Novel hands-free interaction techniques based on the software switch approach for computer access with head movements [original]

Vol.:(0123456789)

1 3

Universal Access in the Information Society

https://doi.org/10.1007/s10209-020-00748-1

LONG PAPER

Novel hands‑free interaction techniques based onthesoftware switch

approach forcomputer access withhead movements

CagdasEsiyok1 · AyhanAskin2· AliyeTosun2· SahinAlbayrak1

Abstract

Head-operated computer accessibility tools (CATs) are useful solutions for the ones with complete head control; but when it

comes to people with only reduced head control, computer access becomes a very challenging task since the users depend on

a single head-gesture like a head nod or a head tilt to interact with a computer. It is obvious that any new interaction technique

based on a single head-gesture will play an important role to develop better CATs to enhance the users’ self-sufficiency and

the quality of life. Therefore, we proposed two novel interaction techniques namely HeadCam and HeadGyro within this

study. In a nutshell, both interaction techniques are based on our software switch approach and can serve like traditional

switches by recognizing head movements via a standard camera or a gyroscope sensor of a smartphone to translate them

into virtual switch presses. A usability study with 36 participants (18 motor-impaired, 18 able-bodied) was also conducted to

collect both objective and subjective evaluation data in this study. While HeadGyro software switch exhibited slightly higher

performance than HeadCam for each objective evaluation metrics, HeadCam was rated better in subjective evaluation. All

participants agreed that the proposed interaction techniques are promising solutions for computer access task.

Keywords Interaction techniques· Universal access· Inclusive design· Switch access· Computer access· Head-operated

access· Software switch· Switch-accessible interface· Head tracking· Hands-free computer access

1 Introduction

According to the World Report on Disability[1] in 2011, it is

estimated that there have been about one billion people with

several disabilities. Besides, about 2% of the world popula-

tion—between 110 and 190 million people—have severe dis-

abilities in functioning. People with motor-impairments—as

a result of amyotrophic lateral sclerosis, carpal tunnel syn-

drome, spinal cord injury or degenerative diseases—require

assistive technology solutions to have a more independent

life. CATs are considered as one of the most efficient exam-

ples of these solutions enabling hands-free computer access.

They are generally based on human–computer interaction

(HCI) techniques where a mouse cursor is controlled by the

user’s complete head control ability. But when it comes to

people with only reduced head movement (i.e., the ones who

cannot operate the mouse cursor by moving head), computer

access becomes a very challenging task since the users have

to interact with a computer by a single head-gesture like a

head nod or a head tilt.

Computers have become indispensable tools with their

immense services in our increasingly digitalized world.

Unfortunately, most people with only minimal head move-

ment lack these services, since they have difficulties to

interact with their computers by means of current solutions.

The World Report on Disability[1] also reveals that 80%

of people with disabilities live in low- and middle-income

countries, which means that the majority of people with only

minimal head movements might not afford most hands-free

HCI solutions[2–4], since they are generally depending on

expensive devices. Although the aim of universal access is

enabling equal opportunity and access to a service or prod-

uct regardless of people’s physical disabilities by reducing

barriers, the high-cost of most current solutions creates a

new barrier financially for the majority of target group. On

the other hand, according to the International Labour Organ-

isation (ILO) statistics[5] published in 2007, an estimated

* Cagdas Esiyok

cagdas.esiyok@dai-labor.de

1 Distributed Artificial Intelligence Laboratory, Technische

Universität Berlin, Berlin, Germany

2 Physical Medicine andRehabilitation Department, Izmir

Katip Celebi University, Izmir, Turkey

Universal Access in the Information Society

1 3

470 million of the world’s working age people live with sev-

eral disabilities. Although there have been many jobs which

are dependent only on computer usage like software coding,

exclusion of millions of working age people with disabili-

ties from the labor force leads to an increase in the Gross

Domestic Product (GDP) lost worldwide. Furthermore, they

lack a paid job which makes them feel more independent by

affording themselves financially. It is obvious that any new

HCI technique based on a single head-gesture will play an

important role to develop better CATs to enable these people

operate a computer for a more inclusive and barrier-free life.

In accordance with our efforts to find a solution for people

with only reduced head control to interact with a computer

by a single head-gesture, we began with a review of the cur-

rent head-operated solutions in Related Works section. We

noticed that the majority of interaction techniques requires a

complete head control ability. In other words, there are lim-

ited solutions which are capable of supporting single head-

gesture access for people with reduced head movements.

As a result of our literature review, we identified two major

problems of current head-operated interaction techniques

with a single head-gesture access support: (1) requirement of

dedicated devices, (2) compatibility with switch-accessible

interfaces. To overcome these problems, we employed our

software switch approach of which first examples were pre-

viously presented in Esiyok etal.’s study[6].

We proposed two novel interaction techniques namely

HeadCam and HeadGyro by following the principles of

the software switch approach. Both interaction techniques,

major problems of the current solutions, and our soft-

ware switch approach were explained in detail in Software

Switches section. In a nutshell, both interaction techniques

can serve like traditional switches by recognizing the head

movements via a standard camera or a gyroscope sensor of

a smartphone to translate them into virtual switch presses.

Furthermore, they do not require a dedicated device and are

compatible with most of switch-accessible interfaces. As

low-cost alternatives, they can be replaced with expensive

traditional head switches for computer access. They are also

capable of recognizing any motion of the other body parts,

such as the user’s shoulder or leg, which makes them quite

flexible switches. By this way, different physical gestures

can be targeted easily, when the user becomes tired. Besides,

both proposed software switches do not require physical

strength to be activated unlike physical switches; especially

HeadGyro can even detect a minimal head movement to

transform it into an emulated switch press.

A usability study with 36 participants (18 motor-

impaired, 18 able-bodied) was conducted in order to col-

lect objective and subjective evaluation data. The SITbench

1.0[7] benchmark was employed for objective evaluation.

Moreover, we also applied a System Usability Scale (SUS)

[8] questionnaire for subjective evaluation. While HeadGyro

showed slightly higher performance than HeadCam for each

objective evaluation metrics, HeadCam was rated better than

HeadGyro in subjective evaluation. All participants agreed

that the idea of controlling a computer via a single head-

gesture without requiring any dedicated device sounded very

promising.

Given that the majority of the current solutions requires

expensive dedicated devices, and that 80% of people with

disabilities live in low- and middle-income countries[1],

proposed software switches are expected to have a con-

siderable impact. Currently, they are the only options for

people with reduced head control (i.e., those who have to

use a switch-based system for computer access) who cannot

afford any dedicated device. On the other hand, considering

there have been many jobs which are dependent only on

computer usage like software coding, any tool for computer

access undoubtedly helps these people to participate in the

labor force, which will result in a decrease in the global

GDP lost. Furthermore, the ones who can perform a paid-job

will feel like they are more independent by affording them-

selves financially. Also, software switches can be employed

as alternative inputs for multi-modal HCIs beyond assis-

tive technology related purposes. Since HeadGyro software

switch is not affected by external factors like light or wind, it

could also be employed for outdoor activities (e.g., operating

a wheelchair).

