ULTRASONIC SENSORS FOR THE ELDERLY AND CAREGIVERS

IN A NURSING HOME

Toshio Hori

Digital Human Research Center, National Institute of Advanced Industrial Science and Technology & CREST, JST.

2-41-6, Aomi, Koto-ku, Tokyo 135-0064, JAPAN

Yoshifumi Nishida

Digital Human Research Center, National Institute of Advanced Industrial Science and Technology & CREST, JST.

2-41-6, Aomi, Koto-ku, Tokyo 135-0064, JAPAN

Keywords:

Intelligent nursing home, ultrasonic 3D tag system, location awareness.

Abstract:

Workloads on caregivers in nursing home are increasing as the imbalance between the number of elderly

people and that of caregivers becomes larger. Excessive workloads on caregivers must be reduced not only

because they become burdens for caregivers but also because they deteriorate the quality of nursing care. One

of such workloads is routine patrol for monitoring the status of the elderly and for detecting accidents on the

elderly as soon as possible. If the number of unnecessary patrols is minimized, caregivers will be able to spend

their time on high touch care and humane communication. The authors have been developing an ultrasonic 3D

tag system which locate ultrasonic tags in real time, and employed the system in a nursing home to monitor

positions of the elderly people. If the system locates the elderly people continuously and robustly, and if it

can notify caregivers about the occurrence of accident-prone activities promptly, caregivers will be releaved

from their unnecessary workloads. This paper describes the research background, system overview, system

implementations, and experimental results.

1 INTRODUCTION

The world has been rapidly aging during the last few

decades and it is projected that this trend will con-

tinue for several more decades. Aging of population

has become one of the most serious and urgent prob-

lems that almost all the countries are facing and must

tackle. Problems incurred by aging are not only the

problem of decreasing working populations but also

the problem related to the shortage of caregivers for

elderly people. The number of the elderly of whom

each caregiver takes care in nursing homes is certainly

increasing and the imbalance between the number of

the elderly and that of caregivers will deteriorate the

quality of nursing care.

In nursing home, caregivers patrol the entire home

regularly to prevent the elderly from meeting with se-

rious accidents and, even if they have accidents, to de-

tect occurrence of accidents promptly. Besides such

patrolling, caregivers must often visit nursing rooms

where nurse calls sounded from to take care of inhab-

itants. The number of caregivers is usually much less

than that of the elderly so they are very busy taking

care of the elderly. There is an overwhelming need

for alleviating such workloads on caregivers.

On the other hand, the authors are involved in re-

search on “human behavior modeling”(Hori, 2001;

Nishida et al., 2004) and have been developing an en-

vironment in which a system observes inhabitants and

recognizes their behaviors(Kitamura et al., 2004). By

installing such observation system into elderly care

homes, we believe that the system can recognize the

behaviors of elderly people and provide some useful

support for caregivers. To be concrete, if the system

can detect accident-prone activities of the elderly be-

forehand and notify caregivers of the occurrence, their

excessive patrols for detecting accidents or accident-

prone activities will be minimized and, consequently,

caregivers will be able to spare their time for high

tough care for and humane communication with the

elderly.

This paper introduces an ultrasonic sensor system

for a nursing home. The system uses ultrasonic sen-

sors to observe activities of the elderly people in a

nursing room. The system is composed of two sub-

systems, a wheelchair locator subsystem and an ultra-

sonic radar subsystem; the former works for tracking

position of a wheelchair which an elderly person uses

and the latter monitors human head’s position on and

around bed. When these two subsystems work coop-

110

Hori T. and Nishida Y. (2005).

ULTRASONIC SENSORS FOR THE ELDERLY AND CAREGIVERS IN A NURSING HOME.

In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 110-115

DOI: 10.5220/0002556201100115

 SciTePress

eratively, activities of the elderly in a nursing room are

monitored continuously and remotely without invad-

ing his/her privacy, so their accidents will be avoided

or minimized if the system can recognize and notify

the accident-prone activities to caregivers beforehand.

This paper is organized as follows: Section 2

presents several works related to location systems and

support technologies. Secondly, Section 3 describes

background of the system and outline of the system,

brieﬂy. Then Section 4 presents the system imple-

mentations and several experimental results and, ﬁ-

nally, Section 5 concludes this paper.

2 RELATED WORKS

Concerning location systems and their technologies,

Hightower and Borriello presented a good survey and

taxonomy in IEEE Computer(Hightower and Bor-

riello, 2001).

