The Role of Haptics in User Input for Simple 3D Interaction Tasks

An Analysis of Interaction Performance and User Experience

Mirjam Augstein

, Thomas Neumayr

, Stephan Vrecer

, Werner Kurschl

and Josef Altmann

Communication and Knowledge Media, University of Applied Sciences Upper Austria, Hagenberg, Austria

Research and Development, University of Applied Sciences Upper Austria, Hagenberg, Austria

Human-Centered Computing, University of Applied Sciences Upper Austria, Hagenberg, Austria

Keywords:

Haptic Interaction, Touchless Interaction, Comparative Study.

Abstract:

Traditionally, input devices allowed for at least a certain degree of haptic experience by involving direct phy-

sical contact between user and device. Recently, touchless interaction gained popularity through readily avai-

lable, cheap devices like the Leap motion controller or Microsoft Kinect. Usually, these devices support more

than two degrees of freedom and are thus especially suitable for interaction tasks in a three-dimensional space.

However, besides the high potential that lies within touchless input techniques, they also involve new challen-

ges (e.g., lack of borders and natural haptic guidance). In this paper, we aim at the identiﬁcation of potentials

and limitations inherent to three different input techniques that involve a varying amount of haptics (i.e., tou-

chful, touchless and semi-touchless input). We present a study conducted with 25 users that focuses on simple

input tasks in a 3D interaction space and analyzes objective interaction performance metrics (e.g., regularity or

time) and subjective User Experience aspects (e.g., dependability or efﬁciency). It reveals parallels as well as

contrasts between the users’ actual interaction performance and perceived UX (e.g., several metrics suggested

haptic input to outperform touchless input while differences regarding UX were not signiﬁcant). The results

are intended to inform other researchers when designing interactive environments.

1 INTRODUCTION

In the past decades, user input seemed to be cou-

pled with at least a certain haptic experience. The

most widely used input devices like mice, keybo-

ards, touch screens or game controllers have in com-

mon that they involve direct physical contact between

the user and the device. Touch-based or haptic in-

put techniques however often suffer from limited in-

teraction options through physical/technological con-

straints. For instance, most touchscreens only allow

for two-dimensional input because the physical de-

vice is a ﬂat surface which makes movement along

the third axis mostly irrelevant (although this might

be up to changes since the advent of 3D touch capa-

bilities with the latest generations of smart phones).

Yet, these physical restrictions also provide a certain

amount of guidance and haptic input devices mostly

offer inherent haptic feedback mechanisms.

During the past few years, touchless input met-

hods and devices, e.g., Microsoft Kinect (further re-

ferred to as “Kinect”) or the Leap motion controller

(further referred to as “Leap”) that theoretically allow

for nearly unlimited input options (e.g., the human

hand theoretically has 27 degrees of freedom (Rehg

and Kanade, 1994)) became attractive, e.g., in game-

based scenarios that involve physical engagement like

sports applications or in therapeutic settings. Tou-

chless input surely offers high potential for a num-

ber of selected application ﬁelds where other input

approaches are difﬁcult to apply (e.g., sports appli-

cations that involve whole body interaction). Howe-

ver, touchless input also bears potential shortcomings

that should be considered, e.g., connected to haptic

perception mentioned before or missing physical con-

straints that could provide orientation for the user du-

ring input. Although there are approaches to over-

come these risks (through haptic feedback for tou-

chless input scenarios, e.g., via focused ultrasound

(Carter et al., 2013)), most readily-available touchless

input devices do not yet support such mechanisms.

In this paper, we aim at investigating the actual

relevance of a haptic experience during user input.

This relevance might differ with varying complexity

of the input tasks. Here, we focus on simple input

tasks in a 3D interaction space. It presents a study

Augstein, M., Neumayr, T., Vrecer, S., Kurschl, W. and Altmann, J.

The Role of Haptics in User Input for Simple 3D Interaction Tasks - An Analysis of Interaction Performance and User Experience.

DOI: 10.5220/0006580500260037

In Proceedings of the 13th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2018) - Volume 2: HUCAPP, pages

26-37

ISBN: 978-989-758-288-2

conducted with 25 participants to compare three input

techniques: i) touchless input, ii) “touchful” input (re-

lying on application of physical pressure to a surface),

and iii) semi-touchless input (combining characteris-

tics of the previous two). We use Leap as touchless in-

put device and two device prototypes (Augstein et al.,

2017a) that have been developed speciﬁcally to ﬁt the

requirements of this and similar studies. The touchful

input device can be considered isometric, i.e., con-

necting the human limb and the device through force

(Zhai, 2008) while the other two can be considered

isotonic, i.e., connecting the interacting hand and the

device through movement (Zhai, 2008).

The study is targeted to analyze i) users’ inte-

raction performance and ii) User Experience (UX)

with each of the three input techniques. The main

aim behind is to study the actual relevance of haptics

during user input (and, secondarily, also the poten-

tials of isometric and isotonic input devices) for the

selected category of tasks. We believe that the com-

bined ﬁndings comprising objective criteria and sub-

jective ones are more conclusive than any of the two

in isolation. A good interaction performance does

not necessarily have to imply a good UX and vice

versa (which is conﬁrmed by the results of our study

where we identiﬁed contradictions between perfor-

mance and UX). In our study we introduce so-called

interaction tests the users have to take. Based on the

raw input data we compute metrics indicative of in-

teraction performance (e.g., time or interaction regu-

larity). To measure UX, the standardized User Expe-

rience Questionnaire is used. We expect the results to

be interesting for researchers and practitioners con-

ceptualizing and designing interactive environments

or input devices.

2 RELATED WORK

This section presents grounding related work on 3D

input as well as studies comparing different input

techniques for 3D input tasks similar to ours by ana-

lyzing interaction performance and UX.

According to (Fr

ohlich et al., 2006), 3D inte-

raction can be classiﬁed into i) navigation and travel,

ii) selection, iii) manipulation (e.g., of an object’s po-

sition) and iv) system control. We categorize the in-

teraction tasks described in this paper (see Section 4)

as manipulation tasks (speciﬁcally, positioning) and

selection tasks. Fr

ohlich et al. divide devices and sen-

sors into isotonic (measuring movement), isometric

(measuring force) and elastic (allowing for movement

and providing a counterforce) sensors. (Zhai, 2008)

describe elastic input sensors as devices with varying

resistance “between the isometric (inﬁnite resistance)

and the isotonic (zero or constant resistance)”. The

devices we use for our study can be considered iso-

metric (see “touchful input” in Section 3.2) and isoto-

nic (see touchless and semi-touchless input in Secti-

ons 3.1 and 3.3). (Bowman et al., 2008) argue that

many input devices are actually a combination of dif-

ferent types of input sensors which they refer to as

“multisensory input”. While we agree on the value of

multisensory input for practical use we explicitly tried

to avoid combinations in our study to be better able to

compare.