This paper proceeds with the Related Works section to

summarize the current head-operated interaction techniques

for computer access. In the Software Switches section, we

identify the common problems of current interaction tech-

niques and introduce our software switch approach with two

software switches called HeadGyro and HeadCam proposed

within this paper. Subsequently, we evaluate both interaction

techniques by presenting objective and subjective evalua-

tion results of our usability study in the Evaluation section.

Finally, we conclude and discuss our study in the Conclusion

and Discussion section.

2 Related works

In this section, from a broader perspective, we reviewed the

current head-operated HCI solutions that provide alternative

means for computer access. We preferred to separate them

into two main groups according to the condition whether

they have a single head-gesture access support.

2.1 Head‑operated interaction techniques

withoutasingle head‑gesture access support

Interaction techniques in this group require a complete

head control ability for hands-free computer access. In

Universal Access in the Information Society

1 3

principle, they translate the users’ head movements into

mouse cursor movements in several ways:

One of the most popular techniques is wearing iner-

tial sensors, such as a gyroscope or an accelerometer

on the head (via a helmet or a cap) to control a mouse

pointer[9–19]. These inertial sensor-based systems are

mostly combined with a different sensor/switch to perform

a mouse click task (e.g., in a way that head movements are

detected by inertial sensors to control mouse pointer, and

mouse clicks are performed by a puff switch). Another

sensor-based solution called Headmaster Plus[20], which

was evaluated in the work by LoPresti etal.[21], consists

of ultrasonic sensors. Briefly, the user wears a headset

including three ultrasonic sensors that wait an ultrasonic

signal from a stationary transmitter on the user’s computer.

In this way, ultrasonic sensors determine the orientation

of the user’s head to convert them into mouse pointer

coordinates.

Using a head pointer—a head-worn stick in princi-

ple—is another solution which permits the users to con-

trol, press or touch any target[22] by head, although this

method is rarely preferred nowadays. Similarly, head-oper-

ated joysticks are alternative tools which enable the users

to point mouse cursor on the screen[23].

On the other hand, a specific part of the user’s face

(e.g., the tip of the nose) or the user’s whole head can be

tracked by a standard camera in order to transform head

movements into mouse cursor movements on a computer

screen[24–44]. Mouse click tasks, such as left or right

click, are generally performed with the dwelling method

(i.e., the user holds the mouse cursor steady for a given

amount of time to perform the click tasks) or with multi-

modal approaches by means of other gestures like eye-

blinks or tooth-clicks.

In addition to the above-mentioned approaches, head

movements can also be followed by special camera-based

systems to control a mouse cursor. In such systems, the

user wears small reflective dots on his/her head/face or an

infrared LED (light-emitting diode) which is placed on a

helmet or a pair of glasses. These reflective dots are illu-

minated by an infrared or near infrared light source, and

then a standard camera[45–47] or an infrared camera[48]

tracks the position of target signals (coming from reflec-

tive dots or an infrared LED) for mouse cursor pointing.

On the other hand, RGB-D cameras as new vision sensor

technologies are also able to do 3D mapping of head posi-

tion to control mouse pointer[49].

2.2 Head‑operated interaction techniques

withasingle head‑gesture access support

For those with only reduced head control, there have

been limited solutions which are able to support single

head-gesture access. Using a traditional button switch via

a scanning interface is a common technique where a head

switch is mounted close to the user’s head in a way that the

user can hit it by tilting the head (or by any activity moving

head)[50, 51]. In addition to traditional hardware switches,

there are just a few software-based solutions[52, 53] dem-

onstrating mouse click tasks with a single head-gesture. In

software-based solutions, first the users are enabled to navi-

gate the mouse cursor to the desired location by vision-based

head tracking methods, and then mouse clicks are emulated

according to the users’ head-gestures as an alternative to

dwelling method.

3 Software switches

This section begins with a subsection which explains how

we handle the detected problems of current interaction tech-

niques by applying our software switch approach. Then, we

introduce the common user interface of both software switch

proposed. Afterward, HeadCam and HeadGyro software

switches are also explained, respectively.

3.1 The software switch approach

Our software switch approach has two principles: an inter-

action technique based on our software switch approach (1)

should not require any dedicated devices, and (2) should be

configurable to be compatible with switch-accessible inter-

faces. By following these principles, we proposed two inter-

action techniques within this study. Detected major problems

of the single head-gesture compatible interaction techniques

(in Sect.2.2) and proposed solutions based on our software

switch approach are presented below:

1. Requirement of Dedicated Devices: The majority of cur-

rent solutions for computer access depend on dedicated

devices which might be hard to afford for the ones living

in low- and middle-income countries[2–4]. The high-

cost of dedicated devices leads to a new financial bar-

rier. Any new efficient solution based on an expensive

device will not make any sense for these people, unless

proposed solutions are affordable for them. Therefore, as

the first principle of our software switch approach, inter-

action techniques for people with reduced head control

should not require any dedicated device beyond standard

computer peripherals like a microphone or a camera. At

this point, as the only reasonable exception, we decided

to exclude smartphones from the dedicated devices list,

because the total number of smartphones—3.2 billions

in 2019[54]—got ahead of the total number of comput-

ers in recent years worldwide[55], which makes them

easy to access for people in even low-income countries.

Universal Access in the Information Society

1 3

Besides, smartphones are capable of providing several

services to the users unlike dedicated devices which

are produced with a specific aim. To sum up, while

software-based solutions[52, 53] do not require any

dedicated devices, traditional button switches are dedi-

cated devices beyond standard computer peripherals. As

low-cost solutions, HeadCam and HeadGyro software

switches are based on a standard camera and a gyro-

scope sensor of a smartphone, respectively;

2. Compatibility with Switch-accessible Interfaces: The

majority of current solutions reported in literature are

only compatible with a specific switch-accessible inter-

face. To make it clear, first the mechanism of a scanning-

based interface and standardization problem should be

understood. In principle, unlike direct selection (such

as typing on a keyboard), the scanning interface high-

lights items one-by-one on the computer screen, and

the user activates the switch when the desired item is

highlighted. Between switch-accessible interface and

the switch, there is a switch adapter which is a dedi-

cated device to transform switch activation signals into

meaningful keyboard presses or mouse clicks. Following

a switch activation, switch adapter emulates a specific

keyboard character or a mouse click event (depending on

the manufacturer of switch interface) and send it to the

computer in order to communicate with switch-accessi-

ble interface. But the main problem in this case is that

there has not been any commonly agreed standard for

the communication between switches and switch-acces-

sible interfaces; while some switch- accessible interfaces

expect to receive a specific keyboard character like

space, the others expect to receive a mouse click. This

standardization problem is partially solved by a switch

driver software permitting the users to assign a specific

character or mouse click—following a switch activa-

tion—which is expected by the target switch-accessible

interface. However, these switch driver software are only

compatible with a limited number of switch adapters

of specific brands, which makes them partial solutions

for the standardization problem. In other words, each

switch adapter requires its specific switch driver soft-

ware. Although current software-based solutions[52,

53]—which are able to emulate mouse clicks—support

single head-gesture and do not require any dedicated

device, they are only compatible with specific switch-

accessible interfaces which can be controlled with a

mouse click as a switch input signal. To the best of our

knowledge, there is not any complete solution for this

standardization problem in literature. Both interaction

techniques proposed within this study can be configur-

able to generate any expected keyboard characters or

mouse clicks, which makes them compatible with most

switch-accessible interfaces. They provide a better

solution to the standardization problem than the current

solution where a switch driver and a traditional switch

are required to purchase. In other words, they are able to

both detect a head-gesture like a traditional switch and

allow the users to assign the expected keyboard charac-

ters or mouse clicks—which will be sent to the switch-

accessible interface—like a switch driver.

