Automotive User Interfaces and Interactive Vehicular Applications
Automotive User Interfaces and Interactive Vehicular Applications
Automotive User Interfaces and Interactive Vehicular Applications
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Speech, buttons or both?<br />
A comparative study of an in-car dialogue system<br />
ABSTRACT<br />
In the DICO project a user test was performed to evaluate<br />
an in-vehicle dialogue system with three different modalities;<br />
speech user interface (SUI), graphical user interface (GUI)<br />
<strong>and</strong> multimodality (MM). Task performance, task completion<br />
time <strong>and</strong> driving ability were measured. We found that<br />
although the GUI was fast <strong>and</strong> easy to use, the participants<br />
experienced that their driving ability was better when using<br />
the SUI <strong>and</strong> multimodal interfaces, <strong>and</strong> the multimodal<br />
interface was preferred over GUI <strong>and</strong> SUI.<br />
Categories <strong>and</strong> Subject Descriptors<br />
H.5.2 [Information <strong>Interfaces</strong> <strong>and</strong> presentation]: <strong>User</strong><br />
interfaces—Evaluation<br />
General Terms<br />
Design<br />
Keywords<br />
Evaluation, in-vehicle, dialogue system, performance, cognitive<br />
load.<br />
1. INTRODUCTION<br />
This paper describes an empirical investigation of an invehicle<br />
dialogue system comparing three different interaction<br />
modes (voice, manual <strong>and</strong> multimodal interaction) with respect<br />
to task performance, (self-estimated) driving ability<br />
<strong>and</strong> task completion time. In order to let the driver keep<br />
her h<strong>and</strong>s on the steering wheel <strong>and</strong> the eyes on the road,<br />
dialogue systems are gaining an increasing interest from the<br />
car manufacturers. At the same time, the usage of mobile<br />
Copyright held by author(s)<br />
<strong>Automotive</strong>UI’11, November 29-December 2, 2011,<br />
Salzburg, Austria<br />
Adjunct Proceedings<br />
Jessica Villing, Staffan Larsson<br />
Department of Philosophy, Linguistics <strong>and</strong><br />
Theory of Science, University of Gothenburg<br />
Olof Wijksgatan 6<br />
Gothenburg, Sweden<br />
{jessica, sl}@ling.gu.se<br />
phones while driving is a much debated issue, for safety reasons.<br />
The main objection is that the mobile phone conversation<br />
takes the attention from the driving task, <strong>and</strong> the concern<br />
is that the usage of in-vehicle dialogue systems would<br />
show a similar pattern. However, since mobile phone conversations<br />
<strong>and</strong> dialogue system interactions are very different<br />
from each other, this type of comparison is not necessarily<br />
very meaningful. Furthermore, the alternative to dialogue<br />
system interaction is not no interaction, but interaction using<br />
manual (haptic) input <strong>and</strong> graphical output (as in traditional<br />
GUIs). To find out if a dialogue system is a reasonable<br />
alternative to manual input we therefore wanted to compare<br />
different user interfaces that is designed for performing the<br />
same tasks.<br />
The overall conclusion is that the multimodal condition gives<br />
the same task performance as the manual condition, the<br />
same driving ability as voice condition, <strong>and</strong> beats both with<br />
respect to driving ability.<br />
2. RELATED WORK<br />
There has been a number of similar studies carried out before.<br />
For example, the SENECA project evaluated manual<br />
vs. speech input, when using a Driver Information System<br />
(DIS) containing radio, CD player, telephone <strong>and</strong> navigation<br />
[2]. 16 subjects compared a manual DIS system with a DIS<br />
system that was equipped with a speech input system. The<br />
findings from the study shows that a multimodal application<br />
is preferable, that safety can be improved by using speech,<br />
<strong>and</strong> that the drivers feel less distracted when using speech.<br />
Another study, commissioned by Nuance <strong>and</strong> performed by<br />
the HMI laboratory of the Technical University of Brunswick,<br />
Germany [8], shows similar results. When comparing a manual<br />
interface to a speech interface, it was found that using<br />
a manual interface resulted in greater distraction <strong>and</strong> an increase<br />
in the number of times drivers looked away from the<br />
road. When using the manual interface the drivers rated<br />
their driving ability as poorer, <strong>and</strong> a Lane Change Task<br />
(LCT) showed a significant decrease in driving performance.<br />
Medenica <strong>and</strong> Kun [5] compared interacting with a police radio<br />
using the existing manual interface to interacting using<br />
their Project54 speech interface, which was adapted to function<br />
in the same way as the manual interface. The graphical<br />
interface, unlike the SUI, resulted in a significant degradation<br />
in driving performance.