(Zhai, 2008) presents an exhaustive compilation

of studies comparing isometric and isotonic input de-

vices. However, he concludes that the literature on

the relative advantages and disadvantages of isome-

tric vs. isotonic devices has not been very conclusive

and argues that the deﬁnite answer may depend on

(among others) the concrete interaction tasks. Thus,

we carefully selected tasks for our study and assured

that they are identical for all three input devices.

(Bowman et al., 2008) also argue that haptics is

“one of the most important types of sensory feed-

back” and distinguish between “active haptic feed-

back” and “passive” or “pseudo-haptics”. Our tou-

chful input technique can be considered passive hap-

tic input, our touchless input technique intentionally

does not involve any haptic experience and our semi-

touchless input technique can be classed somewhere

in between (see Section 3). The comparison of in-

put techniques involving a differing amount of haptics

was our main focus. (Zhai, 2008) further distinguis-

hes between position control, i.e., “control mecha-

nisms by which the human operator controls object

positions directly” and rate control, i.e., mapping “hu-

man input into the velocity of the object movement”.

In our study, all three input techniques and interaction

tasks are based on position control.

Recently, (Tscharn et al., 2016) have evalua-

ted two isotonic input devices (Leap and the 3D

mouse SpaceNavigator) for real world navigation

tasks (using Google Earth). Their study partici-

pants were asked to solve four different interaction

tasks (general movement, targeted movement, spe-

ciﬁed coordinate movement and speciﬁed trajectory

movement). They evaluated navigation efﬁciency

(measured based on the time needed for a task) and

UX (based on the analysis of facial expressions and

the AttrakDiff questionnaire). Their approach is simi-

lar to ours regarding the idea of comparing touchless

and touch-able input devices along interaction perfor-

mance metrics and UX. Although it differs regarding

domain and UX indicators, some of the interaction

tasks and performance metrics are similar. Tscharn et

The Role of Haptics in User Input for Simple 3D Interaction Tasks - An Analysis of Interaction Performance and User Experience

al. found interaction with Leap to be less accurate for

complex tasks, compared to SpaceNavigator while it

had a good UX for simple tasks (which is in line with

our ﬁndings).

Another study on interaction performance and UX

related to touchless (using a ceiling-mounted gesture

recognition device prototype) and touch-able (using

SpaceNavigator) input has been conducted earlier by

(Stannus et al., 2011). They found that regarding in-

teraction performance and UX indicators as scored

by the participants (e.g., naturalness, strain, speed,

and accuracy), the touchless device could not keep

up with touch-able devices. Interestingly, similar to

our ﬁndings, for touchless input, UX-related metrics

(e.g., naturalness) were scored relatively better than

the pure performance indicators. Although study aims

were similar, the study design described by Stannus

et al. differs from ours drastically as we analyze ob-

jective interaction performance metrics automatically

while Stannus et al. asked the participants to rate these

metrics. We believe these “objective” results reported

by users subjectively might be easily biased by users’

UX-related impressions and are thus less reliable.

(Dangeti et al., 2016) discuss bare-hand in-air

gesture-based interaction vs. object-in-hand tangible

interaction for navigation of 3D objects. They pre-

sent the technological differences between the appro-

aches and announce a user study in which they plan

to compare three different interaction methods (tra-

ditional mouse/keyboard interaction, bare hand in-air

and object-in-hand interaction) along objective cri-

teria like interaction speed and accuracy and UX-

related aspects based on interviews. Thus, the plan-

ned methodology seems to be similar to ours, unfor-

tunately, they did not publish any results yet.

(Coelho and Verbeek, 2014) conducted a study

with a traditional mouse and Leap for pointing tasks

similar to ours in a 3D virtual environment. They des-

cribe two different tasks their participants had to per-

form. The ﬁrst task consisted of simple pointing (start

point to target point) while the second task was more

complex (start point to ﬁrst target point to second tar-

get point). For the simpler task, Leap outperformed

the mouse while for the more complex task, the out-

come was contrary. The authors state that the z-axis

can be controlled by the “roll of the mouse”, which

we assume to be the mouse wheel and can therefore

be manipulated isolated from the x- and y-axes. Our

assumption is that there is a major difference for 3D

pointing tasks when using Leap, where all axes are

controlled through the same mechanism (i.e., ﬁnger

or hand tracking). In our study, we tried to avoid such

general differences (not related to haptics) among the

input techniques. To measure usability (and UX),

Coelho and Verbeek used the System Usability Scale

(SUS) questionnaire where Leap scored better than

the mouse which is partly in line with our results.

urst and Helder, 2011) investigated navigation

and selection tasks in a 3D space using mobile de-

vices for moving around virtual objects. However,

while we varied the extent of haptics, H

urst and Hel-

der varied the type of visualization. Thus, the results

are not directly related to our work but the study setup

is similar. They measured objective values from log

data and obtained information about UX.

Another similar approach was applied by (Atkins

et al., 2009) who compared three different interaction

techniques (i.e., mouse wheel, mouse dragging, and

a jog wheel) to navigate through a stack of medical

3D images, however the navigation itself was two-

dimensional. They also used interaction accuracy,

time, and navigation paths as objective measures and

qualitatively evaluated participants’ preferences.

Although we found several studies with aims si-

milar to ours, we did not ﬁnd a comparison where

input activities with different devices have been alig-

ned in a way that the actual difference between set-

tings is the amount of haptics involved. Most stu-

dies compared Leap to a (3D) mouse, however, alt-

hough both isotonic, interaction with Leap inherently

differs from interaction with a mouse e.g., regarding

hand movement, interaction space and DoF. We ai-

med at comparing the input techniques using devices

that allow for almost identical input activities (with

the amount of haptics involved being the actual diffe-

rence). We expect the results to be better generaliza-

ble and more interesting for designers of interactive

environments.

3 INPUT TECHNIQUES

The three input techniques we used for our study have

several aspects in common. They all involve small

movements of the dominant hand in a 3D space and

are based on position control. The 3D manipulation

tasks used in the study (see Section 4) only require the

input devices to analyze the user’s hand position in a

predeﬁned 3D space but allow for users to choose the

most comfortable hand orientation and posture. The

manipulation tasks require movement in the directi-

ons left/right, forward/backward and down/up, see Fi-

gure 1. The interaction space is identical for all set-

tings and the devices have been calibrated within pre-

tests (see Section 5). Differences (that are subject to

our study) can be seen in the amount of haptics invol-

ved. Further, our touchless and semi-touchless input

techniques and devices can be described as isotonic

HUCAPP 2018 - International Conference on Human Computer Interaction Theory and Applications

Figure 1: Input devices and techniques and the three DoF

as required by the manipulation tasks during the user study.

while touchful input is isometric.

