UBC Publications

UBC Publications

UBC Publications

Innovations 2014

Item Metadata

Download

Media
focus-1.0225871.pdf
Metadata
JSON: focus-1.0225871.json
JSON-LD: focus-1.0225871-ld.json
RDF/XML (Pretty): focus-1.0225871-rdf.xml
RDF/JSON: focus-1.0225871-rdf.json
Turtle: focus-1.0225871-turtle.txt
N-Triples: focus-1.0225871-rdf-ntriples.txt
Original Record: focus-1.0225871-source.json
Full Text
focus-1.0225871-fulltext.txt
Citation
focus-1.0225871.ris

Full Text

Array $3.99 CANADA
Also
REHABILITATION ROBOTICS
POWER CONVERSION FOR RENEWABLE ENERGY SYSTEMS
HIGH DYNAMIC RANGE VIDEO
SPORTS PLAYER TRACKING
and more...
magazine
Fall/Winter 2014
eHEALTH:
LEVERAGING NEW
TECHNOLOGIES TO
PROMOTE HEALTH
innovations
innovations
fall/winter 2014
Production Sharon Cavalier
Editor 	 ICICS Administrator
Writer 	 Craig Wilson
ICICS Communication Writer
Design 	 Industry Design
	
	
www.industrydesign.ca
Office ICICS
	 University of British Columbia
	 289-2366 Main Mall
	 Vancouver, BC, Canada V6T 1Z4
Tel: 604-822-6894
Fax: 604-822-9013
Email info@icics.ubc.ca
4–5 director’s desk}
eHEALTH
Leveraging new technologies to promote
health
6 –7
ADAPTIVE USER INTERFACES
Customizations according to
user needs
18–19
BRINGING MACHINE VISION
TO THE CONSUMER
ICICS start-up company
develops devices for consumer
safety and convenience
8–9
REHABILITATION ROBOTICS
FOR STROKE VICTIMS
Robotic therapies that speed recovery
2 Fall/Winter 2014
16–17
MAGIC: 25 Years Later
Human-computer interaction and
graphics centre at ICICS
Welcome to the latest edition of ICICS Innovations
magazine. I would like to take this opportunity to
introduce you to a new model for collaborating with
industry developed by ICICS that takes advantage of
our unique pan-university status.
Under this model, a blanket intellectual-property agreement is negotiated
between the company and UBC, freeing company managers of the need to
negotiate one-on-one agreements with individual researchers. ICICS works
with the company to define one or more broad-based research projects
in support of their R&D goals. We then assemble research teams from
our 160-strong cross-faculty membership, pursue additional government
funding, and manage the projects. Tangible results are realized for companies,
with direct benefits for their innovation strategies. Researchers and students
gain support, industry exposure, and perspective.
This new approach has led to the ICICS/TELUS People and Planet
Friendly Home sustainability initiative, which we profiled in the Fall/
Winter 2012 issue. This broad-based endeavor involves 25 professors and
37 students from 4 different faculties, and is supported by a 6-member
industry consortium led by TELUS. The project continues to grow, attracting
additional industry and government funding. A site has been selected in
the central campus to build a prototype home and broaden the scope of the
integrated research efforts.
We look forward to additional meaningful collaborations with industry
emerging from this new collaboration model. In the meantime, I hope you
enjoy the sampling of UBC research achievements we provide here.
Panos Nasiopoulos, ICICS Director
12–13
HIGH DYNAMIC RANGE VIDEO
Delivering the “wow” factor
14–15
SPORTS PLAYER TRACKING
Advanced computer vision
techniques for tracking
sports players
10–11
POWER CONVERSION FOR RENEWABLE
ENERGY SYSTEMS
Providing the missing pieces
innovations magazine Fall/Winter 2014 3
by Shawn Conner
This is one of a number of questions facing Dr.
Kendall Ho. A professor in the Department of
Emergency Medicine, Ho is also a Director of the
eHealth Strategy Office in the Faculty of Medicine
at UBC. As Director, Dr. Ho leads an interdisciplinary team of
researchers conducting research into eHealth and technology-enabled
knowledge translation. They are focusing on how to
best harness modern technologies to enhance health services
and accelerate the incorporation of the latest health evidence
into routine practices.
“We are very interested in understanding how modern
information technologies, like mobile phones, tablets and
computers, can be used optimally for healthy pursuits,” Dr.
Ho says. “The motto of the strategy office is ‘Health Intercon-nected:
Connecting People to Better Health.’ We use the word
‘better’ as a verb.”
The research helps policy-makers like the B.C. Ministry
of Health’s Patients as Partners branch develop strategies to
engage the general and multicultural public in using technol-ogy
for their own health. “And then we carry out education for
health professionals, including medical students,” Dr. Ho says.
The eHealth Strategy Office opened in September 2008. At
that time, the Dean of UBC’s Faculty of Medicine appointed
Dr. Ho as Founding Director. But Dr. Ho’s work in eHealth