3.2 The user interface

We designed an interface as shown in Fig.1a which

was employed for both software switches. Gamifica-

tion techniques were applied to make software switches

more engaging and fun. An initial state of the interface—

where the user has a stable head position—can be seen

in Fig.1a. The interface includes three dynamic game

elements: (1) the earth, (2) the left and (3) the right red

border lines. All three elements can be controlled by

the user’s head movements called pitch, yaw and roll

(Fig.1b). Sensitivity to control the game elements can be

set according to the user’s head control capability. As the

Fig. 1 a The initial state of

the interface of both software

switches. b Rotational move-

ments of a head

Universal Access in the Information Society

1 3

sensitivity level gets higher, the user can move the game

elements with a slower and minor head movement. The

mission of the game is to save the earth from the grav-

ity of a black hole by moving these three game elements

until the earth intersects with the red border lines. Switch

press and switch release are emulated according to this

intersection situation. In other words, as soon as the earth

intersects with the red border lines, a switch press is emu-

lated until the end of intersection, while a switch release

is emulated once the intersection between the earth and

the red border lines is terminated. The intersection (i.e.,

switch press) is followed by a visual or an auditory sen-

sory feedback provided to the user. In order to calibrate

the earth’s position, we simulated a gravity function that

pulls the earth toward the black hole constantly. The grav-

ity function becomes ineffective during intersection (i.e.,

switch press). Once the intersection is over (i.e., switch

release), the gravity function is reactivated. In this way, if

the user keeps his/her head stable for a while when there

is not any intersection, the earth will be pulled to its ini-

tial position eventually by gravity (i.e., to the center). As

illustrated in Fig.2, each of six different head-gestures

(i.e., rotational movements of the head) results in six dif-

ferent intersection states. While pitch (Fig.2a) and yaw

(Fig.2b) movements control the earth’s position, roll

movements (Fig.2c) operate the position of the right and

the left red border lines.

3.3 HeadCam

HeadCam is based on a real-time video motion tracking

algorithm which is similar with the study by Esiyok etal.

[6]. In principle, the user’s head is tracked by a built-in cam-

era or a standard web-cam to translate the roll movements

of the user’s head (as can be seen in Fig.2c) captured by

the camera into an emulated switch press. Before launch-

ing HeadCam application, in the configuration step, the user

assigns the color of the tracked object through a RGB (red,

green, blue) sphere with specified radius for Euclidean color

filtering. The algorithm of HeadCam is listed step-by-step

below:

– Video frames are taken by a camera with a frame rate

of 15 frames per second and a frame size of

320 ×240

pixels (Fig.3a);

– Euclidean color filtering is applied for each video frames

(Fig.3b). By this way, Euclidean color filtering filters the

colors outside of the RGB sphere with specified center

and radius which are assigned at configuration step. In

other words, it keeps the pixel within the specified color

sphere and fills the other remaining pixels with the black

color;

– Following Euclidean color filtering, video frames are

converted to gray-scale images (Fig.3c);

– All objects are detected in video frames through the

Connected Component Labeling (CCL) method which

groups together pixels belonging to the same connected

component and treats them as separate objects. Follow-

ing object detection, for each object, a rectangle is drawn

according to the edge of the object (Fig.3d);

– The greatest object (i.e., the one whose rectangle has the

largest area) is chosen if there is more than one object

detected (Fig.3e);

– The center point of the rectangle of the greatest object is

tracked in real-time on the frame (Fig.3f);

– Every motion of the greatest object (i.e., center point of

the rectangle) is transformed into the motion of the right

or left red border lines as it is depicted in Fig.2c;

– Once the earth intersects with the red border lines, a

switch press is emulated.

Fig. 2 Six different intersection states of the interface according to rotational movements of a head

Universal Access in the Information Society

1 3

An image processing library called AForge.NET was

employed for filtering (Euclidean color filtering) and object

detection (CCL). HeadCam is compatible with Windows-

based operating systems and was developed under .NET 4.5

framework. Two roll movements of the user’s head (right

and left head tilts) can be easily recognized by HeadCam,

which makes our software switch capable of supporting dou-

ble switch inputs for switch-accessible interfaces.

3.4 HeadGyro

HeadGyro interaction technique, basically, employs 3-axis

gyroscope data of a smartphone—where the smartphone is

placed on the user’s head—to convert the rotational move-

ments of the user’s head into emulated switch presses. The

smartphone can be placed on the user’s head in several ways.

For example, the user can wear a cap which is attached to

the smartphone or a modified belt holding the smartphone as

can be seen in Fig.4. The gyroscope is an important inertial

sensor and mainly used to measure angular velocity of the

sensor in inertial space. In other words, it measures the rate

of change of the sensor’s orientation. Today, inertial sen-

sors like gyroscope are based on microelectromechanical

system (MEMS) technology. They are employed in mod-

ern smartphones frequently since they are small, cheap,

light, and offer low power consumption. In spite of all these

advantages, because of the electromagnetic interference and

the influence of semiconductor thermal noise, MEMS based

solutions might suffer from noise, which affects the accuracy

of the detected angular velocity. We preferred the Kalman

filter, which is a frequently used method in literature[56–59]

for gyroscope data considering the real-time requirements,

to avoid the noise. We also developed a mobile application

depending on the Android operating system—which com-

municates with the computer in a wireless local area network

(WLAN)—to convey the stream 3-axis gyroscope data to

the computer. As can be seen in Fig.1, roll, pitch, and yaw

movements are represented by the angular velocity around

each 3-axis of coordinate system as X, Y, and Z, respec-

tively. The algorithm behind HeadGyro is briefly described

step-by-step below:

– Real-time angular velocity data originated from smart-

phone’s 3-axis gyroscope sensor is drawn by our Android

application;

– The Android application streams this gyroscope data,

which holds three different angular velocity measure-

Fig. 3 Steps of head tracking algorithm: a take video frames via cam-

era; b apply Euclidean color filter for each frame; c convert video

frames to gray-scale; d detect all objects in each frame by connected

component labeling method; e choose the greatest object for each

frame; f track the position of the greatest object

Fig. 4 The placement of the smartphone on the user’s head for Head-

Gyro software switch

Universal Access in the Information Society

1 3

ments from 3-axes (X, Y, Z), wirelessly to HeadGyro

software switch running on computer;

– For each channel (X, Y, Z), the Kalman filter is applied

to reduce the noise as shown in Fig.5;

– Every motion of the user’s head is recognized according

to filtered angular velocity measurements from 3-axes

in HeadGyro, and these measurements are converted

into the motion of the game elements as illustrated in

Fig.2. For example, if the angular velocity originated

from z-axis is measured as a positive value, then the earth

moves to the left side relatively; while it moves to the

right side if the measured angular velocity value is nega-

tive;

– Once the earth intersects with the red border lines, a

switch press is emulated.

