Motion Information Transmission for On-neck Communication

Takahide Ito

, Yuichi Nakamura

, Kazuaki Kondo

, Jonathan Rossiter

, Junichi Akita

and Masashi Toda

Graduate School of Engineering, Kyoto University, Kyoto, Japan

Academic Center for Computing and Media Studies, Kyoto University, Kyoto, Japan

University of Bristol & Bristol Robotics Laboratory, Department of Engineering Mathematics,

University of Bristol, Bristol, U.K.

Kanazawa University, Kanazawa, Japan

Kumamoto University, Kumamoto, Japan

akita@is.t.kanazawa-u.ac.jp, toda@cc.kumamoto-u.ac.jp

Keywords:

User Interface, Motion Communication, Wearable Device, Nursing, Electromyography, Skin-stretcher.

Abstract:

This paper introduces a novel form of communication via a combination of muscle sensing by electromyogra-

phy and stimulation via a skin-stretcher device as a motion monitoring system. After sensing muscle activity

through electromyography, the skin-stretcher device provides a skin sensation that conﬁdentially informs or

induces movements of the user who wears the device. This paper also introduces methods for translating mus-

cle activities to the skin-stretch sensations, and additional ﬁltering to improve the performance. In this study,

we conducted preliminary experiments that demonstrate the potential of our system design.

1 INTRODUCTION

We are developing a skin-stretcher device as a tool for

communication that uses skin sensations to transmit

motion information or induce motions. The device

gently and locally pushes or pulls the skin by contract-

ing or expanding itself via a link structure driven by

a servomotor. Although the device was originally de-

signed to induce a motion that is consistent with skin-

stretch sensation, it can be also used as a somatosen-

sory communication device.

Informing and prompting appropriate motions

provides intuitive support for sports, job training,

and rehabilitation (Kawasaki et al., 2006)(Spelmezan

et al., 2009). Communicating touch or movement

information provides an excellent tool for remote

care, nursing, and intimate communication for fam-

ily members (Doi et al., 2006)(Bentley and Metcalf,

2007). Sharing motion information can also support

group activities that require mutual awareness or joint

force with coordinated timing.

As an effective means for this type of communi-

cation, we focused on skin sensations, in particular,

the sensation of skin stretching. Our past experiments

have shown that good characteristics of the device,

e.g., the head of a person turns proportionally to the

length pulled by the device, can make the human-in-

the-loop design of a human-supporting system easy.

In this study, we focus on an application of our de-

vice to a motion monitoring system for care or nursing

scenarios. In the following sections, we introduce the

idea of motion communication, its design, the transla-

tion of head motion to device actions, and its perfor-

mance as measured via our experiments.

2 COMMUNICATION OF HEAD

MOTION

2.1 Importance of Head Motion

Head motion is a fundamental function for a variety

of movements in our daily lives. The human head

is continuously controlled to maintain an appropriate

position and pose.

A head motion is often an initial step for other im-

portant motions in addition to its own motions. When

we get out of bed, the motion of raising the head pre-

cedes other body motions. Head rotation often reveals

the internal states of a person. For example, we of-

ten expect that the attention of a person is directed

206

Ito, T., Nakamura, Y., Kondo, K., Rossiter, J., Akita, J. and Toda, M.

Motion Information Transmission for On-neck Communication.

DOI: 10.5220/0008417402060213

In Proceedings of the 3rd International Conference on Computer-Human Interaction Research and Applications (CHIRA 2019), pages 206-213

ISBN: 978-989-758-376-6

toward something frontal to the face (Clark, 1996).

Head motions, for example, of seeing a partner and

nodding are natural and essential behaviors in natural

conversations. In caregiving, head motions are one

of the most important signals that require the care-

ful attention of caregivers when monitoring patients.

We consulted staff members at Nishikagawa Hospital,

and they highlighted several actions that require care-

ful attention. For example, movements for rising or

standing need to be noted prevent patients accidents.