WITH THE FAST PACE OF TECHNOLOGY DEVELOPMENTS IN THE
AREA OF HEALTH CARE SERVICES DELIVERY,HOW CAN
STUDENTS, PROFESSIONALS AND PATIENTS KEEP UP?
eHealth:
Leveraging New
Technologies to
Promote Health
by Shaw n Coner
4 Fall/Winter 2014
eHealth Advantages
Using technology to promote health offers advantages across
the sector. For health care professionals, technology offers the
opportunity to measure, quantify, analyze, and disseminate re-search
and knowledge in unprecedented numbers and specif-ics.
Meanwhile, health consumers are finding useful informa-tion
to support their own health, and have become better able
to connect both with other patients with the same diseases,
and with health professionals. And government can use solid
health data in terms of understanding disease patterns and
health trends and where to put its resources. As an example,
Dr. Ho cites the rising number of people with diabetes, and
the subsequent rise in costs of care. “How do you dedicate re-sources
appropriately to the type of services needed?” he asks.
Knowledge from eHealth initiatives can help policy-makers
make these types of decisions.
Technology also offers challenges, however. “Is the infor-mation
accurate?” Dr. Ho says. “How do you deal with inac-curate
information? How do you make sure the information
one obtains is appropriate for that particular individual? How
do we, in this new environment, form this health professional–
patient relationship, so that we can support each other?”
Diabetes Management
Currently, Dr. Ho is spearheading several eHealth-related
projects that look at these and other eHealth challenges. One
project, on diabetes, is supported by the Lawson Foundation,
a Canadian philanthropic organization. The project helps peo-ple
with diabetes and caregivers manage the condition using
sensors, social media and text messaging. The project’s partici-pants
use the sensors to measure their weight, blood pressure,
and glucose level. Social media links them to their fellow pa-tients
and family members to discuss the best ways to manage
the disease. Supportive tips are provided by health profession-als
via text messaging.
This research has found that patients who follow the
program are better able to control their blood pressure, and
improve their A1C levels, an indication of how well they are
able to control their glucose levels over time. “They actually
have some desirable weight reduction,” Dr. Ho says. “They
found that the system is very supportive to their health. At the
eHealth Strategy Office, we don’t invent the sensors ourselves,
we don’t invent social media, but we apply these different
kinds of technology in the health context to support patients.”
Engaging Communities
Dr. Ho is also working with the Ministry of Health’s Patients as
Partners initiative on two programs. One is the Intercultural
Online Health Network (ICON). It uses multimedia channels
such as live forums, Web 2.0, YouTube, and social media to
engage the Chinese and South Asian general public, and to
present material in their native language to support them in
chronic disease management. “We found that seven out of ten
people who attend one or more of our different media chan-nels
want to make a big change in their lives to improve their
health,” Dr. Ho says. “Three out of four people want to make
at least a small change to live a healthier life.” Another sign of
success: some temples in South Asian communities have be-gun
changing their menus to reflect diabetes-related concerns.
Another collaboration with the Ministry of Health looks
at methods to introduce eHealth technology to the general
public. In B.C., patients can now access their own lab results,
on the same day they take the tests. More and more of these
tools, including commercial, health-related apps, are coming.
“Downstream, there will be more opportunities for pa-tients
to access their own information,” Dr. Ho says. “In the
meantime, how do we make the general public aware of these
tools, and how do we make sure we work with them so they
use these tools appropriately?”
Although eHealth has been around for a long time, in
some ways we’re just beginning to understand how to best uti-lize
it. “It’s not only about reading the information, but know-ing
if it’s helpful,” Dr. Ho says. “How do we work with the gen-eral
public in raising eHealth literacy?”
For more information, contact Kendall Ho
at kendall.ho@ubc.ca
had begun before that, in 2003, with the use of videoconfer-encing
for Emergency Health Services.
Of course, we’ve come a long way since then. “Technolo-gies
have come down significantly in costs and gone up in
terms of ease of use,” he says. “The general public has increas-ing
understanding and expectation of using technologies in
