Ubicomp 2005
By John Latta, WAVE
0540 10/7/05
Tokyo , Japan
September 11 – 13, 2005
This is the first Ubicomp to be held in Asia. The Japanese
hosts have gone to great lengths to be good hosts and keep the costs
down for the attendees. It is being held at a first class hotel and the
after hours events are likewise first class. There was a rigorous process
for paper acceptance and only 22 full length papers were accepted. There
are 47 posters and 25 demos. The attendance is near 550 and spans 21
countries. Over 50% came from Japan and with 97 from the US. The event
represents a good cross section of the research in ubiquitous computing.
Sony Looks Forward
The keynote was given by Mario Tokoro, President of the
Sony Computer Science Laboratories. His talk: Tools, Environment and
Society, provides one of the most clearly articulated views of where
computers could go in the next 10+ years.
Three slides were used which covered the:
History of Computing
History of Computer Networks
History of User Interfaces
From these slides it was conclude that major waves in
computer technology take 30 years to mature.
The next slide was on the History of Ubiquitous Computing.
With a beginning in 1990, it was speculated that we are only 15 years
into the development of this wave of computer technology. Numerous
example of technology were cited as building the transition to this
phase. They included:
ActiveBadge, mobile phone with camera, blogs and GPS.
Sony CSL is an independent company founded and supported
by Sony in 1988. Mario Tokoro was one of the founders. It has only
30 researchers divided approximately evenly between engineering and
science. There is no technical support staff and what drives the work
is the ability to contribute to society. This is accomplished by hiring
world class research individuals and making contributions to social
and industrial advancement through fundamental yet applicable research.
The annual turnover is about 10% per year.
The research themes are broken into engineering and science
disciplines.
Engineering
Interactive Devices
Ubiquitous Networking
Computational Photography
Lifeline Computing
Entertainment
Science
Econophysics
Evolution of Language
Systems Brain Science
Statistical Physics of Noise and Delay
Cognitive Robotics
Systems Biology
Mario Tokoro was proud to show how CSL’s contribution
to ubiquitous computing dates back to 1994. He showed video examples
of 5 of the contributions which Sony CSL has made.
The progress of technology comes in 3 phases: first are
the tools, then the environment and then to society. We have just begin
the society phase.
In the tools phase, technology empowers the individual.
In the second phase, computers support the environment
which in turn supports human activities.
The last phase occurs when computers become a means to
construct a society and enrich diverse social relationships.
When computers help form an environment they can be combined
with sensors and actuators which do things. One of the important problems
is that there are societal issues when computers are everywhere and
privacy and security becomes a major issue. There is a dark side to
the creation of such environments, that is, when the environment knows
a great deal about individuals and what they are doing, negative impacts
may be heaped on individuals. Many of these issues have not been fully
addressed by society.
This gives rise to the notion of Intimate Computers which
are a trusted buddy through which the owner interfaces to the environment.
There are two parts to the environment: Closed and Open. In a closed
environment, computers are closely associated with the individual.
In an open environment, computers are a part of broader society. This
was the basis for Tokoro-san’s conclusion that we must only examine
the open environment as a systems issue. Nothing is isolated.
Open was also used to support the notion of a Volitional
Agent – an agent which has its own will for building safer and
more robust systems.
As a means to enter the 3rd phase, building a society,
it was observed that elements of this are already happening: blogs,
Wiki, P2P, VoIP and SNS. One of the society problems is that there
is an anonymity issue. A person may have more than one identity. This
brings rise to the virtual vs. real issue. That is how does one develop
true understanding and a shared common experience on the net when identity
is fluid?
The problem which these issues raise is that policy,
governance and society have not yet established the means by which
to form societies. The well know issues of data protection, security
and privacy have not been adequately addressed.
CSL is now going well beyond its early roots to research
the broader context of the environment and society. New research has
been started in:
Systems Biology – From gene to protein to systems.
This is part of the quest for the fundamental principles of life.
EconoPhysics – The link between macro and micro
economy.
In summary, Mario Tokoro asked of Ubiquitous Computing:
For What?
Human Well-Being
Safety
Excitement/Entertainment
Ease of Use/Convenient Lifestyle
For Whom?
Individual
Society
Earth
How to Build?
Not as individual technologies but as a total system – a
holistic view
His closing words were quite thoughtful:
Science is to understand nature in depth, by searching
the truth.
Engineering is to create value for individuals by making
products.
An Individual is the basis for existence
We need to understand more about each other.
Culture is the man-made value that cannot be detached
from society
A society can only be created and made better by Individuals
We need to create technology that can be used to make
a better society.