Details of the doctor and staff are presented in Ap-

pendix 6.

Even though there are other important body parts

for motion monitoring such as hands or legs, such

body parts move more dynamically and interact with

other objects (Ivanenko et al., 2004)(Kuan et al.,

2010). Such large movements require continuous and

intensive attention to grasp the status of patients.

Based on the above considerations, we investi-

gated head motion as a means of communicating per-

sons’ states.

2.2 Characteristics of Head Motion

Head motions primarily consist of rotation, ﬂexion,

and extension.

We ﬁrst focus on the mechanism of head rotation.

Head motions are caused primarily by the contraction

of the agonist and antagonist muscles around the neck

(Zangemeister et al., 1982). The dynamics of head ro-

tation can be roughly approximated by the following

simple formula:

θ + B

θ + Kθ = T (1)

where I, B, K, θ, and T represent the moment of in-

ertia, viscosity coefﬁcient, elasticity coefﬁcient, rota-

tion angle, and rotation torque of the head, respec-

tively.

We can see the rough relation between the rota-

tion angle and the muscle contraction that provides

the torque T . However, the antagonism of the mus-

cles reduces the rotation torque and increases the stiff-

ness, which is also an important mechanism for body

movements. The antagonism, viscosity, and elastic-

ity cause difﬁculties for direct estimations of the head

movement from the muscle measurements, e.g., elec-

tromyography; however, muscle activities can poten-

tially provide a rich source of information concerning

body movements, intention, and other states of a per-

son.

2.3 Requirements for Motion

Information Transmission

Let us consider the requirements or preferences for

motion information transmission by considering care-

giving applications, one of the promising applications

of such a system.

(a) Transmits Motion Quickly. Timing is one of

the most important factors for caregiving. Patient

movements that require a caregiver’s support or

that might result in dangerous situations need to

be informed as soon as possible; predictions may

even be necessary in some cases.

(b) Provides Natural Stimuli That Are Intuitively

Perceived. Noninvasive stimulation is preferable,

because the caregiver receiving information may

need to concentrate on their own activities.

caregiver is wearing a motion transmitting device

and the device applies a strong force that compels

the user to move counter to the user’s intention,

this may result in dangerous situations or acci-

dents. The device needs to allow the user to ignore

the stimulus and prioritize voluntary movements

if necessary.

(d) Conﬁdentially Transmit Information if Neces-

sary. Patient information is often conﬁdential and

is closed to unrelated persons. Audiovisual com-

munications are often inappropriate because they

may be observed by nearby persons.

These requirements are also considered important in

other situations, such as job training, joint projects,

and rehabilitation support.

Based on these considerations, we propose a novel

communication method by a combination of muscle

sensing via electromyography and stimulation via a

skin-stretcher device. With sensing muscle activity

through electromyography, we can directly capture

muscle contractions for movements and obtain clues

concerning the intentions of the movements. The

skin-stretcher device provides skin sensation that con-

ﬁdentially informs or induces movements of the user

who wears the device. Details are given in the follow-

ing sections.

3 SYSTEM DESIGN

Figure 1 shows an overview of our motion transmit-

ting system. The system comprises a motion-sensing

part using electromyography (EMG) and a motion-

indicating part using a skin-stretcher device.

Motion Information Transmission for On-neck Communication

207

DŽŶŝƚŽƌĞĚ

ƉĞƌƐŽŶ

DŽŶŝƚŽƌŝŶŐ

ƉĞƌƐŽŶ

D'ŵĞĂƐƵƌĞŵĞŶƚ

;^ƚĞƌŶŽĐůĞŝĚŽŵĂƐƚŽŝĚͿ

ĂůĐƵůĂƚĞ

ŵƵƐĐůĞĂĐƚŝǀĂƚŝŽŶ

^ŬŝŶͲƐƚƌĞƚĐŚĞƌ

^ĞůĞĐƚ

ĞŶŚĂŶĐĞŵĞŶƚ

ƚǇƉĞ ;tĂƚĐŚyͿ

ŽŶǀĞƌƐŝŽŶĨƌŽŵŵƵƐĐůĞĂĐƚŝǀĂƚŝŽŶ

ƚŽƐƚŝŵƵůĂƚŝŽŶ

dƌĂŶƐůĂƚĞƚŽ

ƐƚŝŵƵůĂƚŝŽŶƐŝŐŶĂů

&ŝůƚĞƌŝŶŐͬĞŶŚĂŶĐĞŵĞŶƚ

Figure 1: Overview of our motion transmitting system.