3.1 Touchless Interaction

Touchless interaction is described as interaction that

“can take place without mechanical contact between

the human and any part of the artiﬁcial system” (de la

Barr

e et al., 2009). In many cases, a challenge rela-

ted to touchless input is the absence of direct haptic

guidance during the interaction which can also be the

cause of confusion or uncertainty regarding the space

sensitive to the user’s actions. Further, users are in-

formed about the effects of their actions only via the

system’s output (e.g., visually). This visual informa-

tion is present for the other two techniques as well,

for touchless input it is, however, the only source of

feedback. We implemented our touchless setting with

Leap which uses infrared sensors for measurement. It

has a size of 8x3x1cm and is usually placed on the

table in front of the user. In our study, it was used to

track the position of the wrist. Here, the user holds

his/her hand above the device (see Figure 1 (c)) and

moves it according to the tasks. The provision of a

hand rest was not possible as this would have i) inter-

fered with accurate tracking of the user’s hand and ii)

reduced the “touchless” impression for the users.

3.2 Touchful Interaction

We deﬁne touchful interaction as the opposite of tou-

chless interaction, i.e., interaction which involves tho-

rough mechanical contact between human and de-

vice. Thus, it generally includes the majority of the

most widely used input methods and devices such as

mice. Touchful interaction allows for a direct hap-

tic experience and can help the user to orientate and

correctly assess the interaction options and limitati-

ons (e.g., the physical resistance of a button indica-

tes that applying stronger pressure will make no dif-

ference). Yet, these generally helpful boundaries also

limit the interaction range. The concrete manifesta-

tion of touchful input used in our work can be descri-

bed as three-dimensional pressure-based interaction.

The technique aims to combine the advantages of 3D

interaction (as also enabled by touchless techniques)

and pressure-based interaction while reducing the re-

spective shortcomings. It ties in with previous appro-

aches to one-dimensional pressure-based interaction,

see (Augstein et al., 2015) and (Hwang et al., 2013).

This technique allows application of different pres-

sure intensities to a speciﬁed target area.

The device used for our study, “SpongeBox”

(Augstein et al., 2017a), extends this interaction

concept by adding several dimensions/directions:

left/right, forward/backward and down/up. Sponge-

Box has been developed speciﬁcally for purposes of

comparing touchful to touchless input techniques for

simple 3D manipulation tasks. It is based on an Ar-

duino microcontroller and several pressure sensors

and has the physical appearance of a box with open

upper and back walls. The inner walls are covered

with sponges. The user puts a hand in the middle

(see Figure 1 (a)) and can then press against four si-

des of the box (bottom, left, right, forward). The

material used for the walls provides a haptic expe-

rience and physical restriction. Placing the hand in-

side the box without actively applying pressure does

not lead to unintended input, thus a user’s arm can

rest on the sponge during and between interaction

activities. SpongeBox can theoretically distinguish

between 1024 pressure intensities in each direction

which were however reduced signiﬁcantly during the

pre-tests (see Section 5). The conﬁguration used for

the study measured and collected pressure intensities

on the percentage level (0-100%) but used only 13

levels to trigger activities (this was the conﬁguration

perceived as most predictable and comfortable by the

pre-testers). As the interaction space was relatively

small, the movement of the digital object remained

smooth with this conﬁguration.

3.3 Semi-Touchless Interaction

Semi-touchless interaction can be deﬁned as a combi-

nation of the touchless and touchful interaction con-

cepts. It joins the following characteristics: i) phy-

sical constraints and ii) touchless position/movement

analysis. Again, input is done via small movements

of the interacting hand. The directions are equal to

those used for the other two techniques and physical

borders around the area sensitive to input are provi-

The Role of Haptics in User Input for Simple 3D Interaction Tasks - An Analysis of Interaction Performance and User Experience

ded. The input device used in our study is “Squa-

reSense” (Augstein et al., 2017a). Similar to Spon-

geBox, it has been developed speciﬁcally for purpo-

ses of comparing semi-touchless to touchless input

techniques for simple 3D manipulation tasks. It con-

sists of a box with highly sensitive capacitive side,

bottom and front walls the user can freely move the

hand within. The device can recognize touchless mo-

vements but also pressure intensities if the walls are

touched. Thus the device is isometric-isotonic gene-

rally, however, we did not use application of pressure

for the study described here as the devices were con-

ﬁgured in a way that the physical borders marked the

interactive area (which was also aligned to the digital

interaction space, see Section 5).

As with the touchful setting, the user’s arm can

rest on a sponge placed in the rear part of the box

which reduces the physical strain often involved with

touchless input. SquareSense is based on an Arduino

microcontroller and capacitive copper plates ﬁxed to

the walls of the box. Similarly to SpongeBox, Squa-

reSense was conﬁgured in an iterative process during

the pre-tests and used an equal number of 13 distinct

proximity levels for each direction.

4 INTERACTION TESTS AND

PERFORMANCE METRICS

Our user study considers several objective metrics in-

dicative of interaction performance. These metrics

have been selected speciﬁcally to ﬁt simple 3D mani-

pulation tasks. To be able to compute values for these

metrics, our users perform identical so-called “inte-

raction tests” with each input technique. We consider

the metrics to be interesting individually and do not

aim at computing an overall index of performance.

We thus utilize a user modeling framework (Augstein

et al., 2017b) which offers an infrastructure for de-

ﬁning, maintaining and analyzing an arbitrary num-

ber of individual metrics. Each test (see the following

sections) requires the user to move a red cube (see

Figure 2) which acts as a cursor in a 3D space accor-

ding to a particular task. The devices were used for

absolute positioning of the cube in the digital space.

4.1 Reach and Regularity

The ﬁrst test (and corresponding metric) is called Re-

ach. Here, the user has to move the red interactive

cube to the personal maximum comfortable position

in the directions left, right, forward and down (in

all cases starting from the same position in the cen-

ter of the interaction space). The metric Reach has

been selected because the mobility and strength of

a user’s dominant hand (which are decisive for the

values achieved for Reach) are highly individual and

could be signiﬁcantly reduced for users with (tempo-

rary) motor impairments which again can strongly in-

ﬂuence interaction performance. The test is similar to

the “straight navigation task” used by (Tscharn et al.,

2016) in their study on 3D map navigation.