For Kalman fiter, we employed the MathNet.Filtering

library. Like HeadCam software switch, HeadGyro is also

compatible with Windows-based operating systems, and it

was developed in the .NET 4.5 framework. It can provide up

to six switch inputs for switch-accessible interfaces, since

all six rotational head movements can be easily detected by

HeadGyro.

4 Evaluation

A usability study was conducted to collect objective and

subjective data. In this section, firstly we introduce the char-

acteristics of participants. Then, we present the apparatus

used within this study. Afterward, we briefly explain the

SITbench 1.0 and the procedure applied during the evalua-

tion of HeadCam and HeadGyro. At last, we conclude the

section with our experimental findings.

4.1 Participants

Following the approval by the Ethics Committee of the Izmir

Katip Celebi University (Turkey) on 10.10.2018 (decision

number: 332), the usability study was conducted at Medi-

cal Faculty of the University (Turkey). All participants gave

their informed consent before they participated in the study.

Consent for publication of human images in this article

was also received. A total of 36 participants, including 18

females and 18 males, took part in the evaluation of the pro-

posed systems. While the disability group (DG) comprises

18 participants with motor-disabilities whose ages ranged

between 18 and 68, the control group (CG) without disabili-

ties includes 18 people (12 females, 6 males) whose ages

ranged between 18 and 59.

In Table1, age statistics of all participants are summarized

according to groups. We also summarize the main character-

istics of all participants in Table2. As an inclusion criteria, all

voluntary participants in DG had several difficulties control-

ling their heads and thus could not operate a computer with

conventional ways (i.e., with a mouse and a keyboard). They

were all under medical treatment for several motor disabilities,

while the experiments were conducted. On the other hand,

Fig. 5 Two different stream data graphs based on the x-axis of the gyroscope sensor of two different participants when participants nod head.

Blue and red lines represent unfiltered and Kalman filtered gyroscope data, respectively

Table 1 Age statistics of the participants according to groups

Group Gender Mean age Number of

participants

Mix Mix 43.2 (sd = 15.3) 36

Mix Female 39.3 (sd = 14.2) 18

Mix Male 47.1 (sd = 15.7) 18

DG Mix 46.1 (sd = 17.3) 18

DG Female 38.0 (sd = 19.5) 6

DG Male 50.1 (sd = 15.4) 12

CG Mix 40.3 (sd = 12.8) 18

CG Female 39.9 (sd = 11.8) 12

CG Male 41.3 (sd = 15.7) 6

Universal Access in the Information Society

1 3

voluntary participants of CG were generally accompanies of

DG or staff working at the Physical Medicine and Rehabilita-

tion Department. All participants met the following inclusion

criteria: they were able to (1) find a target on the screen; (2)

follow a moving target; (3) maintain gaze on a stable target; (4)

stay focused on tests during experiments. All participants in

DG had difficulties to control their hands. Besides, there were

five participants in DG with reduced head control. We also

applied the mini-mental state examination (MMSE)—30-point

questionnaire for cognitive assessment—to validate whether

the participants can meet the cognitive ability to complete our

tests.

4.2 Apparatus

A laptop computer (Lenovo G505S; CPU: AMD A8-4500M

1.9 GHz; RAM: 6 GB DDR3; screen: LCD 15.6; OS: Win-

dows 10 64 bits; resolution:

1600 ×900

), an integrated camera

(max digital video resolution:

1280 ×720

; Image Sensor Type:

0.3 MP CMOS), and a smartphone with a gyroscope sensor

(Sony Xperia XZ1 Compact; CPU: Qualcomm Snapdragon

835; RAM: 4GB; OS: Android Oreo 8.0) were employed for

the experiments.

4.3 The SITbench 1.0 benchmark

We used the SITbench 1.0[7] benchmark which helps

researchers to evaluate switch-based systems objectively. By

means of this tool, objective evaluation data can be collected

and saved automatically with standardized tests. To this end,

we employed the Tie-Smiley Matching Game (TSMG) and

Hungry Frog Game (HFG) tests of the SITbench 1.0.

4.3.1 TSMG

Briefly, TSMG is a switch-accessible interface based on the

automatic linear scanning method where each smiley is high-

lighted one-by-one for a given scan time. It includes five dif-

ferent templates. As can be seen in Fig.6, the scanning array

of each template consists of 26 smileys in total. Count and

order of red and yellow smileys differ for each template. As an

indirect selection, the user activates the switch when the high-

lighted smiley is the red one. A click sound is also provided

to the user as an auditory prompt once the target red smiley is

highlighted. The mission of the game is to match each smiley

with a tie of the same color (i.e., red to red, yellow to yellow).

To achieve this, the user activates the switch only if the high-

lighted smiley is the red one. Figure6 shows a sample view

after the user completed a trial. Confusion matrix variables as

true positives (TP), false positives (FP), false negatives (FN),

and true negatives (TN) are counted automatically. Then all

performance metrics as accuracy, precision, recall and false-

positive rate are calculated by the SITbench 1.0 according to

the following formulas:

(1)

accuracy

TP +TN +FP +FN

(2)

precision

TP +FP

(3)

recall

TP +FN

Table 2 Main characteristics of the participants

The user Age Gender Disability

DG1 68 Male Hemiplegia

DG2 21 Female Hemiplegia

DG3 59 Male Hemiplegia

DG4 27 Male Tetraplegia

DG5 58 Male Hemiplegia

DG6 57 Male Hemiplegia

DG7 31 Female Hemiplegia

DG8 18 Female Hemiplegia

DG9 34 Female Hemiplegia

DG10 63 Female Hemiplegia

DG11 53 Male Hemiplegia

DG12 65 Male Hemiplegia

DG13 38 Male Hemiplegia

DG14 62 Male Hemiplegia

DG15 61 Female Hemiplegia

DG16 34 Male Hemiplegia

DG17 23 Male Hemiplegia

DG18 58 Male Hemiplegia

CG19 18 Female None

CG20 51 Female None

CG21 43 Female None

CG22 55 Male None

CG23 22 Male None

CG24 27 Male None

CG25 59 Female None

CG26 25 Female None

CG27 32 Male None

CG28 41 Female None

CG29 50 Female None

CG30 28 Female None

CG31 40 Female None

CG32 34 Female None

CG33 45 Female None

CG34 45 Female None

CG35 58 Male None

CG36 52 Male None

Universal Access in the Information Society

1 3

4.3.2 HFG

HFG is the other single-switch-accessible test of the SIT-

bench 1.0 (Fig.7). In a nutshell, a trial includes ten tasks,

and each task is achieved in a way that (a) the user does not

move until a fly appears on the screen, (b) the user acti-

vates the switch as fast as possible once the fly is appeared,

ten tasks of a trial are completed, the SITbench 1.0 meas-

ures the following six evaluation metrics automatically: (1)

average press time of all ten tasks, i.e., the average time

from when the fly appears to when the switch is pressed;