Figure 2: The skin-stretcher device attached to a user’s

neck.

3.1 Motion Indication

Figures 2 and 3 show the appearance and the conﬁgu-

ration of our skin-stretcher device, respectively.

The device is designed to provide a sensation of

skin stretching to convey motion information and to

induce a motion that is consistent with the sensation;

this is in contrast to previous studies such as Refs.

(Mizukami et al., 2007) or (Levesque et al., 2007).

Figure 3: Conﬁguration of the skin-stretcher device. The

end connector is attached to the neck with two sticky pads,

and the device is ﬁxed to the body with a body-mount har-

ness.

Figure 4: The wristwatch device (WatchX).

The primary features of the device are listed be-

low.

• Skin stretching provides a natural and intuitive

stimulus that directly but subtly indicates the di-

rection of the head rotation.

• The device does not apply a strong force that com-

pels the user to move counter to their own inten-

tions. The user can override the stimulus and pri-

oritize voluntary movements if necessary.

• The device can be used not only to induce motions

but also to impart the feeling of motion. The user

can feel the movement of the pads even if the user

does not move as requested by the device.

Despite the above advantages, the device has the

following problems.

• The sensation that the user receives may vary de-

pending on the user’s characteristics, activity, and

the device attachment. The stimuli may be too

small, too annoying, or need to be more focused

on speciﬁc movements.

• Communication via sensations is possible for one-

to-one communication; however, it is difﬁcult to

resolve if motion signals from multiple persons

are transmitted simultaneously and mixed. We re-

quire a mechanism for selecting the signal from

the speciﬁc person that we want to monitor.

To deal with these problems, we use a wristwatch

device (WatchX) that can be used to choose signals

for stimuli and the characteristics of stimuli, such as

gain.

WatchX is an Arduino-based device and has a

small display with three buttons as the output as

shown in Figure 4. It can communicate via Bluetooth

Low Energy.

3.2 Motion Sensing via EMG

We use myoelectric sensing (EMG) based on the fol-

lowing advantages.

• With EMG, we can measure the opposition by the

agonist and antagonist muscles, even if they do

not appear as actual joint torques.

CHIRA 2019 - 3rd International Conference on Computer-Human Interaction Research and Applications

208

Figure 5: EMG logging device and its host (Raspberry Pi).

Nƻ

23

Nƻ

X)

23

Nƻ

Nƻ

Nƻ

Nƻ

23$J$J&O(OHFWURGH

Nƻ

Nƻ

9

9

9

9

9

9

9

9

Figure 6: Schematic of the EMG amplifying circuit.

• EMG signals can be detected from earlier stages

of movements, often even before the body posture

changes.

• EMG signals suggest internal states, such as fa-

tigue or other body conditions.

Despite the above advantages, EMG sensing has

the following problems.

• We need to attach electrodes to the skin at the cor-

rect portions. Contacting the electrode affects the

measurements.

• As mentioned in Section 2.2, estimating the head

rotation angle using only EMG signals is difﬁcult.

Other complementary sensors may be added in fu-

ture designs to address these points.

We measure the sternocleidomastoid muscles,

which primarily provide head rotation torque and

tonus, as presented in previous studies (Nishimoto

et al., 1989)(Pejcic et al., 2016).

Figure 5 shows the EMG logging device and its

host (Raspberry Pi). Figure 6 shows the amplifying

circuit for the EMG.