The WAVE spoke with Tokoro-san after his talk. We asked:
if ubiquitous technology is to make a better environment why is not CSL
doing more work with robotics, in which Sony has a strong position? Yes,
there is technology but societal issues are a barrier. For example, one
of the reasons that Sony has decided not to make QRIO into a product
are the liability issues. These make other factors pale in comparison.
During the question phase one which came up was – How
is it possible to continue to defend such basic research in today’s
quarterly financial reporting environments. The short answer – it
is difficult.
CarpetLAN – Where the individual is a LAN element
NTT DoCoMo and NTT Microsystem Integration Labs described
a novel indoor wireless-like networking and positioning system. This
fit into the category of enabling infrastructure. But it went beyond
this – making the human a part of the network.
The problem with CarpetLAN is addressing the issue that
every object in a room should be connected to a network. This is the
last foot problem.
Today with wireless technology, only a few uses and objects
can be connected to the network – what is needed is the capability
for 1,000’s of objects be connected and this cannot be met by
radio technology.
The solution proposed is a wireless (like) network and
positioning system based on a electric field. In fact, the human body
operates like an Ethernet cable. It is possible to realize cell sizes
as small as 1 meter.
Intrabody communication is made possible when wearable
devices are in contact with the human body via an insulted electrode,
the user stands on a carpet electrode and the wearable devices and
floor transceiver connects to the human body.
Using carpet squares 100 cm square, it is possible to
achieve 20dB separation between the cells. When an individual stands
on the carpet, they make an excellent receiver and signal transmission
route. It is expected that it will be hard to increase the human node
beyond 100Mb/s.
Interference between adjacent human bodies can be a problem
if individuals are standing close together.
The carpet structure consists of a base box with electrodes, a
carpet insulator and a node unit which connects the networking to the
carpet.
CarpetLAN is able to provide up to 10 handovers/sec and
thus used to detect individual moving over the carpet. Graphs were
shown of individual acceleration and angular velocity on the carpet.
A video was shown where an individual was able to play
streaming video on a PDA while standing on the carpet and it stopped
when they walked off.
Position tracking was shown including the ability to
track 3 persons at a time.
One of the concerns about this concept is the current high
pricing. It cost about $5,000 per node and uses 8W. Efforts are being
focused on miniaturization and reductions in cost in power consumption.
Innovative.
Surface Interaction – Is it a Wave of a New Computer
Interface?
MERL (Mitsubishi Electronics Research Laboratories) in
Boston has worked over the last 4 years to develop table top interaction
technology. It has now reached a production state and development kits
will go on sale in October. This is a limited step but an important one
in opening the market to surface interaction technology. The WAVE has
seen examples of this technology at multiple conferences over the last
2 years. Some of the results have been impressive.
MERL’s technology to support surface interaction
is called Diamond Space. This was created in a research laboratory and
is now reaching market due to the demands of those that have seen and
tried it. The first steps are small but significant if surface interaction
is to be an important force in computer interaction. Started 4 years
ago as the combination of hardware which uses a projector to create the
surface, i.e., table top, interaction and a software tool kit which supports
interaction, this technology will go on sale in October. It is a development
that merits watching. If the technology takes off, it could have important
implications for how computers are used in a group environment.
Kathy Ryall, Principal Technical Staff, Technology Lab,
MERL, attended Ubicomp to present a poster paper and show a demo of the
table. The WAVE spoke with Kathy about the state of the technology and
its emergence in the market.
The technology which MERL has developed is Diamond Space
andit consists of the hardware and software to support multi-user and
multi-surface interactions. Currently the hardware uses an overhead
projector and table surface for interaction. But the surface may be
vertical, horizontal or even non-planar. The table component is called
Diamond Touch.
The significant contribution of Diamond Space is that
it enables multi-user co-located or remote computing. Compared to a
traditional single user desktop, each interaction is user differentiated – that
is, the computer knows what every one is doing on the surface. This
is not only essential to provide the interface but also to provide
for a complete audit trail and logging for interaction.
Collaboration is enabled with Diamond Spin because it
makes interaction on an individual level possible within a group setting.
Some of the user results, as the WAVE reported from CHI 2004 in Vienna,
were impressive – Finger Talk research which was done at
Indiana University. This changes the role that the computer has in
a collaborative environment.
MERL has also developed a Java Took Kit called DiamondSpin
which is a higher level set of tools to create a user interface. Some
of the means of interaction addressed with DiamondSpin included surface
space management with “Black-hole” and “Fisheye” approaches.
When the WAVE saw this at CHI in Vienna we were impressed.
In spite of the emphasis on horizontal surface interaction,
the technology has been used in vertical surfaces. Yet, Kathy described
vertical and horizontal surfaces as having different forms of group
interaction. In fact, some projects have developed interactions which
use both horizontal and vertical surfaces. Further, the nature of interaction
is based on the size of the group. MERL has found that these dynamics
can change as the groups go from 2 to 8 devices to 20.