(2) average release time of all ten tasks, i.e., the average

time from when the switch is pressed until it is released; (3)

(4)

false positive rate

FP +TN

the fastest press time within ten tasks; (4) the slowest press

time within ten tasks; (5) the fastest release time within ten

tasks; (6) the slowest release time within ten tasks. HFG

includes five different scenarios. For each scenario of HFG,

waiting times (i.e., the time between when the user starts to

wait the appearance of a fly and when the fly appears on the

screen) differ.

4.4 The SUS questionnaire

The SUS questionnaire[8], which is an industry standard,

consists of ten statements with a five-point Likert scale as

can be seen in Table3. Scale values range from 1 to 5 (1 =

strongly disagree, 2 = disagree, 3 = neither agree nor disa-

gree, 4 = agree, 5 = strongly agree). A SUS score (ranging

from 0 to 100) is calculated based on scale value of the

statements in a way that: (1) score contributions of each

statement are summed where the score contribution is the

scale value minus 1 for statements 1, 3, 5, 7, 9; the score

Fig. 6 A general view of TSMG following a user performance with several mistakes (i.e., with false negatives and false positives)

Fig. 7 A view of HFG in the

end of a trial

Universal Access in the Information Society

1 3

contribution is 5 minus the scale value for statements 2, 4,

6, 8, 10; (2) the sum of the score contributions is multiplied

by 2.5 to calculate the SUS score.

4.5 Procedure

At the beginning, the participants were informed about the

test verbally. Then, we ensured that the participants and

devices were positioned properly. Following a proper posi-

tioning, we let them practice the tests (in a counterbalanced

order) under our guidance, until they feel confident to start

the tests. Afterwards, we applied two tests of the SITbench

1.0 to collect objective data: (1) TSMG: each software

switch was tested by each participant (

n=36

) with the first

three templates of TSMG where scan time was 1000 milli-

seconds; (2) HFG: each software switch was tested by each

participant (

n=36

) with the first three scenarios of HFG.

We applied the tests in the counterbalanced order to avoid

learning and repetition effects. In order to prevent mental

or physical fatigue, we allowed the participants to get rest

up to 5min between the experiments. For each participant,

it took 15–30min to complete the experiments including

breaks. We have not observed any fatigue in any period of

the experiments. At the end of the SITbench 1.0 experi-

ments, we also applied the SUS questionnaire to the par-

ticipants for quantitative subjective evaluation. Besides, we

collected the qualitative subjective data via our observations

and participants’ responses of open-ended questions about

two software switches proposed within this study.

4.6 Objective data based results

As can be seen in Fig.8, according to the results of the

TSMG experiments, HeadGyro demonstrated slightly better

performance than HeadCam in all performance evalua-

tion metrics (accuracy, precision, recall, and false-positive

rate). In terms of accuracy, the mean value of HeadGyro

(

m=0.938

) was greater than HeadCam (

m=0.904

), and

the difference between mean values was found statistically

significant (p < 0.05) according to Student’s t-test for both

software switches. For precision, HeadGyro (

m=0.921

)

exhibited better performance than HeadCam (

m=0.872

and there was a significant difference between means (p <

0.05). Regarding recall, HeadGyro (

m=0.910

) was fol-

lowed by HeadCam (

m=0.863

) with a significant difference

between means (p < 0.05) of both interaction techniques.

For false-positive rate, HeadCam (

m=0.077

) was ahead of

HeadGyro (

m=0.048

), and the difference between means

was significant (p < 0.05).

Figure9 presents the mean values of each software switch

for TSMG depending on the participant groups (Mix, DG,

CG). CG members performed better than DG members for

both software switches according to the mean values through

accuracy, precision and recall evaluation metrics. In false

positive rate score, DG had higher scores than CG for soft-

ware switches, which means that DG members made false

selections more frequently when compared to CG members.

The Student’s t-tests for both interaction techniques through

all evaluation metrics was applied to check whether there

is a significant difference between the performance of DG

members and CG members. The difference between means

between DG and CG was not significant for all metrics.

Likewise, HeadGyro proved a better performance in

comparison to HeadCam for all HFG evaluation metrics

(Fig.10) (average press time, average release time, the fast-

est press time, the slowest press time, the fastest release

time, and the slowest release time). Mean and p-values of

both interaction techniques are presented in Table4 based

Table 3 SUS questionnaire statements with average scale values through participant groups

𝛥

symbol was replaced with HeadGyro and HeadCam, respectively, during assessment

Statements HeadGyro average scale HeadCam average scale

Mix / DG / CG Mix / DG / CG

1. I think that I would like to use

𝛥

frequently 4.11 / 4.16 / 4.06 4.07 / 4.10 / 4.04

2. I found

𝛥

unnecessarily complex 1.16 / 1.10 / 1.22 1.14 / 1.10 / 1.18

3. I thought

𝛥

was easy to use 4.41 / 4.32 / 4.50 4.30 / 4.25 / 4.35

4. I think that I would need the support of a 2.22 / 2.09 / 2.35 1.52 / 1.45 / 1.59

technical person to be able to use

𝛥

5. I found the various functions in

𝛥

were well integrated 4.30 / 4.32 / 4.28 4.30 / 4.26 / 4.34

6. I thought there was too much inconsistency in

𝛥

1.19 / 1.23 / 1.15 1.22 / 1.21 / 1.23

7. I would imagine that most people would learn to use

𝛥

very quickly 4.33 / 4.24 / 4.42 4.33 / 4.27 / 4.39

8. I found

𝛥

very cumbersome to use 1.41 / 1.31 / 1.51 1.11 / 1.10 / 1.12

9. I felt very confident using

𝛥

3.97 / 3.92 / 4.02 4.27 / 4.21 / 4.33

10. I needed to learn a lot of things before I could get going with

𝛥

1.13 / 1.10 / 1.16 1.13 / 1.14 / 1.12

Universal Access in the Information Society

1 3

on HFG experiments. According to p-values based on the

Student’s t-test results of all participants for both interac-

tion techniques, it is demonstrated that there is a statistically

significant difference between the means of HeadGyro and

HeadCam through all evaluation metrics.

4.7 Subjective data based results

Results of the SUS questionnaire as quantitative subjective

data are listed in Table3. The average scale values acquired

from all participants are represented according to HeadGyro

and HeadCam through participant groups as mix, DG, and

CG. For mix group, the average SUS scores are calculated

as 85.0 and 87,9 for HeadGyro and HeadCam, respectively.