The Active Bat system(Harter et al., 1999) devel-

oped at AT&T Cambridge is the system which is al-

most same to ours. The system uses ultrasonic emit-

ters, Active Bats, and a grid of ceiling-mounted ultra-

sonic receivers. Bats are attached to objects or per-

sonnel and their positions are computed with the ac-

curacy of approximately 3cm.

The Cricket Location Support System(Priyantha

et al., 2000) takes the opposite approach. In this

system, ultrasonic emitters are installed in the envi-

ronment and receivers are embedded in the objects.

So triangulation computation is performed on the ob-

jects’ side and, therefore, privacy of humans who have

objects is not invaded.

As for support technologies, many interesting arti-

cles are posted in IEEE Pervasive Computing.

Stanford introduced an elder care home, the Oat-

ﬁeld Estates(Stanford, 2002). The home had em-

ployed IR and RF wireless communication technolo-

gies not only for surveillance but also for monitoring

health condition of the elderly.

The CareMedia project is an on-going research

project at CMU(Hauptmann et al., 2004). They in-

stalled four cameras and microphones in a nursing

home in Pittsburgh and recorded video images over a

week in their preliminary study. Analyzing the video,

they tried to track the elderly’s movement and to ex-

tract their activities in the nursing home.

SIMBAD (Smart Inactivity Monitor using Array-

Based Detectors) is a sensor system developed by Six-

smith(Sixsmith and Johnson, 2004). They utilize IR

sensor arrays as thermal imaging detectors to observe

(in)activities of inhabitants.

Similar systems are too numerous to mention. We

believe that this is because these support technologies

are getting a lot of attention all over the world.

TouchSensor

FloorMat

Sensor

(a) Touch sensor on a

handrail of a bed

(b) Floor mat sensor at

bedside

Figure 1: Sensors used in a nursing home

3 ULTRASONIC SENSOR

SYSTEM FOR ELDERLY CARE

HOMES

This section ﬁrstly presents the background of this

project. Then the Ultra Badge System developed by

the authors is introduced brieﬂy and, lastly, we outline

our proposed system for elderly care homes.

3.1 Background of the Project

The authors started the “Sensorized Elderly Care

Home” project in the mid-summer of 2003 and have

been implementing a system which reduces work-

loads on caregivers using a sensor network. This is

a cooperative project with a nursing home in Tokyo.

To begin with, we interviewed caregivers of the

nursing home and categorized accidents of the elderly

into the following two cases:

1. Falls—Users of wheelchair or walker tend to fall

down when they transfer from their wheelchair/

walker to their bed or a toilet seat and vice versa.

2. Wanders—Elderly people who suffer from senile

dementia, such as Alzheimer’s disease, tend to

wander around day and night.

To detect these accidents, several kinds of contact-

type sensors are already used in the nursing home. For

example, touch sensors (Figure 1(a)) were attached to

handrails of beds and ﬂoor mat type pressure sensors

(Figure 1(b)) were put at bed side. Both sensors had

been used for detecting the elderly falling down from

his/her bed or trying to sneak out of the bed. However,

those sensors had little effect on preventing accidents

from occurring because: (1) the elderly, though suf-

fering from senile dementia, often learn how to escape

being detected by the sensors, (2) the sensors often

emit false alarms when, for example, a person turns

over and touches the sensor by chance while sleeping,

and (3) the sensors detect the occurrence of accidents

but cannot prevent accidents from occurring even if

they work correctly.

ULTRASONIC SENSORS FOR THE ELDERLY AND CAREGIVERS IN A NURSING HOME

111

(a) Tiny type (1×1×2cm;

comparison with an US

cent coin)

(b) Long life battery type

(Active for 2mon. with

Li-Ion battery)

Figure 2: Variety of Ultra Badges

According to the caregivers, more than 85% of the

residents are suffering from senile dementia so they

cannot expect the residents to put on any devices.

Therefore, we have to monitor the elderly without

putting any devices on their bodies. And, moreover,

cameras are unacceptable because they may invade

the privacy of the elderly.

3.2 The Ultra Badge System

The authors have been developing an ultrasonic 3D

tag system(Nishida et al., 2004; Hori et al., 2003)

which tracks positions of multiple ultrasonic tags con-

tinuously for monitoring human behaviors. We call it

Ultra Badge System.