At the touchful (isometric) setting, a user has

to apply the highest amount of pressure he/she can

(comfortably) achieve in all four directions to move

the cube in the respective direction. The maximum

values achieved for the four directions are stored as

ReachLeft, ReachRight, ReachDown, and ReachFor-

ward. At the (semi-)touchless (isotonic) settings, the

test requires a user to move the cube by moving the

hand. The values are stored in percent of the sy-

stem’s global maximum (the red cube cannot be mo-

ved out of the interaction space, if it has reached the

maximum position, the resulting Reach value is 100%

and the cube will stop there). The results could also

be used to individually adapt the ContinuousRegula-

rity test which was done within a similar study with

people with impairments where Reach was quite di-

verse.

Additionally, within the same test, the system ana-

lyzes deviations from the straight-most path between

initial and maximum position and uses it to compute

a Regularity metric (again, for all interaction directi-

ons and measured in percent). A straight path would

result in a Regularity of 100%. The path (including

the respective target and actual positions) is analy-

zed at every time stamp between initial and end po-

sition. The deviation from the straight path is avera-

ged over all time stamps and subtracted from an ini-

tial value of 100%. Regularity is computed for each

direction (RegularityLeft, RegularityRight, Regulari-

tyDown, RegularityForward).

4.2 Continuous Regularity

The second test ContinuousRegularity shares some

characteristics with the “rotation navigation task”

used by (Tscharn et al., 2016) and requires users to

follow a green target cube (see Figure 2) over a path

that reaches all relevant areas of the 3D interaction

space. The target path starts at the initial position in

the center of the interaction space, then moves around.

The computation algorithmically matches the one

of Regularity, which is, however, tested for each of

the four directions individually only while Continu-

ousRegularity requires a user to perform a coordina-

ted, continuous and interruption-free movement that

covers all directions. To be able to identify pre-

HUCAPP 2018 - International Conference on Human Computer Interaction Theory and Applications

Figure 2: Visualization users see during interaction.

ferred directions (e.g., left or right) or dimensions

(e.g., left/right or down/up), we compute individual

metrics (ContinuousRegularityLeftForward, Continu-

ousRegularityLeftDown, ContinuousRegularityRight-

Forward and ContinuousRegularityRightDown) in

addition to the general ContinuousRegularity metric

which considers the whole path.

4.3 Time

Finally, the third test called Time requires users to re-

ach target cubes that consecutively appear in each of

the four directions after a certain delay as quickly as

possible. Again, the red interactive cube is positio-

ned in the center of the interactive space initially and

is moved back there after a target cube has been re-

ached. The target cubes are big enough to be easily

reached as it was not the aim to ﬁnd out the optimum

size of the target area but the interaction speed in case

the target is of sufﬁcient size. The metrics are called

TimeLeft, TimeRight, TimeForward, and TimeDown.

5 USER STUDY

A two-fold user study was conducted in summer

of 2016 to compare touchful, touchless and semi-

touchless input for pre-speciﬁed manipulation tasks

in a 3D interaction space. As mentioned earlier, the

three devices used have been thoroughly tested and

adjusted prior to the actual study. The pre-tests took

place with three test users that did not participate in

the actual study and aimed at i) ensuring compara-

bility among the three devices related to spatial re-

solution, reactivity, interaction space and movement

range, and ii) identiﬁcation of device conﬁgurations

convenient for the user. Thus, we e.g., tested for all

devices whether they reacted to user input in the ex-

pected way (e.g., without delays and with the physical

effort involved being within a comfortable range). As

the sensation of physical demand or fatigue is indivi-

dual it was important to not only test with one pre-test

user. Thus we tested with three users (including also a

user without previous experience with alternative in-

put methods and devices) and chose a setting that was

described as comfortable by all of them. Further, the

pre-tests involved an iterative adjustment of the devi-

ces’ conﬁguration (e.g., regarding interaction space,

reactivity and sensitivity to user input) in order to en-

sure comparability. For instance, we adjusted the mi-

nimum pressure intensity required to trigger an input

activity for the touchful setting and smoothed the va-

lues delivered by the semi-touchless device. Also, we

aligned the physical with the digital interaction space

and made sure that the physical space was equal also

for the touchless setting (which inherently did not in-

volve physical constraints).

5.1 Research Questions

The user study was tailored to evaluate two aspects re-

lated to the three input techniques and devices. First,

it should identify strengths and weaknesses of the in-

dividual devices/techniques, based on the interaction

performance indicators Reach, Regularity, Continu-

ousRegularity and Time. In the second part, UX re-

lated to each of the input devices/techniques should

be evaluated, using the categories Efﬁciency, Perspi-

cuity, Dependability, Stimulation and Novelty deﬁned

by the standardized User Experience Questionnaire

(UEQ)

and described later. As i) the interaction tests

were identical for all devices, ii) the users’ required

interaction activities were comparable for all techni-

ques and iii) the users had enough time to get familiar

with the input techniques prior to the actual tests, the

main difference between the three input techniques is

the degree of haptics involved. Additionally, a diffe-

rence can be seen between touchful and the other two

techniques as it relies on force application (isometric)

whereas the other two rely on movements (isotonic).

Based on our expectations and ﬁndings of related stu-

dies, we formulated seven hypotheses:

• H1: We expect users to be able to interact fastest

with touchless input.

• H2: We expect Reach to be about equally high

with all three input techniques.

• H3: We expect interaction Regularity to be better

with input techniques that involve haptics.

• H4: We expect ContinuousRegularity to be better

with input techniques that involve haptics.

• H5: We expect UX to be generally better with in-

put techniques that involve haptics.

http://www.ueq-online.org

The Role of Haptics in User Input for Simple 3D Interaction Tasks - An Analysis of Interaction Performance and User Experience

• H6: We expect Stimulation to be better with input

techniques that involve haptics.

• H7: We expect Dependability to be better with

input techniques that involve haptics.

Regarding H1, we became aware that related studies

report divergent results, e.g., (Tscharn et al., 2016)

found users to be slower with Leap for navigation

tasks, compared to a 3D mouse whereas in the study

of (Coelho and Verbeek, 2014) who compared Leap to

a common mouse, users were faster with Leap at tasks

comparable to ours. Although (Zhai, 2008) concludes

that “human response with an isometric device is fas-

ter than with a comparable isotonic one since no trans-

port of limb or device is needed” we expected users to

be faster with touchless input in our study. First, they

did not have to overcome physical resistance to hit the

target and second, the distance users had to transport

limb (i.e., their hand) over was equal for all three in-

put techniques. Regarding H2, we expected Reach to

be about equally high with all input techniques as the

participants of the study did not have any known im-

pairments (reducing interaction range) and analyzed

it just to conﬁrm this assumption. Regarding H3 and

H4, we expected (Continuous)Regularity to be better

with input techniques that provide physical restricti-

ons which constitute some form of guidance. Regar-

ding H5, H6 and H7, we expected general UX, and

Dependability and Stimulation in particular to be bet-

ter with input techniques that involve a haptic expe-

rience because prior research suggests that even sim-

ple haptic stimulation can contribute to the communi-

cation of emotional information, see, e.g., (Salminen

et al., 2008).