In DG, the average SUS score is 85,3 for HeadGyro, while

it is 87.8 for HeadCam. On the other hand, in CG, the aver-

age SUS score is calculated as 84.7 for HeadGyro, while it

is calculated as 88.0 for HeadCam. According to the SUS

adjective rating scale[60], all SUS scores can be considered

as excellent. After the experiments, all participants agreed

that both proposed interaction techniques are promising

solutions for computer access tasks. They also declared

that they were looking forward to experience both software

switches to control a computer. Regarding to experiments

with the SITbench 1.0, five participants stated that they

would perform better if the scanning speed of the TSMG test

was set to a slower value, while four participants suggested

to increase the size of smileys. All participants were pleased

with the visual and auditory sensory feedback provided to

the user during tests once the switch is activated or the target

is appeared. While 31 of all participants declared that they

would prefer to use HeadCam for computer access, 5 of them

chose HeadGyro as their favorite software switch. They all

agreed that gamification techniques made software switches

more engaging. None of the participants experienced any

fatigue during tests.

5 Conclusion anddiscussion

Hands-free computer access via head movements is already

a challenging task in comparison to conventional ways, but

when it comes to people with limited head control, computer

access becomes a more challenging task since the users are

obliged to interact with a computer by a single head-gesture

like a head nod or a head tilt. On the other hand, the high-

cost of dedicated devices—employed by the majority of

current head-operated HCI solutions—creates a new bar-

rier, although the aim of universal access is to break the

barriers to enable equal opportunity and access for people

with disabilities.

Alternative computer access methods can provide several

useful services for people with motor disabilities in every

part of life, such as communication and education. Any

new interaction techniques enabling computer access with

minimal head movements will obviously help to enhance the

quality of life and the self-sufficiency of people with reduced

head control ability alone. Therefore, we proposed two novel

interaction techniques namely HeadGyro and HeadCam

which depend on the gyroscope sensor of a smartphone and

a standard camera, respectively. Both interaction techniques

are based on our software switch approach that provides a

comprehensive solution to the following problems of the

current single head-gesture based interaction techniques:

(1) requirement of dedicated devices and (2) compatibility

with switch-accessible interfaces. In accordance with the

two principles of our software switch approach, HeadGyro

and HeadCam software switches (1) do not require any dedi-

cated devices and (2) are configurable to be compatible with

switch accessible interfaces. In a nutshell, both software

switches can serve like traditional switches by recognizing

head movements via a standard camera or a gyroscope sen-

sor of a smartphone to transform them into virtual switch

presses.

Fig. 8 Mean values of interaction techniques acquired from all participants through evaluation metrics of TSMG including accuracy, precision,

recall, and false positive rate (*p < 0.05)

Universal Access in the Information Society

1 3

Fig. 9 Mean values of the software switches through evaluation metrics (accuracy, precision, recall, false positive rate) according to the partici-

pant groups (Mix, DG, CG)

Universal Access in the Information Society

1 3

According to the evaluation data of the conducted usabil-

ity study with 36 participants (18 motor-impaired, 18 able-

bodied), HeadGyro showed slightly better performance than

HeadCam in objective evaluation, while HeadCam was rated

better than HeadGyro in subjective evaluation. Furthermore,

31 of all participants declared that they would prefer to use

HeadCam for computer access, while 5 of them selected

HeadGyro. Based on our observations, the reasons behind

this situation are as follows: (1) The head control ability is

the key factor for this situation. Those who have complete

head control ability (31 participants) rated HeadCam, while

the ones with reduced head control (5 participants) pre-

ferred HeadGyro since it is more sensitive and thus capable

of recognizing tiny head movements; (2) Those with com-

plete head control can easily activate the software switch

via a standard camera. As expected, wearing a smartphone

on head was found an unnecessary solution by the partici-

pants as long as their head control capability remains unim-

paired or their head movements can be detected by Head-

Cam. However, HeadGyro can be advantageous if (1) the

users cannot move their head enough to be recognized by

a camera, or (2) the external factors (e.g., low/high light or

any moving object behind the user) cannot be tolerated by

camera-based tracking. As can be concluded from the results

of objective evaluation, HeadGyro works in a more sensitive

way in comparison to HeadCam.

Both software switches can serve as the only low-cost

options for people with limited head control who can-

not afford the systems depending on high-cost dedicated

devices. Beyond head motions, proposed software switches

can be quite flexible by recognizing the other body motions

to transform them into emulated switch presses. This flex-

ibility also permits the user to change the targeted body

motion once the user becomes tired. On the other hand,

proposed software switches can also be employed by multi-

modal systems as new input techniques beyond the assistive

technology area (e.g., as a new input for a computer video

game). As another application domain, HeadGyro software

switch might be preferred during outdoor activities, since it

is quite durable against external factors like low light, high

noise, and air conditions. As a future work, any other physi-

cal gesture—which is well-controlled by the user—can be

targeted to evaluate the efficiency and usability of the pro-

posed interaction techniques. Both software switches can

also be employed by a single-switch accessible CAT to see

their performance in a real-life scenario.

Acknowledgements Open Access funding provided by Projekt DEAL.

We would like to thank Dr. Brijnesh Jain and Dr. Fikret Sivrikaya for

their valuable suggestions. The authors are grateful to the participants

for their valuable time. The first author holds the Ministry of National

Education Scholarship of the Turkish Republic.

Fig. 10 Mean values of two software switches for all participants through HFG evaluation metrics (average press time, the fastest press time, the

slowest press time, average release time, the fastest release time, and the slowest release time) (*p < 0.05; **p < 0.01)

Table 4 Mean values of HeadGyro and HeadCam through HFG eval-

uation metrics (average press time, the fastest press time, the slow-

est press time, average release time, the fastest release time, and the

slowest release time) for all participants

Metric type HeadGyro HeadCam

Average press 0.514 0.582

The fastest press 0.402 0.424

The slowest press 0.670 0.775

Average release 0.204 0.255

The fastest release 0.140 0.176

The slowest release 0.281 0.342

Universal Access in the Information Society

1 3

Compliance with ethical standards

Conflicts of interest On behalf of all authors, the corresponding author

states that there is no conflict of interest.

Open Access This article is licensed under a Creative Commons Attri-

bution 4.0 International License, which permits use, sharing, adapta-

tion, distribution and reproduction in any medium or format, as long

as you give appropriate credit to the original author(s) and the source,

provide a link to the Creative Commons licence, and indicate if changes

were made. The images or other third party material in this article are

included in the article’s Creative Commons licence, unless indicated

otherwise in a credit line to the material. If material is not included in

the article’s Creative Commons licence and your intended use is not

permitted by statutory regulation or exceeds the permitted use, you will

need to obtain permission directly from the copyright holder. To view a

copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.