The Ultra Badge System consists of ultrasonic re-

ceivers embedded in the environment and wireless

small ultrasonic emitters. We named the emitter Ul-

tra Badge. Figure 2 shows two types of Ultra Badges

developed by the authors. The left picture is a tiny

type whose size is 1cm×1cm×2cm, and the right one

is a long life battery type which uses a Li-Ion battery

of mobile phone and is active for two months using

sleep mode.

By embedding many receivers whose 3D positions

are known beforehand in an environment, the system

can obtain a badge’s position from distance data using

multi-lateration technique. Experiments conducted in

our laboratory exhibited that the accuracy of position

estimation was about 20 to 80mm and the resolution

was about 15mm in horizontal directions and about

5mm in vertical direction by using a robust estimator,

RANSAC(Fischler and Bolles, 1981).

3.3 Outline of the Ultrasonic Sensor

System for Detecting Falls in a

Nursing Home

From the interview with caregivers, we summarized

the targets of our sensor system as follows:

• The system prevents falls from occurring in the

nursing home,

Detection

Areas

Ultrasonic

Signalfor

Locator

Ultrasonic

Signalfor

Radar

Ultrasonic

Receivers

Network

Ultrasonic

Emitters

Nurse

Call

Figure 3: Schematic diagram of the Ultrasonic Sensor Net-

work System for a nursing home

• The system monitors beds, bedsides and the side of

toilet seats because the locations where falls occur

are relatively limited to those places,

• The system monitors the elderly remotely and does

not expect them to put any devices on their bodies,

• The system expects the elderly to use wheelchairs

or walkers,

• The system uses an existing nurse call system to

send alarms to caregivers when it detects accident-

prone activities, and

• The system must respect and must not invade the

privacy of the elderly.

We designed an Ultrasonic Sensor Network System

for the Nursing Home based on the above consider-

ations. Figure 3 is a schematic diagram of the sys-

tem. The system consists of (1) embedded ultrasonic

receivers, (2) embedded/wireless ultrasonic emitters

(Ultra Badges) and (3) a sensor network which con-

nects receivers and emitters. In addition to these main

components, the system uses a synchronous signal

generator to synchronize the emitters/receivers inter-

nal timers, an RF transmitter to activate wireless emit-

ters, and a PC for data processing.

The system is composed of two subsystems: wheel-

chair locator and ultrasonic radar. The wheel-

chair locator subsystem uses Ultra Badges attached

to wheelchairs to monitor their position remotely and

the ultrasonic radar subsystem uses ultrasonic emit-

ters embedded in the ceiling to monitor activities of

the elderly on their beds. The subsystems are con-

nected to a nurse call system and they call caregivers

whenever they detect predeﬁned events leading to ac-

cidents. Details of these subsystems are described in

Section 4.

In this system, information transmitted through

the air is wireless emitters’ IDs only, and ultrasonic

pulses sent from any emitters do not carry any infor-

mation on the individuals being monitored. The sys-

ICEIS 2005 - HUMAN-COMPUTER INTERACTION

112

Ultrasonic

Receivers

Ultrasonic

Receiver

Ultrasonic

3D Tag

Figure 4: Ultrasonic receivers on the ceiling Figure 5: Enlarged view of a

receiver

Figure 6: A wheelchair and an Ul-

tra Badge attached

tem does not collect unnecessarily rich information,

such as the images taken by cameras. We can there-

fore conclude that the level of privacy invasion is min-

imized in this ultrasonic sensor network system.

4 IMPLEMENTATIONS OF THE

SYSTEM AND EXPERIMENTAL

RESULTS

This section presents a wheelchair locator subsystem

and an ultrasonic radar subsystem in detail. The aim

of the former subsystem is to detect the wheelchair

approaching one of areas where falls are likely to oc-

cur and the aim of the latter is to detect the elderly

getting up and going out of their beds.

4.1 Wheelchair Locator Subsystem

The authors installed the Ultra Badge System for

monitoring positions of a wheelchair in one room of

the nursing home. We embedded 99 ultrasonic re-

ceivers on the ceiling of the room. Appearance of the

ceiling and an enlarged view of a receiver are shown

in Figure 4 and 5, respectively. All the other appara-

tus are hidden behind the ceiling so the room looks

the same as the other rooms except for the ceiling.