5.2 Procedure and Methodology

The within-subjects study took place in a controlled

lab setting. Participants were asked to do the three in-

teraction tests with all devices. We used a counterba-

lanced (latin square) order in which the devices were

presented to the them to prevent a bias due to practi-

cing effects and control other position effects like fati-

gue. Before the tests, users got an introduction by the

test supervisor, explaining device and input techni-

que, and could become familiar with it. When they

were ready, the tests started. The results were auto-

matically recorded and analyzed. In total, 25 users

did 27 interaction tasks each (9 tasks with three de-

vices), resulting in a total number of 675 tasks that

were analyzed. The interaction during the phase of

getting familiar with a setting was not recorded. Af-

ter the tests with a device, users answered the stan-

dardized UEQ, see e.g., (Laugwitz et al., 2008). The

UEQ aims at i) enabling quick assessment, ii) cap-

turing comprehensive impression of UX, and iii) en-

couraging simple and immediate feedback (Laugwitz

et al., 2008). It is available in 15 languages and com-

prises the following UX aspects (“scales”): attracti-

veness, perspicuity, efﬁciency, dependability, stimu-

lation, and novelty. Each scale is represented by a

number of “items” that have the form of a semantic

differential, using a seven-point scale between -3 and

3. For our study we excluded the scale attractiveness

as we considered the related items less important for

our objectives (after making sure that the UEQ allows

for reduction of categories). Our users answered 20

items related to the remaining ﬁve scales:

• Perspicuity: Is it easy to get familiar with the pro-

duct? Is it easy to learn how to use it? Exam-

ple items are not understandable/understandable

or easy/difﬁcult to learn.

• Efﬁciency: Can users solve their tasks without un-

necessary effort? Example items for this category

are fast/slow or impractical/practical.

• Dependability: Does the user feel in control of

the interaction? Example items are unpredic-

table/predictable or meets expectations/does not

meet expectations.

• Stimulation: Is it motivating to use the product?

Example items for this category are not interes-

ting/interesting or motivating/demotivating.

• Novelty: Is the product innovative and creative?

Does the product catch users’ interests? Example

items are creative/dull or conservative/innovative.

The procedure was repeated for all input techniques.

Afterwards, we collected basic demographic informa-

tion. The tests took 15 to 30 minutes per person.

5.3 Participants

We could recruit 25 volunteers, aged between 19 and

49 (AV = 34.12, SD = 9.35), 8 male and 17 female.

All participants were right-handed and used the right

hand for interaction. Most participants were staff or

students of the university. The user group was excep-

tionally diverse regarding experiences with alternative

input techniques (reaching from a UX professor to an

ofﬁce assistant without any previous experiences with

alternative input devices). About 24% of the users had

previous experience with Leap (only 3 users had used

it more thoroughly). Also 24% had used SpongeBox

once before (due to an earlier user test). In order not to

cause a bias based on previous experiences, we allo-

wed all participants to get familiar with the respective

device as long as they needed before the test.

HUCAPP 2018 - International Conference on Human Computer Interaction Theory and Applications

6 RESULTS

This section summarizes the results of the user study.

6.1 Interaction Performance

We will discuss the results based on the metrics des-

cribed earlier, distinguishing between the categories

Reach, Regularity, ContinuousRegularity and Time

(see details in Table 1 and Table 2). To statisti-

cally analyze the differences between the three input

techniques, we ran Friedman’s tests for these crite-

ria. The test was chosen due to the characteristics

of the data which partly violate the normal distribu-

tion prerequisite of ANOVA. However, to examine

the results’ stability, we repeated the analysis with a

repeated-measures ANOVA which conﬁrmed all re-

sults.

The tests revealed signiﬁcant

differences bet-

ween the three input techniques regarding Reach

(χ

(2) = 13.317, p = .001

∗

), Regularity (χ

(2) =

44.163, p = .000

∗∗

), and ContinuousRegularity

(χ

(2) = 13.520, p = .001

∗

) while the differences for

Time were not signiﬁcant (χ

(2) = 5.040, p = .08).

Post-hoc analysis with Wilcoxon signed-rank tests

was conducted with a Bonferroni correction applied,

resulting in a signiﬁcance level set at p < .017. More

detailed results are reported in the following sections.

6.1.1 Reach

We report the values for the four Reach metrics (see

Table 1) and their aggregation (see Table 2). Statis-

tically signiﬁcant differences for Reach were found

only between the semi-touchless and touchful (Z =

−2.947, p = .003

) techniques. Differences between

touchful/touchless and semi-touchless/touchless were

not signiﬁcant (Z = −2.130, p = .033 and Z =

−2.060, p = .039). The results generally suggest that

participants could reach the maximum positions well

in all directions and with all techniques.

6.1.2 Regularity

Regularity is measured related to four directions (see

Table 1). The results for these metrics are then avera-

ged for an overall result for Regularity as reported

in Table 2. Generally, the semi-touchless techni-

que clearly outperformed the other two. It resul-

ted in a mean value of 93.68% (SD = 11.04), while

2∗

denotes a signiﬁcance level set at p < .05,

∗∗

a signi-

ﬁcance level set at p < .001

As a signiﬁcance level of p < .017 has been identiﬁed,

we do not use the standard APA notation (p < .05).

the touchful technique gained 82.13% (SD = 14.94)

and touchless input resulted in a mean of 43.91%

(SD = 16.9). 20 of the 25 participants gained their

individually best result with the semi-touchless input

technique, two with the touchful technique, and the

remaining three gained equally good results with the

semi-touchless and touchful techniques (none with

the touchless one). The differences between the in-

put techniques were signiﬁcant for touchful and semi-

touchless (Z = −3.072, p = .002), touchful and tou-

chless (Z = −4.372, p = .000) and semi-touchless

and touchless (Z = −4.372, p = .000) input.

For three directions (left, right, forward), Regula-

rity was best with semi-touchless input (with means

between 95.15% and 100%, however with high varia-

tion in two cases, see Table 1). The touchful technique

scored best for RegularityDown (M = 85.87%, which

is slightly better than the average result with the semi-

touchless setting, M = 84.26%, both with high vari-

ation). The touchless technique scored worst for all

directions (with means between 24.25% and 62.11%

for forward and right). This trend is conﬁrmed by the

participants’ individual results: no user gained any of

their individually best results with touchless input.