References

1. World report on disability. https ://web.archi ve.org/web/20191

11011 2641/https ://www.who.int/disab iliti es/world _repor t/2011/

repor t.pdf. Accessed 24 Nov 2019

2. Li, W., Sellers, C.: Improving assistive technology economics for

people with disabilities: harnessing the voluntary and education

sectors. In: 2009 IEEE Toronto International Conference Science

and Technology for Humanity (TIC-STH), pp. 789–794. IEEE

(2009)

3. Borg, J., Östergren, P.O.: Users’ perspectives on the provision of

assistive technologies in bangladesh: awareness, providers, costs

and barriers. Disabil. Rehabilit. Assist. Technol. 10(4), 301–308

(2015)

4. de Witte, L., Steel, E., Gupta, S., Ramos, V.D., Roentgen, U.:

Assistive technology provision: towards an international frame-

work for assuring availability and accessibility of affordable high-

quality assistive technology. Disabil. Rehabil. Assist. Technol.

13(5), 467–472 (2018)

5. International labour organization. https ://web.archi ve.org/

web/20180 42501 4312/http://www.ilo.org/wcmsp 5/group s/publi

c/---dgrep or ts/---dcomm /docum ents/publi catio n/wcms_08770

7.pdf. Accessed 24 Nov 2019

6. Esiyok, C., Askin, A., Tosun, A., Albayrak, S.: Software switches:

novel hands-free interaction techniques for quadriplegics based on

respiration-machine interaction. Univ. Access Inf. Soc. (2019).

https ://doi.org/10.1007/s1020 9-019-00645 -2

7. Esiyok, C., Albayrak, S.: Sitbench 1.0: a novel switch-based inter-

action technique benchmark. J. Healthcare Eng. (2019). https ://

doi.org/10.1155/2019/50751 63

8. Brooke, J., etal.: Sus-a quick and dirty usability scale. Usabil.

Eval. Ind. 189(194), 4–7 (1996)

9. WookKim, Y., HyunCho, J.: A novel development of head-set

type computer mouse using gyro sensors for the handicapped. In:

2nd Annual International IEEE-EMBS Special Topic Conference

on Microtechnologies in Medicine and Biology. Proceedings (Cat.

No. 02EX578), pp. 356–359. IEEE (2002)

10. Antunes, R.A., Palma, L.B., Duarte-Ramos, H., Gil, P.: Intelli-

gent hci device for assistive technology. In: Doctoral Conference

on Computing, Electrical and Industrial Systems, pp. 157–168.

Springer (2019)

11. Sancha-Ros, S., García-Garaluz, E.: Computer access and alter-

native and augmentative communication (aac) for people with

disabilities: a multi-modal hardware and software solution. In:

International Work-Conference on Artificial Neural Networks, pp.

605–610. Springer (2015)

12. Quha zono. https ://web.archi ve.org/web/20200 70715 4711/https

://www.quha.com/produ cts-2/zono/. Accessed 7 July 2020

13. Rosas-Cholula, G., Ramirez-Cortes, J., Alarcon-Aquino, V.,

Gomez-Gil, P., Rangel-Magdaleno, J., Reyes-Garcia, C.: Gyro-

scope-driven mouse pointer with an emotiv® eeg headset and data

analysis based on empirical mode decomposition. Sensors 13(8),

10561–10583 (2013)

14. Gerdtman, C., Bäcklund, Y., Lindén, M.: A gyro sensor based

computer mouse with a usb interface: a technical aid for motor-

disabled people. Technol. Disabil. 24(2), 117–127 (2012)

15. Zhang, T., Li, L., Yan, H.: The hci method for upper limb disabili-

ties based on emg and gyros. In: 2014 IEEE 13th International

Workshop on Advanced Motion Control (AMC), pp. 434–439.

IEEE (2014)

16. Castillo, A., Cortez, G., Diaz, D., Espíritu, R., Ilisastigui, K.,

O’Bard, B., George, K.: Hands free mouse. In: 2016 IEEE 13th

International Conference on Wearable and Implantable Body Sen-

sor Networks (BSN), pp. 109–114. IEEE (2016)

17. Sim, N., Gavriel, C., Abbott, W.W., Faisal, A.A.: The head

mouse—head gaze estimation in-the-wild with low-cost inertial

sensors for bmi use. In: 2013 6th International IEEE/EMBS Con-

ference on Neural Engineering (NER), pp. 735–738. IEEE (2013)

18. Raya, R., Roa, J., Rocon, E., Ceres, R., Pons, J.L.: Wearable iner-

tial mouse for children with physical and cognitive impairments.

Sens. Actuators A Phys. 162(2), 248–259 (2010)

19. Blackmon, F.R., Weeks, M.: Target acquisition by a hands-free

wireless tilt mouse. In: 2009 IEEE International Conference on

Systems, Man and Cybernetics, pp. 33–38. IEEE (2009)

20. Headmaster plus. http://web.archi ve.org/web/20191 12500 1741/

https ://abled ata.acl.gov/produ ct/headm aster -plus-model -hm-2p/.

Accessed 11 Dec 2019

21. LoPresti, E.F., Brienza, D.M., Angelo, J.: Head-operated com-

puter controls: effect of control method on performance for sub-

jects with and without disability. Interact. Comput. 14(4), 359–

377 (2002)

22. Andres, R.O., Hartung, K.J.: Prediction of head movement time

using fitts’ law. Hum. Factors 31(6), 703–713 (1989)

23. Jouse2. https ://web.ar c hi ve.or g/ w eb /20191 11214 4721 /http://tekny

ka.com/Jouse 2/. Accessed 24 Nov 2019

24. Kjeldsen, R.: Improvements in vision-based pointer control. In:

Proceedings of the 8th International ACM SIGACCESS Confer-

ence on Computers and Accessibility, pp. 189–196. ACM (2006)

25. Tu, J., Huang, T., Tao, H.: Face as mouse through visual face

tracking. In: The 2nd Canadian Conference on Computer and

Robot Vision (CRV’05), pp. 339–346. IEEE (2005)

26. Javanovic, R., MacKenzie, I.S.: Markermouse: mouse cursor con-

trol using a head-mounted marker. In: International Conference on

Computers for Handicapped Persons, pp. 49–56. Springer (2010)

27. Toyama, K.: Look, ma-no hands! hands-free cursor control with

real-time 3d face tracking. PUI98 (1998)

28. Gorodnichy, D.O., Malik, S., Roth, G.: Nouse -use your nose as

a mouse- a new technology for hands-free games and interfaces.

In: Proceedings of Internationl Conference on Vision Interface

(VI’2002), pp. 354–361 (2002)

29. Morris, T., Chauhan, V.: Facial feature tracking for cursor control.

J. Netw. Comput. Appl. 29(1), 62–80 (2006)

30. Kumar, R., Kumar, A.: Black pearl: an alternative for mouse and

keyboard. J. Graph. Vis. Image Process. 8, 1–6 (2008)

31. El-Afifi, L., Karaki, M., Korban, J., alAlaoui, M.A.: Hands-free

interface-a fast and accurate tracking procedure for real time

human computer interaction. In: Proceedings of the Fourth IEEE

International Symposium on Signal Processing and Information

Technology, pp. 517–520. IEEE (2004)

Universal Access in the Information Society

1 3

32. Pallejà, T., Soler, E.R., Teixidó, M., Tresanchez, M., Del Viso,

A.F., Sánchez, C.R., Palacin, J.: Using the optical flow to imple-

ment a relative virtual mouse controlled by head movements. J.

