and exemplary embodiments of the invention, Will be.

FIG. 1 is a block diagram of a stepping motion classifying apparatus according to an embodiment of the present invention. Referring to FIG.

As shown in FIG. 1, the walking motion classifying apparatus includes an inertial sensor unit 110, a grounding type detecting unit 120, and a walking motion determining unit 130. Here, the step motion classification device may be, for example, a mobile device such as a smart phone or a wrist type electronic device such as an electronic watch.

The inertial sensor unit 110 is inertial sensors provided in the walking motion classifying device and outputs inertia data based on the motion of the walking motion classifying device. The inertial sensor unit 110 can output inertia data for a predetermined sampling time, for example, every 50 msec.

The navigation system detecting unit 120 detects the navigation system based on inertial data output from the inertial sensor unit 110, for example, the navigation system of the mobile device. There are many types of attitudes to carry mobile devices, such as people putting in their pockets (front / back), people carrying their hands, people carrying their bags, and so on. Therefore, if the current possession method of the mobile device can be detected and known, it is possible to accurately determine whether the mobile device is actually walking or moving, or the like.

The step operation determination unit 130 determines whether the user is actually walking or moving from the inertial data of the step motion classification apparatus.

In FIG. 1, the sub-system detection unit 120 is illustrated, but this is for the purpose of illustrating the present invention and may not necessarily be a necessary configuration for the present invention.

Fig. 2 is a block diagram of the step-motion classifying apparatus shown in Fig. 1 in more detail. Fig. 3 is a diagram showing the motion characteristic calculating part shown in Fig. 2 in more detail, , 4 (b) and 4 (c) are graphs illustrating graphs according to an embodiment of the present invention.

2, the inertial sensor unit 110 may include an acceleration sensor 212, a gyro sensor 214, and a geomagnetic sensor 216. As shown in FIG.

이며, 여기서 a_x, a_y 및 a_z는 X축, Y축 및 Z축 각각의 가속도 성분을 나타낸다.The acceleration sensor 212 measures the acceleration (gravitational acceleration + motion acceleration) of each of the X-axis, Y-axis, and Z-axis of the step-motion classifying apparatus. The acceleration vector output from the acceleration sensor 212 is

, Where a _x , a _y, and a _z represent the acceleration components of the X, Y, and Z axes, respectively.

이며, 여기서 w_x, w_y 및 w_z는 X축, Y축 및 Z축 각각의 각속도 성분을 나타낸다.The gyro sensor 214 is a sensor for measuring the angular velocity of each of the X-axis, the Y-axis, and the Z-axis of the step-motion classifying apparatus. The angular velocity vector output from the gyro sensor 214 is

, Where w _x , w _y, and w _z represent the angular velocity components of the X, Y, and Z axes, respectively.

이며, 여기서 m_x, m_y 및 m_z는 X축, Y축 및 Z축 각각의 지자기 성분을 나타낸다.The geomagnetic sensor 216 is an electronic compass capable of measuring an azimuth using a geomagnetic field in a walking motion classifying device. The geomagnetism vector outputted from the geomagnetic sensor 216 is

, Where m _x , m _y, and m _z represent the geomagnetism components of the X axis, Y axis, and Z axis, respectively.

2, the parental feature detection unit 120 includes an angular velocity feature calculation part 222 and a gravitational acceleration feature calculation part 224 in order to detect the manner in which the current user has the step motion classification device .

및 각속도 벡터 크기 값의 일정한 개수(N1)의 과거 샘플들의 분산 값

을 계산하며, 또한 X축 각속도 성분의 일정 개수(N2)의 분산 값

을 계산한다.The angular velocity feature calculation part 222 calculates the magnitude of the angular velocity vector

And a variance value of past samples of a certain number Nl of angular velocity vector magnitude values

And also calculates the variance value (N2) of a certain number N2 of X-axis angular velocity components

.

을 계산한다. 예를 들면, 샘플링 시간 50msec인 경우에 1.5초 동안 샘플링을 하면 N=30이다. 중력 가속도 특징 계산 파트(224)는 또한, 중력 가속도 벡터로부터 X축 및 Y축의 중력 가속도 평균값

을 계산하고, 아울러 Z축 중력 가속도 성분의 일정 개수의 분산 값

을 계산한다.The gravitational acceleration characteristic calculation part 224 calculates the characteristic values of the gravitational acceleration vector for a predetermined time. The gravitational acceleration characteristic calculation part 224 obtains the gravitational acceleration vector from the acceleration vector input from the acceleration sensor 212, and calculates average values of the X and Y axes of the gravitational acceleration vector of the current time-

. For example, if the sampling time is 50 msec and sampling is performed for 1.5 seconds, N = 30. The gravitational acceleration characteristic calculation part 224 also calculates the gravitational acceleration average value of the X and Y axes from the gravitational acceleration vector

And also calculates a dispersion value of a certain number of Z-axis gravity acceleration components

.