6.1.3 Continuous Regularity

Continuous regularity is measured for all four directi-

ons individually (see Table 1) and further aggregated

to ContinuousRegularity (see Table 2). For touchful

input, the mean values for the different directions

were between 76.47% (right forward) and 87.67%

(left down). For semi-touchless input, the results

were between 72.89% (right forward) and 86.41%

(left down). For touchless input, the results ranged

from 79.61% (right forward) to 82.50% (left down).

Interestingly, even if the differences were in most ca-

ses not too high, the coordinated movement was most

difﬁcult in the direction right/forward with all devi-

ces and least difﬁcult in the direction left/down for

all 25 participants. We attribute this to the fact that

for right-handed people a movement (or force ap-

plication) of the right hand to the left (i.e., towards

the body) is easier than to the right (away from the

body). For the aggregated ContinuousRegularity me-

tric, touchful input scored best on average, resulting in

a mean of 82.34% (SD = 3.84). The touchless techni-

que was slightly worse (M = 81.41%, SD = 16.49)

and semi-touchless was ranked third (M = 79.64%,

SD = 4.1). The statistical analysis showed that the

differences between touchful and semi-touchless in-

put were signiﬁcant (Z = −2.4082, p = .016). The

other comparisons revealed no signiﬁcant differen-

ces (Z = −1.574, p = .115 for touchful/touchless and

Z = −2.301, p = .021 for semi-touchless/touchless).

The Role of Haptics in User Input for Simple 3D Interaction Tasks - An Analysis of Interaction Performance and User Experience

Table 1: The computed values for mean, median and standard deviation of all metrics averaged for 25 participants. All metrics

except for the ones for Time (which are measured in milliseconds), are measured in percent. LF stands for left forward, LD

for left down, RF for right forward and RD for right down respectively.

Metric Touchful Semi-Touchless Touchless

M Mdn SD M Mdn SD M Mdn SD

ReachDown 100.00 100.00 0.00 100.00 100.00 0.00 98.33 100.00 4.08

ReachLeft 95.33 100.00 19.73 100.00 100.00 0.00 99.00 100.00 4.90

ReachRight 91.33 100.00 12.80 100.00 100.00 0.00 99.67 100.00 1.63

ReachForward 100.00 100.00 0.00 100.00 100.00 0.00 100.00 100.00 0.00

RegularityDown 86.87 100.00 21.70 84.26 100.00 34.28 28.78 28.24 21.97

RegularityLeft 91.85 100.00 22.00 100.00 100.00 0.00 60.48 66.97 26.61

RegularityRight 72.30 93.13 34.31 95.30 100.00 14.81 62.11 72.50 32.21

RegularityForward 77.49 100.00 30.20 95.15 100.00 19.87 24.25 19.89 22.74

Cont.RegularityLF 82.18 83.84 6.12 83.58 84.46 3.40 81.43 85.10 14.02

Cont.RegularityLD 87.67 88.30 3.57 86.41 87.95 4.67 82.50 91.40 23.05

Cont.RegularityRF 76.47 75.92 4.44 72.89 73.38 3.75 79.61 82.20 6.56

Cont.RegularityRD 83.05 82.84 4.61 75.68 80.26 13.02 82.08 91.13 24.36

TimeDown 1272 1002 901 1391 1156 856 1832 1251 1681

TimeLeft 795 653 353 940 801 458 1679 1100 2729

TimeRight 1114 1107 466 978 900 380 1797 950 3862

TimeForward 1356 1014 1032 1451 1151 783 2819 1401 3024

6.1.4 Time

We measure the average time a user took to reach each

of the target cubes in the Time test. Interestingly and

in opposition to our expectation (but consistent with

the ﬁndings of (Tscharn et al., 2016) or (Stannus et al.,

2011)), users were not fastest with touchless input;

the statistical analysis did actually not reveal any sig-

niﬁcant differences. We initially expected users to be

faster with touchless input due to the absence of phy-

sical resistance. Based on the results we assume that

this missing resistance led to an uncertainty regarding

whether the goal had already been reached or not (alt-

hough we visualized this). Due to the high variation

we searched for outliers and found a user who had

troubles with touchless input in general. Removing

this user from the data set reduces the variation con-

siderably but does not affect statistical signiﬁcance.

6.2 User Experience

UEQ includes a customized data analysis tool and

methodology which we primarily used for the analy-

sis of our results reported in this section. Although

usually analyzed on the descriptive level (which is

sufﬁcient for the evaluation of one condition), we ad-

ditionally ran a statistical analysis of the UEQ data,

in order to be better able to judge the actual differen-

ces between the settings on a statistical level. For this

analysis we used a repeated-measures ANOVA (as the

UEQ data met the ANOVA’s prerequisites).

6.2.1 UEQ Summary

The results for the UEQ are summarized in Figure 3

and Table 3. The scale ranges from −3.0 (conside-

red “horribly bad”) to 3.0 (“extremely good”). Ho-

wever, the authors of the UEQ consider it “extremely

unlikely” to observe answers < −2.0 or > 2.0 in real

applications. Values between −0.8 and 0.8 represent

a neutral evaluation (visualized yellow in Figure 3),

values > 0.8 represent a positive (green) and values

< −0.8 a negative one (red). Touchful input was eva-

luated positively for Perspicuity, Stimulation and No-

velty and neutrally for Efﬁciency and Dependability.

For semi-touchless input, the results were neutral for

Dependability and positive for all other categories (es-

pecially for Stimulation and Novelty). For touchless

input, all categories were evaluated positively.

6.2.2 Statistical Analysis

The ANOVA revealed signiﬁcant differences for De-

pendability (p = .022

∗4

) and Novelty (p = .036

∗

The pairwise comparison showed signiﬁcant differen-

ces only between the touchful and the touchless in-

put technique (p = .031

∗

). The differences for No-

velty were signiﬁcant only at the group level (pairwise

comparisons did not show signiﬁcant differences with

e.g., p = .059 for of semi-touchless vs. touchless in-

put). The statistical results seem to lower the weight

Greenhouse-Geisser correction was applied because

Dependability data violated the sphericity prerequisite.

HUCAPP 2018 - International Conference on Human Computer Interaction Theory and Applications

Table 2: Aggregated results for mean, median and standard deviation of Reach, Regularity, ContinuousRegularity and Time

averaged for the 25 participants, measured in percent (or milliseconds, for Time).

Aggregated Metric Touchful Semi-Touchless Touchless

M Mdn SD M Mdn SD M Mdn SD

Reach 96.67 100.00 6.36 100.00 100.00 0.00 99.25 100.00 1.85

Regularity 82.13 86.76 14.94 93.68 100.00 11.04 43.91 47.30 16.90

ContinuousRegularity 82.34 82.41 3.84 79.64 80.45 4.10 81.41 86.68 16.49

Time 1134 1026 466 1190 1081 484 2032 1250 2442

(a) Touchful (b) Semi-Touchless (c) Touchless

Figure 3: The UEQ results for all input techniques. The error bars represent the 95% conﬁdence intervals of the scale means.

