南台科技大學 資訊工程系 Posture Monitoring System for Context Awareness in Mobile Computing Authors: Jonghun Baek and Byoung-Ju Yun Adviser: Yu-Chiang Li Speaker:

南台科技大學 資訊工程系

Posture Monitoring System for Context Awareness in Mobile ComputingAuthors:Jonghun Baek and Byoung-Ju YunAdviser: Yu-Chiang Li Speaker: Gung-Shian LinDate:2011/01/14IEEE Transactions on Instrumentation and Measurement, VOL. 59, NO. 6, JUNE 2010

2

Outline

Introduction1

Sensors2

TAMA3

User Posture Monitoring4

Recognition results5

Conclusion6

3

1. Introduction

The posture of a user is one of the contextual information that can be used for mobile applications and the treatment of idiopathic scoliosis.

This paper describes a method for monitoring the posture of a user during operation of a mobile device in three activities such as sitting, standing, and walking.

4

1. Introduction

The user posture monitoring system (UPMS) proposed in this paper is based on two major technologies. The first involves a tilt-angle measurement algorithm

(TAMA) using an accelerometer.

The second technology is an effective signal-processing method that eliminates the motion acceleration component of the accelerometer signal using a second-order Butterworth low-pass filter (SLPF).

5

2. Sensors

Typical output values of the accelerometer due to gravity.

6

3. TAMA

It used the reference vectors defined as the acceleration values measured at 0◦ of the X- and Y -axes compensated at the datum angle, respectively.

7

3. TAMA

Signal Processing for Measuring the Tilt Angle

8

3. TAMA

Data Collection Method

The time-series acceleration data from the accelerometer was gathered for approximately 30 s for each degree at a sampling rate of the 100 samples/s, and it is termed the training data set.

9

3. TAMA

Compensation and Reference Vectors We define the offset errors and the reference vectors as the

model parameters of the TAMA.

10

3. TAMA

The equations for the model parameters and compensation for each axis in each datum angle.

11

3. TAMA

Table shows the values of the model parameters obtained at each datum angle using the training data set.

12

3. TAMA

Estimation Time To estimate the posture of a user during mobile computing,

the accelerometer was attached to a PDA, and the TAMA was implemented on it.

13

3. TAMA

Performance Evaluation Table shows the tilt angles measured by the TAMA with 1-s

estimation time and 180◦ datum angle.

14

3. TAMA

These results were compared with the previous research [7] in the range of 0◦ to 70◦ using evaluation factors.

15

4. User Posture Monitoring

System Architecture

16

4. User Posture Monitoring

Data Collection Method The training data sets were collected in our scenario from

five subjects that were asked to perform a test: after the initial state of about 5 s, the subjects watched the movie played out by the PDA for about 15 s.

17

4. User Posture Monitoring

Motion Acceleration Component Elimination The frequency response curves have their peak values at a

specific frequency component when the pole values were complex numbers.

18

4. User Posture Monitoring

If the pole values were real numbers and the poles were moved to the left half-plane in the z-plane.

19

4. User Posture Monitoring

When poles were moved to the right half-plane, the skirt characteristic of the SLPF was better, and the SLPF allowed passing the very small low-frequency component.

20

4. User Posture Monitoring

An experiment was conducted to eliminate the motion acceleration component according to moving of the pole values of the SLPF.

(a) Original time-series acceleration data.(b)–(e) Time-series acceleration data after filtering: (b) p1 = −0.5 − j0.5, p2 = −0.5 + j0.5; (c) p1 = p2 = −0.6; (d) p1 = p2 = 0.7; (e) p1 = p2 = 0.97.

21

4. User Posture Monitoring

To find out the proper pole values of the SLPF, the pole values were investigated in the range of 0.95 to 0.99.

22

4. User Posture Monitoring

Posture Recognition in Three Activities To determine the range of θ for the posture of a user, a

series of threshold analysis tests were run. The θ in each activity was calculated by the TAMA with the

training data set.

23

4. User Posture Monitoring

The threshold analyses were performed on the training data sets to estimate the posture of a user in each activity, and we examined the values of the optimal threshold to determine the convergence of the posture.

24

5. Recognition results

Two evaluation factors were used as follows: the ratio of the number of “Display ON” to the number of

trials. the ratio of the number of “Display ON” to the number of

malfunctions (“Display OFF”).

25

5. Recognition results

The recognition accuracy of the UPMS.

26

6. Conclusion

The TAMA can be used to estimate not only the posture of users with a mobile device, as mentioned in this paper, but also the posture of scoliosis patients and the bent spine posture of musicians, athletes, or public people.

The proposed UI using context-aware computing can automatically recognize the posture of a mobile device user with good accuracy.

南台科技大學 資訊工程系