health. As technologies have become more mainstream, gov-ernments
and health organizations have been increasingly
adopting eHealth programs. Health professionals are becom-ing
much more adept at using technology.”
innovations magazine Fall/Winter 2014 5
Adaptive User
Interfaces
AS THE INFORMATION SOCIETY HURTLES FORWARD,
COMPUTER USERS OFTEN FEEL OVERWHELMED. COMPUTER
SCIENCE PROFESSOR CRISTINA CONATI IS WORKING TO
MAKE COMPLEX INTERACTIVE SYSTEMS MORE EFFECTIVE
BY ADAPTING THEM TO USER NEEDS.
6 Fall/Winter 2014
In the field of user-adaptive interfaces, there are two
main areas of inquiry, spanning disciplines that
include artificial intelligence, cognitive science,
and human–computer interaction (HCI). In the
first, a user model has to be established, based on the user’s in-teractions
with the system—the machine needs to learn some-thing
about who it is dealing with, using artificial-intelligence
techniques. In the second, the system needs to adapt to the user’s
perceived needs by, for example, changing how the information
is presented, customizing interface functionalities, or providing
explicit suggestions that might help the user. Conati believes that
these user-adaptive interactions must strike a balance between
automated adaptations and those controlled by the user. “You can
see it as a continuum,” she says. “At one extreme, the machine un-derstands
what is going on with the user and is in complete con-trol
of how to adapt to it. At the other, the user does all the work
and has good tools to customize the interface as they like. Neither
approach works perfectly, so we aim for a ‘mixed-initiative inter-action’,
or a balance between the two.”
Adapting Visual Information
to the User
The exponential increase in the amount of widely available digital
information made possible by technological advances has created
a need for new tools to facilitate its interpretation, such as visual
analytics. Conati has been exploring user-adaptive information-visualization
scenarios in some of her recent work, where the
user interprets data that are presented visually. Since different us-ers
have different visualization preferences and abilities, it can be
useful to alter the visualization accordingly. In developing these
visualizations, Conati collaborates closely with fellow UBC Com-puter
Science professor Giuseppe Carenini. UBC cognitive psy-chologist
James Enns has lent his expertise on user study design,
and the selection of user parameters to guide the adaptations—
Conati is currently focusing on perceptual speed, visual working
memory, and verbal working memory, and may consider ad-ditional
features in future work. Through extensive lab studies,
these features are used to develop user models, which help guide
real-time adaptations as the user interacts with a visualization.
Gaze data obtained by an eye tracker, for example, may indicate
that a user has low perceptual speed, since they refer frequently
to additional information—such as a graph’s legend—to interpret
patterns in visual data. In combination with their interactions
with the system, the user can then be “classified” and the visual
information modified accordingly.
While real-time user-adaptive interactions are the ideal, Conati
concedes that again a balance must be struck, in this case between
“making an early prediction based on developed user models in
order to help the user as soon as possible, and acquiring enough
data to make it fairly accurate. The ultimate user model would
know as many things as possible about the user: emotional reac-tions,
perceptual abilities, the task they are working on, and their
expected performance, in order to adapt and personalize the in-teraction.”
Urban Planning Applications
Conati collaborates with Vancouver-based company Metroquest,
which has developed a community-engagement platform for
planning projects that is used by municipalities, as well as archi-tectural
and engineering firms, across North America. The visu-ally
based platform is designed to solicit public input on planning
options, such as regional transportation plans, both online and
in public kiosks. The company is providing system access and
technical support to Conati, to help her develop adaptations that
will keep their target audiences engaged, limiting complication as
necessary and responding to user preferences. As a testbed, Co-nati
is using a customized Metroquest system to canvass student
opinion regarding public transit improvements to UBC.
Potential real-world applications of Conati’s work are growing
daily. “We are moving toward the big data era,” she observes,
“where we are going to be surrounded by data that need to be in-terpreted
by a wide range of individuals. Health-tracking data, for