Table 3: UEQ results averaged for the 25 participants.

UEQ-

Category

Touchful Semi-

Touchless

Perspicuity 1.560 1.520 1.740

Efﬁciency 0.670 0.873 1.140

Dependability 0.670 0.790 1.430

Stimulation 1.390 1.570 1.140

Novelty 1.440 1.580 1.170

of the UEQ results drastically at ﬁrst glance. Yet it is

surprising that the scores for three of ﬁve categories

are best for touchless input (even if this effect was not

statistically signiﬁcant for Perspicuity and Efﬁciency).

We had initially expected a contradictory result with

a strong tendency towards haptic scoring better than

touchless input. The results could indicate that UX

was inﬂuenced less than other factors by performance.

6.2.3 Comparison to UEQ Benchmarks

The UEQ data analysis tool also provides a bench-

mark data set that compares the evaluated system

against the responses of 9905 individuals from 246

studies on interactive products. The benchmark data

set helps to draw conclusions about the relative (UX)

quality of the evaluated product in comparison to ot-

hers. This was especially interesting to us as the sta-

tistical analysis revealed less conclusive insights than

expected. The results classify each of the evaluated

categories as excellent, good, above average, below

average, or bad (e.g., “good” means that 10% of the

results are better, 75% are worse). Figure 4 shows that

except for Dependability for touchful input (which is

rated “bad” in comparison), all UX aspects are at least

in an “okay” area. All aspects apart from Dependa-

bility and Efﬁciency for touchful and semi-touchless

input range from “good” to “excellent”. For touchless

input, the results for all categories are at least “above

average” (“good” for Perspicuity and Novelty).

6.3 Findings Related to the Hypotheses

We summarize the results regarding our hypotheses

as follows. We had to reject H1 as we did not ﬁnd a

signiﬁcant difference in Time. H2 was partially con-

ﬁrmed. Although the results showed unexpected sig-

niﬁcant differences between touchful (isometric) and

semi-touchless (isotonic) input, the differences bet-

ween the techniques that involve a haptic experience

and touchless input were not signiﬁcant. We could

conﬁrm H3 as touchful and semi-touchless were sig-

niﬁcantly better than touchless input regarding Regu-

larity. H4 had to be rejected as we did not ﬁnd signi-

ﬁcant differences between touchful and touchless as

well as semi-touchless and touchless input regarding

ContinuousRegularity. H5 and H7 also had to be re-

jected as the UX part of the evaluation revealed that

while two categories were actually better with tou-

chful and semi-touchless input, the remaining three

(including Dependability) were better with touchless

input. H6 could only partially be conﬁrmed. Sti-

mulation actually tended to be better for both input

techniques that involve a haptic experience which was

however relativized by the statistical analysis. We do

not reject H6 because the comparison with the UEQ

benchmarks categorized touchful and semi-touchless

as “excellent” while touchless input was only “good”.

The Role of Haptics in User Input for Simple 3D Interaction Tasks - An Analysis of Interaction Performance and User Experience

(a) Touchful (b) Semi-Touchless (c) Touchless

Figure 4: Comparison with the benchmark data set.

7 CONCLUSIONS

This section summarizes the work presented in this

paper and discusses its impact and limitations.

7.1 Summary

In this paper we have described and compared three

input techniques that rely on a different degree of hap-

tics. Aiming at analyzing the impact of a haptic ex-

perience on objective and subjective factors related

to interaction, we conducted a two-fold user study.

The ﬁrst part focused on objective interaction per-

formance metrics, the second emphasized perceived

UX. It was quite surprising that the results for the two

parts were relatively contradictory. While the semi-

touchless and touchful input techniques both outper-

formed touchless input regarding selected interaction

performance metrics (and touchless input was not sig-

niﬁcantly better than any of the other two at any me-

tric), three of ﬁve UX categories were evaluated bet-

ter in tendency for touchless input. Some factors that

can contribute to the ambivalence between objective

and subjective results are discussed as follows. (Ma-

cLean, 2008) describes that for humans “precision re-

quires resistance”, i.e., absolute position control is

difﬁcult and humans need something solid to push

against when we want to accomplish ﬁne position

control. Touchless input lacks this kind of supporting

guidance, which could have led to the inferior per-

formance. However, a lot of research has dealt with

the emotional quality of haptics in human-technology

interaction (Salminen et al., 2008) and showed that

there is an impact on human perception of interaction

when haptics is involved. This would support the

fact that the more emotion-leaning categories (grou-

ped under “hedonic quality” by UEQ) Stimulation and

Novelty were rated better for the techniques involving

a haptic experience, whereas the rather analytic ca-

tegories (“pragmatic quality”) Perspicuity, Efﬁciency,

and Dependability received a contrary result.

Based on our observations we draw the following

conclusions related to simple 3D manipulation tasks:

i) haptics does play an important role, especially in

terms of Regularity, ii) it however seems to be more

important for interaction performance than for UX,

and iii) a lower-performing technique can still retain

a good UX. It seems that in our study even the well-

known issue of missing physical constraints with tou-

chless input inﬂuenced interaction performance more

strongly than UX. Although in several cases our par-

ticipants did actually leave the sensitive area acci-

dentally (which negatively affected interaction per-

formance), they e.g., rated “good” 1.5 for the items

predictable/unpredictable and “excellent” 1.8 for un-

derstandable/not understandable. The only UX as-

pects that might have been affected negatively can be

found at fast/slow which got a score of 0.8 and mo-

tivating/demotivating (0.4). Although the UX scores

for touchful and semi-touchless input were in most

cases below the ones for touchless, most participants

verbally expressed that they liked the prototypes a lot.

Regarding the comparison of isometric and isotonic

devices, our study did not reveal signiﬁcant insights.

7.2 Impact and Limitations

Regarding the results of our study, these generally al-

low for conclusions that apply to other input settings

as well. The metrics chosen for the user study are re-

latively general and should be applicable for other in-

put settings, although their signiﬁcance might differ.

Reach, Time, and Regularity are metrics that are e.g.,

also relevant for interaction based on full-body mo-

vement with considerable more DoF or for conven-

tional mouse/keyboard or touch-based input (2DoF).

ContinuousRegularity is only important for settings

that allow for continuous movement-based input.