We used the index of the muscle activation, which

is converted from the raw EMG signal, as follows

(Stroeve, 1996).

˙e =

u − e

(2)

˙a =



e−a

act

(e ≥ a)

e−a

deac

(e < a)

(3)

Here, u, e, and a indicate the absolute value of the

raw EMG signal, the intermediate value (the excita-

tion), and the muscle activation, respectively. From a

0 5 10 15

time(sec)

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0.1

0.2

0.3

0.4

EMG signal(mV)

-80

-60

-40

-20

left - rotation angle amount(degree) - right

quick head rotation

right sternocleidomastoid(activation)

left sternocleidomastoid(activation)

head roration angle

Figure 7: Example of raw EMG signals and the head ro-

tation angle, the red and black lines show the myoelectric

signals during head rotation.

0 5 10 15

time(sec)

normalized muscle activation - %MVC(%)

-80

-60

-40

-20

left - rotation angle amount(degree) - right

quick head rotation

right sternocleidomastoid(activation)

left sternocleidomastoid(activation)

head roration angle

Figure 8: Example of muscle activations and the head ro-

tation angle (converted from the example in Figure 7, the

red and black lines show the muscle activations during head

rotation.

previous study (Stroeve, 1996), we used τ

= 40ms,

act

= 10ms, and τ

deac

= 50ms for Eqs. (2) and (3).

Figures 7 and 8 show a sample of a raw EMG sig-

nal and its conversation result. The red and black lines

in Figures 7 and 8 show the signals of the right and

left sternocleidomastoid muscles, respectively, and

the blue line shows the head rotation angle measured

using a magnetic positional sensor.

4 TRANSLATION FROM EMG TO

STIMULATION

Sometimes the user (monitoring the motion informa-

tion) may need to quickly obtain every detail of the

motion information, while at other times, the user

may need to concentrate on their own activities. The

strength, timing, and characteristics of the stimuli

provided by the device need to change according to

the user’s needs.

Motion Information Transmission for On-neck Communication

209

^ƚŝŵƵůĂƚŝŽŶ

;/ŶŝƚŝĂůůĞŶŐƚŚͿ

;ĞǀŝĐĞůĞŶŐƚŚͿ

,ĞĂĚƌŽƚĂƚŝŽŶĐĂƵƐĞĚ

ďǇƚŚĞĚĞǀŝĐĞ

;/ŶŝƚŝĂů ƌŽƚĂƚŝŽŶͿ

;ƌĂĚŝƵƐͿ

;ŝƐƚĂŶĐĞͿ

;EĂƚƵƌĂůŚĞĂĚƌŽƚĂƚŝŽŶͿ

Figure 9: Geometry and action of the device. Calculation

of θ

is explained in Appendix 6.

Therefore, the system needs to provide a variety

of transmission modes and leave their selection to the

user.

4.1 Skin-stretch as Stimulation

The input and output of the system are the EMG sig-

nal and the stimulation via the skin-stretch device, re-

spectively.

Figure 9 shows the geometry of the skin-stretcher

device and the head. Based on the head rotation angle

of the monitored person, skin stretch is applied to the

monitoring user. If the monitoring user rotates their

head to an angle at which they feel no skin stretch,

this results in a similar head rotation to that of the

monitored person, even though such an action is not

requested most of the time.

Here, we use the notation in Eqs. (4) and (6) and

denote the head rotation angle of the monitored per-

son as θ

, and the natural head rotation angle at which

the monitoring user feels no skin stretch as θ

. We

calculated θ

as it is explained in Appendix 6, and

it represents the output stimulus. In addition, we as-

sume that the muscle activation and

have the same

dimension, as we mentioned in Section 2.2.

Based on the above discussion, we considered

several methods of translation from EMG to the skin

stretch as follows.