UCS 14(19), 3127–3141 (2008)

33. Su, M.C., Su, S.Y., Chen, G.D.: A low-cost vision-based human-

computer interface for people with severe disabilities. Biomed.

Eng. Appl. Basis Commun. 17(06), 284–292 (2005)

34. Lin, Y.P., Chao, Y.P., Lin, C.C., Chen, J.H.: Webcam mouse using

face and eye tracking in various illumination environments. In:

2005 IEEE Engineering in Medicine and Biology 27th Annual

Conference, pp. 3738–3741. IEEE (2006)

35. Varona, J., Manresa-Yee, C., Perales, F.J.: Hands-free vision-based

interface for computer accessibility. J. Netw. Comput. Appl. 31(4),

357–374 (2008)

36. Betke, M., Gips, J., Fleming, P.: The camera mouse: visual track-

ing of body features to provide computer access for people with

severe disabilities. IEEE Trans. Neural Syst. Rehabilit. Eng. 10(1),

1–10 (2002)

37. Song, Y., Luo, Y., Lin, J.: Detection of movements of head and

mouth to provide computer access for disabled. In: 2011 Interna-

tional Conference on Technologies and Applications of Artificial

Intelligence, pp. 223–226. IEEE (2011)

38. Perini, E., Soria, S., Prati, A., Cucchiara, R.: Facemouse: A

human-computer interface for tetraplegic people. In: European

Conference on Computer Vision, pp. 99–108. Springer (2006)

39. Alhamzawi, H.A.: Control mouse cursor by head movement:

development and implementation. Appl. Med. Inform. 40 (2018)

40. Kim, H., Ryu, D.: Computer control by tracking head movements

for the disabled. In: International Conference on Computers for

Handicapped Persons, pp. 709–715. Springer (2006)

41. Strumiłło, P., Pajor, T.: A vision-based head movement tracking

system for human-computer interfacing. In: 2012 Joint Confer-

ence New Trends In Audio & Video and Signal Processing: Algo-

rithms, Architectures, Arrangements and Applications (NTAV/

SPA), pp. 143–147. IEEE (2012)

42. Kim, S., Park, M., Anumas, S., Yoo, J.: Head mouse system based

on gyro-and opto-sensors. In: 2010 3rd International Conference

on Biomedical Engineering and Informatics, vol.4, pp. 1503–

1506. IEEE (2010)

43. Manresa-Yee, C., Varona, J., Perales, F.J.: Towards hands-free

interfaces based on real-time robust facial gesture recognition. In:

International Conference on Articulated Motion and Deformable

Objects, pp. 504–513. Springer (2006)

44. Loewenich, F., Maire, F.: Hands-free mouse-pointer manipulation

using motion-tracking and speech recognition. In: Proceedings of

the 19th Australasian Conference on Computer-Human Interac-

tion: Entertaining User Interfaces, pp. 295–302. ACM (2007)

45. Headmouse nano. https ://web.archi ve.org/w eb/20191 00610 5058/

https ://www.or in.com/acces s/headm ouse/. Accessed on 24 Nov

2019

46. Walsh, E., Daems, W., Steckel, J.: An optical head-pose tracking

sensor for pointing devices using ir-led based markers and a low-

cost camera. In: 2015 IEEE Sensors, pp. 1–4. IEEE (2015)

47. Trackerpro. https ://web.archi ve.org/web/20160 73102 3751/https

://www.ablen etinc .com/track erpro . Accessed 24 Nov 2019

48. Smartnav. https ://web.archi ve.org/web/20190 60710 0209/http://

www.natur alpoi nt.com/smart nav/. Accessed 24 Nov 2019

49. Li, S., Ngan, K.N., Sheng, L.: A head pose tracking system using

rgb-d camera. In: International Conference on Computer Vision

Systems, pp. 153–162. Springer (2013)

50. Head switch. http://web.archi ve.org/web/20191 12500 2431/https

://enabl ingde vices .com/produ ct/head-switc h/. Accessed 11 Dec

2019

51. Easy flex dual ultimate switch. http://web.archi ve.org/web/20191

11404 0944/https ://enabl ingde vices .com/produ ct/easy-flex-dual-

ultim ate-switc h/. Accessed 24 Nov 2019

52. Gorodnichy, D., Dubrofsky, E., Mohammad, A.: Working with a

computer hands-free using the nouse perceptual vision interface.

In: International Workshop on Video Processing and Recognition

(VideoRec’07) (2007)

53. Fu, Y., Huang, T.S.: Hmouse: head tracking driven virtual com-

puter mouse. In: 2007 IEEE Workshop on Applications of Com-

puter Vision (WACV’07), Austin, TX, pp. 30. IEEE (2007). https

://doi.org/10.1109/WACV.2007.29

54. Statista: Smartphone users worldwide 2016-2021. https ://web.

archi ve.org/web/20191 12002 1523/https ://www.stati sta.com/stati

stics /33069 5/numbe r-of-smart phone -users -world wide/. Accessed

24 Nov 2019

55. Bröhl, C., Rasche, P., Jablonski, J., Theis, S., Wille, M., Mertens,

A.: Desktop pc, tablet pc, or smartphone? an analysis of use pref-

erences in daily activities for different technology generations

of a worldwide sample. In: International Conference on Human

Aspects of IT for the Aged Population, pp. 3–20. Springer (2018)

56. Cao, H., Lv, H., Sun, Q.: Model design based on mems gyroscope

random error. In: 2015 IEEE International Conference on Infor-

mation and Automation, pp. 2176–2181. IEEE (2015)

57. Ruan, X.g., Yu, M.m.: Modeling research of mems gyro drift

based on kalman filter. In: The 26th Chinese Control and Deci-

sion Conference (2014 CCDC), pp. 2949–2952. IEEE (2014)

58. Kownacki, C.: Optimization approach to adapt kalman filters for

the real-time application of accelerometer and gyroscope signals’

filtering. Digital Signal Process. 21(1), 131–140 (2011)

59. Xue, L., Jiang, C.Y., Chang, H.L., Yang, Y., Qin, W., Yuan, W.Z.:

A novel kalman filter for combining outputs of mems gyroscope

array. Measurement 45(4), 745–754 (2012)

60. Bangor, A., Kortum, P., Miller, J.: Determining what individual

sus scores mean: adding an adjective rating scale. J. Usabil. Stud.

4(3), 114–123 (2009)

Publisher’s Note Springer Nature remains neutral with regard to

jurisdictional claims in published maps and institutional affiliations.