The base-paper type detection unit 120 detects feature values calculated in the angular velocity feature calculation part 222 and feature values calculated in the gravitational acceleration feature calculation part 224 with predetermined thresholds to detect the base type .

The step motion determination unit 130 may include a linear acceleration size calculation part 232 and a motion characteristic calculation part 234 to determine whether the robot is actually walking or walking.

이다.The linear acceleration magnitude calculation part 232 can obtain a linear acceleration sensor in which a gravity component is removed by using a low pass filter (LPF) or the like that passes only a certain frequency component or less. The linear acceleration vector can be obtained by filtering the gravity components in the X-axis, Y-axis, and Z-axis components, and the linear acceleration vector

to be.

The motion characteristic calculation part 234 calculates a motion characteristic using the geomagnetism value input from the geomagnetism sensor 216 and the acceleration value input from the acceleration sensor 212. [ The detailed configuration of the motion characteristic calculation part 234 is shown in Fig.

3, the motion characteristic calculation part 234 includes a geomagnetism size dispersion calculation module 312, a linear acceleration magnitude sum calculation module 314, a complementary filter module 316, a filter output dispersion calculation module 318, An averaging module 320, a threshold value module 322, and a feature value comparison module 324.

로 표시되며, 이 분산 값의 그래프가 도 4(a)에 도시되어 있다.The geomagnetism size variance calculation module 312 calculates a variance value for a past NI (for example, 12) dataset of the magnitude value of the geomagnetism vector input from the geomagnetism sensor 216,

, And a graph of the dispersion value is shown in Fig. 4 (a).

은 다음의 수식을 이용하여 계산되며, 이 데이터 합의 그래프가 도 4(b)에 도시되어 있으며, 여기서 k는 스케일링 계수이다.Meanwhile, the linear acceleration magnitude sum calculation module 314 calculates N1 data sum of the magnitude values of the linear acceleration vector outputted from the linear acceleration magnitude calculation part 232. [ This data sum

Is calculated using the following equation, and the graph of this data sum is shown in Fig. 4 (b), where k is a scaling factor.

을 얻을 수 있으며, 여기서 필터 상수 α는 0보다 크고 1보다 작은 값으로 결정되는데, 0.2±0.1인 것이 바람직하다. The complementary filter module 316 applies a complementary filter to the variance value input from the geomagnetism size variance calculation module 312 and the data sum input from the linear acceleration magnitude sum calculation module 314,

Where the filter constant alpha is greater than 0 and less than 1, preferably 0.2 + - 0.1.

을 계산하고 버퍼에 저장한다.The complementary filter module 316 compares the filter output value

And stores it in the buffer.

을 계산한다.The filter output variance calculation module 318 calculates a variance value of the N1 filter output values stored in the buffer

.

의 평균값

을 계산한다. 이 평균값이 움직임 특성 계산 파트(234)에서 얻어지는 최종적인 움직임 특성값이다.The average calculation module 320 calculates the final four variance values calculated in the filter output variance calculation module 318

Average of

. This average value is the final motion characteristic value obtained by the motion characteristic calculation part 234.

FIG. 4 (c) is a diagram showing motion characteristic values according to actual walking and standing walking. In FIG. 4 (c), it can be seen that the actual step and the footstep are clearly distinguished based on the threshold value.

As shown in FIG. 3, the motion characteristic calculation part 234 may further include a threshold value module 322 and a feature value comparison module 324. The threshold value module 322 may include a storage unit storing a threshold value or a register and the characteristic value comparison module may compare the final motion characteristic value obtained by the average calculation module 320 and the threshold value of the threshold value module 322 To output a characteristic code, for example, 111 or 333.

The motion characteristic calculation part 234 of the step movement determination part 130 can change the threshold value of the threshold value module 322 according to the holding method detected by the holding method detecting part 120. [

5 is a flowchart illustrating a method of classifying a stepping motion according to an embodiment of the present invention.

The user drives the step motion classification program displayed on the display unit of the step motion classification apparatus (S502). When the step motion classification program is activated, the acceleration sensor 212, the gyro sensor 214, and the geomagnetic sensor 216 of the step motion classification device output respective inertia data every 50 ms (S504).

The linear acceleration size calculation part 232 obtains a linear acceleration vector by filtering the gravity components in the X-axis, Y-axis, and Z-axis components input from the acceleration sensor 212 (S506).

The linear acceleration size calculation part 232 calculates the size of the linear acceleration vector (S508).

은 다음의 수식을 이용하여 계산되며, 이 데이터 합의 그래프가 도 4(b)에 도시되어 있으며, 여기서 k는 스케일링 계수이다.The linear acceleration magnitude sum calculation module 314 calculates N1 data sum of the magnitude values of the linear acceleration vectors output from the linear acceleration magnitude calculation portion 232 (S510). This data sum

Is calculated using the following equation, and the graph of this data sum is shown in Fig. 4 (b), where k is a scaling factor.