An Ultra Badge was attached to the wheelchair that

the subject of this experiment uses. Figure 6 shows

the wheelchair and an Ultra Badge attached to the

back of the seat. The system activates the badge at

5 Hz and tracks its position continuously.

In the room, we deﬁned detection areas in front of

the bed that the subject uses and at the entrance of

a toilet. The subsystem is connected to a nurse call

system and, when it detects the wheelchair entering

one of the areas, it notiﬁes caregivers that the subject

is approaching the place where accident is likely to

occur. Though we have not measured the time from

the nurse call alarm to caregivers’ arrival, it seems that

Toilet

Wash-

stand

Bed

Figure 7: Trajectory of the wheelchair

both the position of detection areas and the timing of

alarm work for caregivers to support the subject.

4.2 Experimental Results of the

Wheelchair Locator Subsystem

This subsystem has been running since January, 2004,

and taking logs of wheelchair’s position. Figure 7 is

its trajectory obtained during the operation. The line

segments are the trajectory and purple dots are the re-

ceivers’ positions on the ceiling. Yellow and blue rec-

tangles indicate bed positions and detection areas, re-

spectively. The subject uses the lower-left bed so we

preset the detection areas in front of the bed and at the

entrance of the toilet.

This experimental data was obtained from 8:00AM

to 10:00AM on Aug. 15th, 2004 and, according to the

log, the system correctly sent alarms when the wheel-

chair got into either detection area. So we may say

that the system was almost working ﬁne.

ULTRASONIC SENSORS FOR THE ELDERLY AND CAREGIVERS IN A NURSING HOME

113

250

200

345

250

220

900

1800

1200

receiver

transmitter

Ultrasonic Radar

Figure 8: Ultrasonic emitters/receivers on the ceiling

During the operation, the system detected the other

events that even caregivers did not know at all. One

of such events that the system disclosed was the fol-

lowing:

In one night, the system sent alarms several

times so a caregiver went to the nursing room to

ﬁnd that the subject was going to the toilet with

her wheelchair.

The contact-type sensors used for the subject could

not have detected such events at all so no caregivers

had thought the subject went out of bed in the night.

But, with the assistance of this system, they found that

the subject had symptoms of frequent urination.

Such sneaking out could not be detected with a

wearable-type device if the subject (deliberately) for-

got to put the device on. Cameras could have detected

the events if their installation was permitted in the

nursing room, but they need someone to keep watch-

ing all monitors to detect such events. So we are con-

ﬁdent that this is a good example which exhibits the

effectiveness and the superiority of our system.

4.3 Ultrasonic Radar Subsystem

The wheelchair locator subsystem is a system to lo-

cate wheelchair so, theoretically, it cannot detect the

activity of the elderly on their beds. But, of course,

falls from bed must also be avoided and we need to

monitor the activity on beds to detect such accidents.

So we developed this ultrasonic radar subsystem.

The subsystem comprises several ultrasonic emit-

ters and receivers embedded on a ceiling. The sub-

system activates emitters one by one and measures

time-of-ﬂight of ultrasonic pulses, just the same as the

wheelchair locator subsystem. When the system ob-

tains the distance data, the reﬂection point is located

on a spheroid whose foci are the emitter and the re-

ceiver. So the position of the reﬂection point is cal-

culated by solving simultaneous equations of at least

three spheroids.

Figure 9: Evaluation of localization accuracy of human

head

x[cm]

y[cm]

z[cm]

200

100

150

130

140

150

160

170

location of a human head

x [cm]

y [cm]

z [cm]

200

100

150

130

140

150

160

170

Figure 10: Tracking a head of moving human

Figure 8 depicts our experimental setup. In this

setup, we set 18 ultrasonic emitters and 32 receivers

on a panel, and placed it as a ceiling at 230 cm height

from the ﬂoor level. In this experimental system, the

sampling frequency is 1 Hz, i.e. every emitter is acti-

vated every one second in turn.

This subsystem is still under development and is

not installed in the nursing home. So the experimental

results presented in the next subsection were obtained

by this experimental setup.

4.4 Experimental Results of the

Ultrasonic Radar Subsystem

The accuracy of localization was measured to eval-

uate the ability to detect the human head location.

Figure 9 shows the results, where the blue dots and

the red crosses indicate true positions of the head and

positions measured by the subsystem, respectively.

All the distance errors between corresponding points

were within 5 cm so we think that the system has suf-

ﬁcient accuracy to locate human head.

As the subsystem calculates the reﬂection point pe-

riodically, it can obtain a trajectory of a human head,

though the resolution is not so precise because of the

sampling rate. Figure 10 illustrates the tracking re-

sults. The upper parts show the trajectories when the

subject moved as shown in the lower parts.