The technology has created its own “religious
issues” which are the subject of much debate in MERL. These include:
Vertical vs. Horizontal Surfaces
Single vs. Multi-user Interaction
Size of Group and the Interaction Dynamics
Ultimately, the market will decide many of the factors
which drive these issues.
One of the impediments to the market is the lack of multi-user
support in Windows. For true collaboration to take place, Windows should
support the independent interactions by N users. As Kathy asked:
What is the user experience like when 4 mice are plugged
into a USB hub on one machine?
As word of the technology spread in 2001, MERL made
a decision to deploy units, basically for free, to evaluate the technology
in ways that it could not do. In 2002 – 2003, 90 table top units
were built of which about 50 went to research labs outside of Mitsubishi
Electric. 90% of those included the hardware and SDK. Another 10% also
received the Diamond Spin tool kit. Many of these systems are still
in use today. Some of the areas of use include:
A CEO that shows how an executive can use such technology
in daily work, including one-on-one interactions with another person
at his table.
Elderly remote interactions where two seniors can interact
remotely with their grandchildren.
Augumented reality which has a 3D overlay to a scene.
A museum in Italy which uses the surface for story
telling.
A means for remote interaction between children.
Teams of biologists which use the surface for interaction.
The table as a means to interact with physical objects
that are placed on it.
Using the table to conduct opportunistic browsing of
photo databases.
The table as a game play surface.
GIS applications of the table for single and multi-person
interaction with large spatial information.
This latter use has generated considerable interest in
the GIS community and they were one of the most vocal in requesting
that units be made available of sale.
This seed effort has generated considerable feedback
and more continues to be received by MERL. One of the advantages of
such a deployment is that the current design is quite robust – it
has been made “bullet-proof” by use and the subsequent
refinement.
The Developer Kits, which will go on sale in October,
will cost approximately $10,000. There will be two display sizes available – 80cm
and 107cm diagonal. The kits will be manufactured by a 3rd party in
the US. The hardware will support 4 users simultaneously. Accessories
will also be available to augment the table.
The gaming market for table interaction is a natural
one because most games are multi-person. A demo unit has been set up
for gaming. Further, there have been other activities in the game space
that Kathy was not a liberty to discuss.
As sign of the growing interest in the table as a user
interface paradigm is the conference.
First IEEE International Workshop on Horizontal
Interactive Human-Computer System (TableTop 2006)
January 5-7, 2006
Adelaide, South Australia
http://www.tinmith.net/tabletop2006
Mitsubishi Electric in Japan has yet to identify how
it will support the commercial technology. MERL continues to get strong
support from its parent company but it is unclear if Diamond Space
market is large enough to merit a major product effort by the parent.
Due to its size, it seeks very large new markets in order for it to
impact the business. Thus, the Developer Kits are a means to explore
the market and let it be further refined by the users. It could well
be that market development will take place by licensing, a spin-off
or commercialization with Mitsubishi Electric. The next 6 months, which
will see the response to the developer kit, is an important market
indicator.
One of the strengths that Mitsubishi Electric brings
is its projector technology. This is one of the keys to Diamond Space
and is, at the same time, responsible for the highest cost component.
If the market should develop, Mitsubishi Electric is in a strong position
to drive the cost down with volume.
At the same time all of this is happening, MERL is looking
to the next generation. This includes:
Support for rear projection.
Ever increasing demands for larger surfaces, both
table and wall.
Much finer grain user input of which a stylist is
frequently mentioned.
On the software side, support for FLASH is expected
in the next 6 months
The WAVE asked what is the competition? At this time there
are basically one-off systems and nothing which is close to commercial
scale. Thus, Diamond Space is positioned to see its research transition
to a commercial venture – with the first small step being made
with the developer kits. The market is getting ready to respond. MERL
will be watching with great interest to see what the future of the technology
it has created will be.
Fujitsu – Making the Palm a Biometric
Tucked away in the demo section of Ubicomp was a demonstration
of a palm vein biometric. This is not a research effort but a product
that has been on sale for a year. The experience was positive and the
claimed performance even more so.
It all seemed so simple – an inverted plastic square
device that one lays their hand on. Three clicks and the user is registered.
Then, as I shifted my hand from right to left, I was denied entry. This
is the first time we have seen “Palm Vein” as a biometric.
In commercial deployment for a year, it has been deployed in Japan at
bank ATMs and is now being deployed in other applications worldwide.
The detection is by IR and the sensor is quite simple.
In fact, the next version of the sensor will be reduced so that it can
fit on a notebook. The claimed results are:
FRR - .01%
FAR - .00008%
This seems incredible and independent testing would be
merited.