(1) Rotation Angle: The approximated rotation an-

gle θ

of the monitored person is given to the

monitoring user. θ

is approximated by the

second-order integral of torque

T d t which is

also approximated by the muscle activation

This is intuitive and easy to understand; however,

if the postures (the head rotation angles) of both

persons are very different, the stimulus may be

too extreme. Conversely, when the motion of the

monitored person is small, the stimuli often be-

come to be too small to be perceived.

(2) Rotation Angular Velocity: The angular veloc-

ity

can be approximated by

T d t, which can

also be approximated by

a. This method pro-

vides strong stimuli if the head rotation of the

monitored person is fast and weak stimuli if the

head rotation is slow. We expect that motions can

be characterized by their velocity and are less af-

fected by the pose changes of both persons. This

method may partially transmit motion informa-

tion; however, estimations of the actual rotation

angle are difﬁcult.

(3) Rotation Torque: The rotation torque T , which

is derived from a, is a rough approximation of the

angular acceleration. This method transmits de-

tails of even small motions, even though it does

not provide the actual head rotation angle. Con-

sequently, this method transmits every attempt at

movement but can be annoying to the monitoring

user.

In addition, we considered a method for transmit-

ting simultaneous contractions of both the right and

left sternocleidomastoid muscle.

(4) Co-contraction: Muscle activation on both

sides are transmitted, and both sides of the skin

are pushed to the front by the device. This method

also enables the transmission of motion attempts

for ﬂexion, extension, and a state in which the

neck is strained with the stiffness increased.

4.2 Filtering and the Emphasis of

Signals

The following problems arise from the utilization of

skin sensation via the simple attachment of a device

to the skin.

• The condition of the device installation may affect

the user’s perception.

• Small displacements of the skin are often imper-

ceptible, which results in a region of impercepti-

bility.

• The sensation of skin displacement gradually be-

comes imperceptible if no change occurs.

• Frequent or continuous stimulation can be annoy-

ing and feel unpleasant.

To handle those problems, we considered the ap-

plication of ﬁlters to selectively emphasize or smooth

the signals.

CHIRA 2019 - 3rd International Conference on Computer-Human Interaction Research and Applications

210

(a) Emphasizing. Considering the above problems

(1), (2), and (3), we consider a ﬁlter that empha-

sizes the output to a sufﬁcient level as follows.



(θ

> θ

)

(θ

≤ θ

)

(4)

Here, θ

, θ

, and θ

are the stimulus deﬁned

in the previous section, the emphasized stimulus,

and the threshold, respectively.

(b) Thining. The input signal can be quantized and

the frequency of the stimulated outputs can be de-

creased as follows.

= b

reso

c ·C

reso

(5)

Here, θ

, θ

, and C

reso

are the input, output, and

unit of quantization, respectively.

angle is large, the user requires muscle contrac-

tion to keep prevent rotation countering the elas-

ticity of the neck mechanism. This muscle con-

traction may result in a continuous stimulus. This

ﬁlter suppresses this effect by emphasizing the

initial change and attenuating the rest of the sig-

nal.



(t ≤ t

)

exp(−

t−t

dec

) (t > t

)

(6)

Here, θ

, θ

, t, t

, and τ

dec

are the input, out-

put, elapsed time after muscle activation, thresh-

old of the time, and time constant for attenuation,

respectively.

5 PRELIMINARY EXPERIMENTS

5.1 Experimental Setting

We conducted preliminary experiments in the follow-

ing setting. We asked one participant to perform the

following three behaviors as a monitored person and

measured the muscle activations using EMG.

(a) Simple Head Movement: Rotating the head

right and left, and shrugging the shoulders.

(b) Conversation: Rotating the head in a conversa-

tional situation.

motions.

Then, we asked the rest of the participants to wear

the device, discriminate the activity, and subjectively

evaluate the quality of the system based on the follow-

ing points; which task is performed, and their feeling

concerning the effectiveness and comfort of the stim-

ulation. During the experiment, participants did noth-

ing except feel the stimuli of the system while remain-

ing seated.

Typical responses from the participants concern-

ing the input-output translation methods were as fol-

lows.

