os, T. 
anning 
)botics 
1720— 
-Based 
mputer 
fed '95 
m, C. 
| H.-P. 
rungen 
rurgie, 
edings, 
aft für 
itza, I. 
und 
riffen*, 
71.101, 
orn, C. 
M.W. 
in der 
und 
Aedizin 
'hmann 
S, “The 
Reality 
edicine 
, 1998, 
ANALYSIS OF TIME-SPACE EFFECT OF WALKING 
USING ACCELEROMETERS AND GYRO SENSOR SYSTEM 
Masako Tsuruoka®*, Yuriko Tsuruoka®, Ryosuke Shibasaki’, Yoshifumi Yasuoka*, Shunji Murai* 
^ Institute of Industrial Science, University of Tokyo, 4-6-1, Komaba, Meguro-ku, Tokyo, Japan - 
(*masako, yyasuoka) @iis.u-tokyo.ac.jp, shibasaki@skl.iis.u-tokyo.ac.jp, shmurai@nifty.ne.jp 
^ Institute of Statistical Mathematics,4-6-7, Minamiazabu, Minato-ku, Tokyo, Japan - 
tsuruoka@ism.ac.jp 
Commission V, WG V/3 
KEY WORDS: Accelerometer, AR Modelling, Gyro Sensor, Measurement System, Movement, Spectral Analysis 
ABSTRACT: 
The walking sequence becomes less stable under the influence of 
analysed on the basis of two new approaches. The first approac 
a problematic part of the lower limb or back. This problem was 
h is a newly developed measurement system, consisting of two 
accelerometers for walking analysis (MSA). It was a small, wearable, personal computer-assisted system (WPC). The WPC was 
worn on a subject’s wrist, and the sensors were fixed to a subject’s knees. However, it was difficult to analyse directly the 
characteristics of movement stability using the raw measurement data. The second approach was an analysis of impulse response 
using AR modelling. In the case of a normal right knee, the response value of left knee was small, because the impulse seemed to be 
absorbed and wasn’t transmitted to the left knee, and vice versa. The value and duration of the impulse response curves have 
showed which knee is stronger and to what degree. It was observed that the movement of center body gravity (COG) was also 
influenced by physical and mental stress. The 1/f (f frequency) fluctuation, i.e., an index to stable walking was disappeared while 
the subjects had a high pulse rate or strong breathing as a result of stress. 
1. INTRODUCTION 
The fluctuation of movement, i.., the feedback control 
movement of left and right legs in the walking sequence 
becomes less stable under the influence of a problematic part of 
the lower limb or back. The CCD video cameras are used for 
the 3-D measurement of human movement. However, there are 
some problems: the measured distance is not enough for the 
walking analysis because cameras are fixed at place, and it is 
necessary to process many images. Also sometimes markers on 
joints seem to disappear suddenly against camera angles in 
walking Also sometimes markers on joints seem to disappear 
suddenly against camera angles while walking. 
For medical evaluation in rehabilitation, it is expected to 
measure without cords and to secure some extent of the walking 
distance and the ascending and descending of stairs in daily 
movement. This problem was analysed on the basis of two new 
approaches. The first approach is newly developed 
measurement system of human movement. They are small 
wearable personal computer (WPC)-assisted systems consisting 
of two accelerometers (Figure 1) or a gyro sensor. While 
walking or standing, the WPC is worn on a subject's wrist, the 
sensors are fixed to a subject's left and right knees or back, and 
batteries are inserted into a subject's waist pouch. | While 
walking, the time series data of accelerations in two axes of 
both of the subject’s knees or three angles of back are recorded 
into the WPC. 
However, it is difficult to analyse directly the characteristics of 
movement stability using the raw serial measurement data, 
because of their complex fluctuations in feedback control 
movement among their variables. The second approach is an 
analysis of impulse response or power spectrum using 
autoregressive (AR) modelling. AR modelling defines the best 
prediction. Giving the power spectral density, which expresses 
the characteristics of a sequential system concisely, 
decomposing it into periodic components. An analysis of 
impulse response was done with a particular focus on the 
feedback control movement: the movement of left and right 
knees in walking balance. Another analysis of power spectral 
was done a particular focus on the feedback control movement: 
the movement of center body gravity in standing balance after 
given physical and metal stress. 
   
  
  
  
5 Batteries 
WPC 
Right & Left Aaccelerometers 
Figure 1. Measurement of Walking Balance 
using two Accelerometers 
—281—