Last: Mobile HCI 2004: Device Differences and Web Pages
Next: Mobile HCI 04: Speech Interfaces
September 16, 2004
Mobile HCI 2004: Novel Interaction Techniques
MobiVR: large virtual touch screen for mobile devices
Tiiu Koskela
Tampere University
Wireless handheld near-eye microdisplay (two eye non seethrough)
pointing as the input method
two hand operation - pointing and selection
focus groups - early adopters, teens, non-technically minded
raised concerns on social and health issues, acceptance
size of device more important than size of the screen
two-handed coordinated use hard
missing feedback of cursor control
missing support for social activities
physical strain
requirements for next prototype:
only one eye covered
direct object manipulation
natural gestures
multimodal input methods
augmented reality
feedback in visual and aural
*****
Handy: a new Interaction Device for Vehicular Information Systems
Sergio Di Martino
current in-car systems have heavy visual workload on users
traditional HCI approaches cannot be effectively applied to the automotive domain
related work
Alfa Romeo Connect
audi / volkswagen
jaguar s-type
bmw iDrive
no standard at all
working with Fiat research centre Elasis
requirements:
reduce distraction
easy to use for naive users
quick to use for expert users
cost effective to industrialise within 2-3 years
interactions -
search for hardware on dashboard
place hand/finger on (small) control
interact to achive goal
try to minimise dis traction of the first two
handy - a remote device placed on the driver seat arm rest behind the gearstick
a rotary wheel and 4 soft buttons
main advantage is that position is standardised and displacement of fingers always the same
handy provides at most 7 actions (fits with 7+-2 idea - idrive has 10)
interaction modality and view modality, depending on whether controls needed to interact are shown
flicks into interaction modality when hand is placed on handy
limited haptic feedback on scroll wheel
going to be tested in a car simulation
main drawback so far - only 3 buttons leads to deeper menu hierarchies
*****
Interactive Positioning based on Object Visibility
Christian Kray
GPS doesn't work [hooray]
use the user as the information source
interact with the user about what they can see
need objects and position hypotheses
assumptions-
user can determine object visibility
system can compute object visibility
problem-
very large number of potential questions
select *good* objects for questions
users a matrix, that is reduced as information is found out
functional prototype (Deep Map)
significantly improve precision of positional information in case of imprecise sesnor data or in absence of any data
field trial (19m precision in street of 170m length in three steps)
in future, visibility check could be done by computer (cameraphone sends images to server)
*****
IDeixis - Searching the Web with Images for Location-Based Information
Konrad Tollmar
CSAIL, MIT
send a picture to the server, get back links to webpages with images similar to the picture you took
landmarks pretty easy to recognise from a computer perspective
normally occupy most of the image
people tend to take images from the same location
asked students to take pictures of three locations on campus
about 60% accuracy in image matching (difference between similar images and relevant images)
(Nokia sponsored project)
'irrelevent' web pages - photo albums - no info about what iamge is about
interview study with tourists - how do people use maps and guides?
moved to standard websearch, keyword boosting through keywork extraction from found webpages, and MapQuest
finding location images
many pictures on the web
we can remove thos ewith people in
and those that are man-made
Trackbacks
TrackBack URL for this entry:
http://www.anti-mega.com/cgi-bin/mt/addmttb.cgi/1109