• All translation methods worked well for simple

head movements, and the identiﬁcation of the task

was easy.

• It was easy to distinguish conversational behav-

iors from other behaviors.

• Identifying the behavior of getting up required

concentration because the motion happens over a

short period of time and the signals were small.

• The translation method (1) sometimes gave am-

biguous stimuli for identifying the standing up

and getting up tasks.

• Fast stimuli, i.e., pulling the skin faster, made

recognition easier.

As highlighted above, translation method (1) did

not provide sufﬁcient stimulations for complex or

continuous motions because small motions are often

difﬁcult to perceive with this method. Quantization

slightly improves the performance; however the per-

formance was not satisfactory compared to transla-

tion methods (2) and (3). We did not ﬁnd clear per-

formance differences between translation methods (2)

and (3); we need further experiments to address this

point.

For the ﬁltering methods, typical impressions of

the participants were as follows.

• The Emphasis ﬁlter helped with the perception of

small movements.

• Continuous small stimuli were unpleasant, and ﬁl-

tering (the thinning ﬁlter) was necessary.

• Emphasizing the initial part of the head move-

ments helped with the identiﬁcation of behaviors,

e.g., this helped with distinguishing actual head

rotations from increased stiffness for stabilization.

The above results imply that the emphasis ﬁlter

and quantization had the beneﬁcial effects that we ex-

pected. These experiments and results are prelimi-

nary and do not cover all the necessary situations for

our purposes. Further experiments for systematic ver-

iﬁcations and evaluations are necessary, and we are

planning to continue our investigation.

Motion Information Transmission for On-neck Communication

211

6 CONCLUSIONS

In this paper, we introduced a novel system for mo-

tion monitoring. The proposed system transmits mo-

tion information using muscle sensing via EMG and

skin sensations transmitted via the skin-stretcher de-

vice. We also introduced methods to translate muscle

activities to skin-stretch sensations as well as ﬁltering

to improve the performance of the device.

Our preliminary experimental results show the po-

tential of our system design.

We need further investigations with systematic ex-

periments to verify our framework and examine the

performance of our system.

ACKNOWLEDGEMENTS

We would like to thank Dr. Tomotake Otsuka and the

staffs in Nishikagawa Hospital for useful discussions

and suggestions.

REFERENCES

Bentley, F. R. and Metcalf, C. J. (2007). Sharing motion in-

formation with close family and friends. In Proceed-

ings of the SIGCHI Conference on Human Factors in

Computing Systems, CHI ’07, pages 1361–1370, New

York, NY, USA. ACM.

Clark, H. H. (1996). Using Language. ’Using’ Linguistic

Books. Cambridge University Press.

Doi, M., Inoue, H., Aoki, Y., and Oshiro, O. (2006). Video

surveillance system for elderly person living alone by

person tracking and fall detection. IEEJ Transactions

on Sensors and Micromachines, 126(8):457–463.

Ivanenko, Y. P., Poppele, R. E., and Lacquaniti, F. (2004).

Five basic muscle activation patterns account for mus-

cle activity during human locomotion. The Journal of

Physiology, 556(1):267–282.

Kawasaki, H., Kimura, H., Ito, S., Nishimoto, Y., Hayashi,

H., and Sakaeda, H. (2006). Hand rehabilitation sup-

port system based on self-motion control, with a clin-

ical case report. In 2006 World Automation Congress,

pages 1–6.

Kuan, J., Huang, T., and Huang, H. (2010). Human inten-

tion estimation method for a new compliant rehabil-

itation and assistive robot. In Proceedings of SICE

Annual Conference 2010, pages 2348–2353.

Levesque, V., Pasquero, J., and Hayward, V. (2007).

Braille display by lateral skin deformation with the

stress2 tactile transducer. In Second Joint EuroHap-

tics Conference and Symposium on Haptic Interfaces

for Virtual Environment and Teleoperator Systems

(WHC’07), pages 115–120.