ICEIS 2005 - HUMAN-COMPUTER INTERACTION

114

Figure 11: Coordination of the wheelchair locator subsys-

tem and the ultrasonic radar subsystem

4.5 Integration of Two Subsystems

When these two subsystems work cooperatively, they

can monitor the activities of the elderly on beds and

in the room. We integrated them into an experimen-

tal system shown in Figure 11. The ﬁgure illustrates

an example of cooperation of two subsystems. In the

ﬁgure, a person is sitting on a bed and transferring to

a wheelchair. The red rectangle indicates the position

of his head and the white cross indicates the position

of the Ultra Badge attached to the wheelchair.

One of the advantage of this integration is that

both subsystems use the same ultrasonic technology;

that is, installing a large-scale ultrasonic sensor net-

work system sufﬁces both subsystems’ requirements.

The total system is expected to work for detecting

accident-prone activities and for preventing accidents

from occurring in a nursing home.

5 CONCLUSIONS

This paper presented an ultrasonic sensor network

system for nursing homes. We developed two subsys-

tems. The wheelchair locator subsystem was installed

in a nursing home and has been running successfully.

The ultrasonic radar subsystem is still under develop-

ment but will be installed in the nursing home soon as

its experimental results satisﬁed our requirements.

The authors think that accidents on the elderly will

be avoided or minimized when both subsystems work

cooperatively; i.e., the total system monitors the el-

derly, recognizes accident-prone activities and noti-

ﬁes their occurrence to caregivers beforehand. If this

sensor network system works in place of caregivers,

their workloads are alleviated and they will be able to

spare their time for high touch care for the elderly.

The authors believe that nursing homes will be

fully sensorized in the near future.

ACKNOWLEDGMENT

The authors wish to express our great gratitude to Mr.

Yuichi Motomura at the nursing home Aizenen for his

great help on conducting this project. We express our

appreciation to Mr. Hiroshi Aizawa and Mr. Shin’ichi

Murakami for their fruitful research results.

REFERENCES

Fischler, M. A. and Bolles, R. C. (1981). Random sample

consensus: A paradigm for model ﬁtting with app li-

cations to image analysis and automated cartography.

Communications of the ACM, 24(6):381–395.

Harter, A., Hopper, A., Steggles, P., Ward, A., and Webster,

P. (1999). The anatomy of a context-aware applica-

tion. In Proceedings of the 5th Annual ACM/IEEE

International Conference on Mobile Computing and

Networking, pages 59–68.

Hauptmann, A. G., Gao, J., Yan, R., Qi, Y., Yand, J., and

Wactlar, H. D. (2004). Automated analysis of nursing

home observations. Pervasive Computing, 3(2):15–

21.

Hightower, J. and Borriello, G. (2001). Location systems

for ubiquitous computing. Computer, 34(8):57–66.

Hori, T. (2001). Overview of digital human modeling. In

Tutorial Note, Workshop on “Digital Human Model-

ing”at the IEEE/RSJ International Conference on In-

telligent Robots and Systems 2001.

Hori, T., Nishida, Y., Kanade, T., and Akiyama, K. (2003).

Multi-lateration for multiplexed ultrasonic sensors. In

Proceedings of the 2nd IEEE International Confer-

ence on Sensors, pages 1219–1224.

Kitamura, K., Nishida, Y., Kimura, M., and Mizoguchi, H.

(2004). Real world sensorization and virtualization

for observing human activities. In Proceedings of the

6th International Conference on Enterprise Informa-

tion Systems 2004, volume 5, pages 15–20.

Nishida, Y., Kitamura, K., Hori, T., Nishitani, A., Kanade,

T., and Mizoguchi, H. (2004). Quick realization of

function for detecting human activity events by ultra-

sonic 3d tag and stereo vision. In Second IEEE Inter-

national Conference on Pervasive Computing, pages

43–54.

Priyantha, N. B., Chakraborty, A., and Balakrishnan, H.

(2000). The cricket location-support system. In Pro-

ceedings of the 6th Annual International Conference

on Mobile Computing and Networking, pages 32–43.

Sixsmith, A. and Johnson, N. (2004). A smart sensor to

detect the falls of the elderly. Pervasive Computing,

3(2):42–47.

Stanford, V. (2002). Using pervasive computing to deliver

elder care. Pervasive Computing, 1(1):10–13.

ULTRASONIC SENSORS FOR THE ELDERLY AND CAREGIVERS IN A NURSING HOME

115