One limitation related to the interaction tasks is

that except for left/right, they focused on one di-

rection per dimension (i.e., forward and down). This

was on purpose as for touchful input it would have

been impossible to offer an upper wall without de-

priving users of the visual control of their hand. Re-

garding the direction back, adding another pressable

wall in the back would have been possible but dras-

tically changed the user’s interaction activity. In or-

der to ensure comparability of the input techniques

and prevent a bias based on the devices rather than

the input techniques, we focused on the directions

where activities were of sufﬁcient similarity for all de-

vices. Regarding UX, the good results for touchless

HUCAPP 2018 - International Conference on Human Computer Interaction Theory and Applications

input suggest that a missing haptic experience does

not necessarily have to lead to a bad overall UX. To

the contrary, the study has shown that a worse per-

forming device/input technique can still have a good

perceived UX. The good UX results for both touchful

and semi-touchless input in the category Perspicuity

have shown that users understood the input technique

quickly. Also, the good results for touchful and es-

pecially semi-touchless input in the category Stimu-

lation are conclusive (e.g., scores of 1.1 and 1.0 for

motivating/demotivating for semi-touchless and tou-

chful, 0.4 for touchless input). Some results (e.g., the

good results for semi-touchless and touchful input for

Novelty) might be less signiﬁcant as they might have

been biased by the prototypic nature of the devices

and the resulting “novel” impression. Yet, this poten-

tial bias is only present in a small part of the results.

Most semantic differentials suggest a clear focus on

the interaction with the device, not its appearance.

7.3 Subsequent Work

We expect our results to generalize to interaction tasks

in a similar complexity category while they might dif-

fer with varying complexity and DoF, which should

be analyzed in consecutive studies. We believe that

such studies will reveal interesting ﬁndings on the

dependence of need for haptics on task complexity

and expect that with increasing complexity of the in-

teraction tasks, the importance of haptics increases

as well. We already conducted a qualitative study

with users with motor and cognitive impairments, see

(Augstein et al., 2017a), which revealed a much stron-

ger dependence on haptic guidance, compared to non-

impaired users: almost all participants gained their

best results for Regularity, ContinuousRegularity and

Time with touchful input. Regarding subjective im-

pressions, results were relatively balanced.

REFERENCES

Atkins, M. S., Fernquist, J., Kirkpatrick, A. E., and For-

ster, B. B. (2009). Evaluating Interaction Techniques

for Stack Mode Viewing. Journal of Digital Imaging,

22(4).

Augstein, M., Kern, D., Neumayr, T., Kurschl, W., and

Altmann, J. (2015). Measuring Physical Pressure in

Smart Phone Interaction for People with Impairments.

In Mensch und Computer 2015, Workshopband, Stutt-

gart. Oldenbourg Wissenschaftsverlag.

Augstein, M., Neumayr, T., and Burger, T. (2017a). The

Role of Haptics in User Input for People with Motor

and Cognitive Impairments. In Proc. of AAATE 2017.

Augstein, M., Neumayr, T., Kern, D., Kurschl, W., Alt-

mann, J., and Burger, T. (2017b). An Analysis and

Modeling Framework for Personalized Interaction. In

IUI 2017 Companion: Proc. of the 22nd Int. Conf. on

Intelligent User Interfaces, Limassol, Cyprus.

Bowman, D., Coquillart, S., Fr

ohlich, B., Hirose, M., Kita-

mura, Y., Kiyokawa, K., and St

urzlinger, W. (2008).

3d user interfaces: New directions and perspectives.

IEEE Computer Graphics and Applications, 28(6).

Carter, T., Seah, S. A., Long, B., Drinkwater, B., and Subra-

mania, S. (2013). Urtra haptics: Multi-point mid-air

haptic feedback for touch surfacces. In Proceedings

of the 26th Annual ACM UIST Symposium. ACM.

Coelho, J. C. and Verbeek, F. (2014). Pointing Task Eva-

luation of Leap Motion Controller in 3D Virtual Envi-

ronment. In Proceedings of the ChiSparks 2014 Con-

ference, The Hague, Netherlands.

Dangeti, S., Chen, Y. V., and Zheng, C. (2016). Comparing

Bare-Hand-in-Air Gesture and Object-in-Hand Tangi-

ble User Interaction for Navigation of 3D Objects in

Modeling. In Proceedings of the TEI ’16. ACM.

de la Barr

e, R., Chojecki, P., Leiner, U., M

uhlbach, L., and

Ruschin, D. (2009). Touchless Interaction – Novel

Chances and Challenges. Springer Berlin Heidelberg.

ohlich, B., Hochstrate, J., Kulik, A., and Huckauf, A.

(2006). On 3d input devices. IEEE Computer

Graphics and Applications, 26(2).

urst, W. and Helder, M. (2011). Mobile 3D Graphics and

Virtual Reality Interaction. In Proceedings of the 8th

International Conference on Advances in Computer

Entertainment Technology, page 28. ACM.

Hwang, S., Bianchi, A., Ahn, M., and Wohn, K. (2013).

MagPen: Magnetically Driven Pen Interaction On and

Around Conventional Smartphones. In Proc. of the

15th Int. Conf. on Human-Computer Interaction with

Mobile Devices and Services, Munich, Germany.

Laugwitz, B., Held, T., and Schrepp, M. (2008). Con-

struction and evaluation of a user experience question-

naire. In Holzinger, A., editor, Proceedings of USAB

2008.

MacLean, K. E. (2008). Haptic Interaction Design for Ever-

yday Interfaces. Reviews of Human Factors and Ergo-

nomics, 4(1).

Rehg, J. and Kanade, T. (1994). Visual tracking of High

DOF Articulated Structures: an Application to Human

Hand Tracking. In Proc. of the 3rd European Confe-

rence on Computer Vision, Stockholm, Sweden.

Salminen, K., Surakka, V., Lylykangas, J., Raisamo, J.,

Saarinen, R., Raisamo, R., Rantala, J., and Evreinov,

G. (2008). Emotional and Behavioral Responses to

Haptic Stimulation. In Proc. of CHI’08. ACM.

Stannus, S., Rolf, D., Lucieer, A., and Chinthammit, W.

(2011). Gestural Navigation in Google Earth. In Pro-

ceedings of the 23rd Australian Computer-Human In-

teraction Conference.

Tscharn, R., Schaper, P., Sauerstein, J., Steinke, S., Stiers-

dorfer, S., Scheller, C., and Huynh, H. T. (2016). User

Experience of 3D Map Navigation – Bare-Hand Inte-

raction or Touchable Device? In Mensch und Compu-

ter 2016. GI.

Zhai, S. (2008). Human Performance in Six Degree of Free-

dom Input Control. PhD thesis, University of Toronto.

The Role of Haptics in User Input for Simple 3D Interaction Tasks - An Analysis of Interaction Performance and User Experience