Mizukami, Y., Uchida, K., and Sawada, H. (2007). Tac-

tile transmission by higher-level perception using vi-

bration of shape memory alloy thread. IPSJ journal,

48(12):3739–3749.

Nishimoto, K., Kobayashi, S., Hashimoto, T., Okubo, M.,

Ohori, T., Kikunaga, H., Nishimoto, H., and Koda, T.

(1989). Surface electromyographic analysis for sitting

from supine, side lying and prone positions. Physical

Therapy Japan, 16(5):317–322.

Pejcic, N., Djuric-Jovicic, M., Miljkovic, N., Popovi

c, D.,

and Petrovi

c, V. (2016). Posture in dentists: Sitting

vs. standing positions during dentistry work - an emg

study. Srpski arhiv za celokupno lekarstvo, 144:181–

187.

Spelmezan, D., Jacobs, M., Hilgers, A., and Borchers, J.

(2009). Tactile motion instructions for physical ac-

tivities. In Proceedings of the SIGCHI Conference

on Human Factors in Computing Systems, CHI ’09,

pages 2243–2252, New York, NY, USA. ACM.

Stroeve, S. (1996). Learning combined feedback and feed-

forward control of a musculoskeletal system. Biolog-

ical Cybernetics, 75(1):73–83.

Zangemeister, W. H., Stark, L., Meienberg, O., and Waite,

T. (1982). Neural control of head rotation: Elec-

tromyographic evidence. Journal of the Neurological

Sciences, 55(1):1–14.

APPENDIX

Opinions of Doctors and Caregiving

Staffs in Nishikagawa Hospital

We consulted doctors and the nursing staffs members

in Nishikagawa Hospital in the city of Mitoyo, Ka-

gawa, Japan. We discussed the purpose and require-

ments of motion monitoring and obtained the follow-

ing opinions.

• The Detecting and Support of Getting Up,

Standing Up, and Starting Walking Behaviors

Are Extremely Important. Elderly people tend

to have accidents during the initial steps of these

behaviors, and they have a lower likelihood of

accidents after they have passed the initial steps.

Therefore, motion monitoring of these behaviors

would greatly help nursing or caregiving staff

members.

• Give Support Only When Necessary. Ideal

caregiving involves letting patients act by them-

selves except in situations in which they really

need or want help. Motion motoring is also use-

ful in that patients can behave naturally and care-

givers can notice situations where assistance is

necessary. This contributes to maintaining the pa-

tient’s self-esteem and quality of life.

CHIRA 2019 - 3rd International Conference on Computer-Human Interaction Research and Applications

212

Calculating the Natural Head Rotation

Figure 9 shows the geometry of the device and the

head. The device is initially attached with the con-

dition that the extension rod is parallel to the tangent

of the neck. We approximate the relation between the

natural head rotation and the device length by the fol-

lowing formulas:

+ x

= L

(7)

cos



− θ



= −

+ L

− x

2RL

(8)

where R is the approximated radius of the neck, L is

the distance between the center of the device and the

center of the neck, x and x

are the actual and the ini-

tial length of the skin-stretcher device, respectively.

is the head rotation at which the user feels no skin

stretch. It is represented by the angle between the ori-

entation from the center of the neck toward the pad

and the vertical axis as shown in Figure 9. θ

is the

initial head rotation.

As we consider the above mechanism as a man-

machine system, the natural head rotation angle of

monitoring person θ

is the head rotation angle re-

quested by the skin displacement, which is caused

by θ

− θ

. We do not directly use the rotation angle

of the servo motor for calculating the input value be-

cause the extension rod and the servo motor are con-

nected by the elastic link, i.e., the rotation angle of

the servo motor and the displacement of the rod do

not hold stable relationships. Alternatively, the dis-

placement ∆x is directly measured by the potentiome-

ter installed in the device. Thus the natural head rota-

tion angle θ

can be estimated using ∆x by the above

equations.

Motion Information Transmission for On-neck Communication

213