로 표시되며, 이 분산 값의 그래프가 도 4(a)에 도시되어 있다.The geomagnetism size dispersion calculation module 312 calculates a dispersion value of the past NI data set of the magnitude value of the geomagnetism vector input from the geomagnetic sensor 216 (S512)

, And a graph of the dispersion value is shown in Fig. 4 (a).

을 얻을 수 있으며(S514), 여기서 필터 상수 α는 0보다 크고 1보다 작은 값으로 결정되는데, 0.2±0.1인 것이 바람직하다. The complementary filter module 316 applies a complementary filter to the variance value input from the geomagnetism size variance calculation module 312 and the data sum input from the linear acceleration magnitude sum calculation module 314,

(S514), where the filter constant? Is determined to be greater than 0 and less than 1, preferably 0.2 + - 0.1.

을 계산하고 버퍼에 저장한다.The complementary filter module 316 compares the filter output value

And stores it in the buffer.

을 계산한다(S516).The filter output variance calculation module 318 calculates a variance value of the N1 filter output values stored in the buffer

(S516).

의 평균값

을 계산한다(S518). 이 평균값이 움직임 특성 계산 파트(234)에서 얻어지는 최종적인 움직임 특성값이다.The average calculation module 320 calculates the final four variance values calculated in the filter output variance calculation module 318

Average of

(S518). This average value is the final motion characteristic value obtained by the motion characteristic calculation part 234.

The characteristic value comparison module compares the final motion characteristic value obtained in the average calculation module 320 with the threshold value in the threshold value module 322 (S520). In this case, the characteristic value comparison module outputs a characteristic code, for example, 111 if the final motion characteristic value obtained by the average calculation module 320 is greater than a threshold value in the threshold value module 322, for example, 0.045, , And if it is 0.045 or less, a feature code, for example, 333 is output to indicate that it is a standstill.

The scope of protection of the present invention should be interpreted according to the claims. It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit and scope of the invention as defined by the appended claims. It should be interpreted that it is included in the scope of right.

110: inertia sensor unit 120:
130: step operation determination unit 212: acceleration sensor
214: Gyro sensor 216: and geomagnetic sensor
222: Angular velocity feature calculation part 224: Gravity acceleration feature calculation part
232: Linear Acceleration Size Calculation Part 234: Motion Characteristic Calculation Part

Claims (7)

An inertial sensor unit including an acceleration sensor and a geomagnetic sensor,
Calculating a magnitude of the linear acceleration from the linear acceleration vector obtained by removing the gravity acceleration from the acceleration vector output from the acceleration sensor, calculating a magnitude value of the geomagnetism vector output from the geomagnetism sensor, And a gesture motion determiner for distinguishing between a gesture motion that actually moves forward and a gesture motion that repeats in the first place using the magnitude value of the geomagnetism vector,
The step motion determiner continuously calculates a magnitude value of the linear acceleration for a predetermined time interval, calculates a data sum of magnitude values of the linear acceleration, continuously obtains the magnitude value of the geomagnetism during the predetermined time interval, Calculating a variance value for the set, and applying a complementary filter to the calculated linear acceleration-related data sum and the calculated geomagnetism-related variance value. delete The method according to claim 1,
The step operation determining unit stores a filter output value to which the complementary filter is applied in a buffer, calculates a variance value for a predetermined number of output values, calculates an average value using the calculated variance values, Wherein the step motion classification device comprises: The method of claim 3,
Wherein the step of determining the step of gait is to compare the gait operation-related characteristic value with a predetermined threshold value to determine whether the gait is a step that is actually moving forward or a step that is repeated in place. 5. The method of claim 4,
Wherein the inertial sensor unit further comprises a gyro sensor,
Further comprising a detection means for detecting a holding mode of the walking motion classification apparatus using the magnitude of the angular velocity of the gyro sensor and the magnitude of the gravitational acceleration obtained from the acceleration vector of the acceleration sensor,
Wherein the retention mode detecting unit changes the predetermined threshold value to be compared with the step-motion-related feature value in the step-motion determining unit according to the detected retention mode. A method of classifying a stepping motion in an apparatus including an acceleration sensor and a geomagnetic sensor,
Calculating a magnitude value of a linear acceleration from a linear acceleration vector obtained by removing gravitational acceleration from an acceleration vector output from the acceleration sensor;
Calculating a magnitude value of a geomagnetism vector output from the geomagnetic sensor,
The step of calculating the magnitude of the linear acceleration and the magnitude value of the geomagnetism vector calculated in the step of calculating the geomagnetism magnitude value may be used to calculate the magnitude of the linear acceleration, Comprising the steps of:
Wherein the step of classifying the stepping motion includes continuously calculating a magnitude value of the linear acceleration for a predetermined time interval, calculating a data sum of magnitude values of the linear acceleration, continuously calculating the magnitude value of the geomagnetism during the predetermined time interval, Calculating a variance value for a set of magnitude values, and applying a complementary filter to the calculated linear acceleration-related data sum and the calculated geomagnetism-related variance value to classify the stepping motion. delete

Images