When we used it, the advantage was clear – enrollment
and use was easy, as much so if not better than a fingerprint.
The WAVE spoke with Fujitsu about their palm vein biometric
technology.
They will target the computer log on market with the
new sensor that can be integrated into a notebook.
The sensor does actually image the hand with IR to detect
the veins. The template created for matching is actually quite small.
One of the advantages is that the algorithm for vein detection actually
has a means to determine if blood is flowing and they feel it is not
subject to liveness spoofs.
The characteristics of the lens to do the imaging seems
unusual in that the focal length from the bottom of the hand jig to
the sensor is only about a cm. This could imply a fresnel lens but
no information was gained on the imaging design or the size of the
sensor in the current configuration.
No information was provided on the notebook sensor other
than it will be about 1/4th the size of the existing one and it will
also image the palm. Thus, such information as
How can the distance be maintained from the palm to
the sensor be maintained in a notebook to allow for imaging?
Will the smaller sensor image cover the full palm?
These remain unanswered. The response was - contact with
the technical team.
Fujitsu recognizes that in order to get acceptance in
a world wide market the sensor must be subject to a number of independent
tests to verify performance and to evaluate other issues such as enrollment
and system integration. Fujitsu has not undertaken to have such tests
done. They have not yet identified a firm that will do this work.
Fujitsu will exhibit at the Biometric Consortium conference
next week. This has implications that the company is seeking to enter
the broader biometric market.
The Role of Context in Information Management in the Home
The University of Calgary presented the best assessment
of information management in the home seen to date. But there is no bottom
line – just rich insights.
Kathryn Eliot, University of Calgary, Calgary, Alberta,
Canada presented a paper - Time, Ownership and Awareness: The Value of
Contextual Locations in the Home. A study was done of 10 households,
29 individuals with widely varying demographics and types of homes. What
the study sought to do was to classify information elements in the home
according to the following:
What is it?
Whose is it?
What needs to be done with it?
When do I/others need to interact with it?
The major value of this research is that they were able
to categorize the various types of information in the home.
Reminders, Alerts and To-do Lists
Awareness and Scheduling
Visual Displays
Notices
Resource Coordination
Each information type has its own goal in the home. This
was especially rich and meaningful in home management. What was also
important is the role of context. The research found that information
had value “…because it was there.” and there are specific
locations in the home for information and these locations are also related
to the contextual information. This became a filter and means to manage
the information.
Another set of factors included: Time, Ownership and Awareness.
This was also rich in meaning. For example, awareness had elements of
presence and monitoring.
One of the most important implications of the study is
that existing technology – that is, the PC, has significant limitations
because of its inability to take advantage of the importance of context
and location.
The overall implication of the study is that home information,
and all its complexity, is an opportunity for truly ubiquitous computing.
There are also some new uses for current technology.
In the question session it was asked – how does this
directly impact the development of new appliances for the home? The response
was not insightful and it is clear much thought, and likely more research
is required, to translate the insights gained into action. Yet, this
is the most insightful of the work we have seen on home information.
WAVE Comments
Here at Ubicomp, this is the crowd of serious believers
that ubiquitous computing will be the future of computing and the role
it will play in individual lives and society. Yet, we struggled with
the idea that ubiquitous computing has even reached the point of being
able to make important contributions to society and life. Data points
at Ubicomp include:
A number of the papers were just trivial. In spite of
the fact that there was a rigorous selection process, it seemed that
the criteria was on accomplishing the façade of the scientific
method, not the same context that ubiquitous computing promises to
bring value to society.
A substantial percentage of the papers were on location
technology. Yet, when it was asked – I am having a hard time
understanding where this is leading us, no one wanted to address the
question.
Some of the research was biased. For example, Intel Research
did a study of Global cities. The demographic focus was on individuals
aged 22 – 32 with no children. This population was characterized
as that which cities such as Los Angeles, London and Tokyo were “designed
for.” The WAVE spoke with the author and asked – could
you give this paper before the city council and represent tht the city
was not designed for families, ethnic diversity and a broader age group?
There were a number of questions – did you consider
this work and how does it relate to your findings? Many seemed unaware
of past research.
Many of the papers reached the conclusion – it
is on the cell phone, or this is the best place to provide it because
we all carry cell phones. This certainly reinforces the value of cellular
technology and its near ubiquity, but it hardly assesses the role that
ubiquitous computing can play.
There is no strategic context. The only paper which provided
this was by the CEO of Sony CSL.
For the ubiquitous computing to impact society it must
bring value and have long term impact. This was missing at Ubicomp. At
times there were small hints of value but overall the movement forward
was random at best.
|