example, will need to be personalized in such a way that anyone
can benefit from it, whatever their skill level.”
For more information, contact Cristina Conati
at conati@cs.ubc.ca
innovations magazine Fall/Winter 2014 7
Rehabilitation
Robotics for
Stroke Victims
by Shawn Conner
How do you make rehabilitation involving, and
entertaining? “Rehabilitation is really hard
work,” Van der Loos says. “We need to give peo-ple
reasons to keep at it. There’s only so much
you can do intrinsically—we can encourage them with, ‘You
have to get better, stronger.’”
Van der Loos and his team in the RREACH lab, which
is part of the CARIS (Collaborative Advanced Robotics and
Intelligent Systems) lab at UBC, have come up with a novel
solution: the FEATHERS (Functional Engagement in Assisted
Therapy through Exercise Robotics) project, which uses video
games and social media to facilitate and encourage physical
rehabilitation.
By altering the hardware of store-bought controllers and
the software “between the measurement of the position and
movement of the cursor,” Van der Loos and his colleagues
have made it necessary for the player to manipulate two con-trollers
simultaneously, thus ensuring that both arms are em-ployed
equally.
“The software takes bimanual motions and maps those
to the movement of one cursor,” Van der Loos says. “If you
have one weak and one strong arm, and you’re asked to do
bimanual motions, you’re likely to do smaller-range, less pre-cise
motions with your weak side. The cursor gets mapped to
the hand that moves the least to promote making your hands
move in an even and symmetrical way to get the cursor to
move.”
The target beneficiaries of the project are survivors of
stroke and teenagers with cerebral palsy. As of publication,
the project is in the midst of its final testing phase, a six-month
design study.
Another important component of FEATHERS is the so-
Mechanical engineering
professor Machiel Van der
Loos is using robotics to
help people recover from
debilitating impairments.
8 Fall/Winter 2014
cial media aspect. Players are able to log in and share results
not just with their physiotherapists but also other players.
“Having some extrinsic motivators like games and social
media are really powerful tools to keep people engaged,” Van
der Loos says.
To change the shape of the controller itself, FEATHERS
brought in an Emily Carr design team to create a new shell
“more appropriate to the kinds of movements used in physi-cal
therapy. The shape is easier to put into a hand that may be
perennially clenched due to flexors stronger than extensors [a
result of stroke].”
RREACH (which stands for Robotics for Rehabilita-tion,
Exercise and Assessment in Collaborative Healthcare) is
also collaborating with other UBC labs and departments on
FEATHERS, including the Department of Kinesiology and
the Department of Physical Therapy.
“Many people who live with an impairment don’t have a
therapist,” he says. “So they use whatever they can glean off
the Internet, or they depend on however hard their parent or
spouse pushes them to get better. But the Canadian healthcare
system has nothing dialled in for the kinds of repetitions that
are needed to effect some change. You need thousands and
thousands of repetitions. In gaming, you get thousands and
thousands of repetitions.”
Since the benefits reside in using the controllers and not
in the games themselves, anything from popular titles like
Candy Crush Saga and Farmville to more complicated fare,
like first-person-shooter-type games, can be played using the
modified system. “We’re saying, ‘Play whatever you want. Get
a handful of games you like, and include the types of move-ments
that are going to further your rehabilitation so there’s
carryover to your daily tasks.’”
Standing Balance and
Sit-to-Stand Research
Besides FEATHERS, Van der Loos is engaged in a number of
other robotics-and-rehab projects. One is STARR (Standing-balance
Training and Assessment using RISER), which uses a
robot to develop new therapies to achieve standing balance.
That robot, RISER (Robot for Interactive Sensory Engage-ment
and Rehabilitation), measures and can control standing
balance.
“It’s a way of exploring the human sensorimotor systems,
and how all of these different inputs are integrated to perform
what we consider to be really mundane tasks,” Van der Loos
says.
The research can be used to help people who have trouble
with balance, including stroke victims, diabetics, or people
who have other neuropathies. Van der Loos is working with
colleagues from the Department of Physical Therapy, includ-ing
a standing-balance group run by Professor Jayne Garland.
Van der Loos is also involved in sit-to-stand asymmetry
research. Still in its early stages, the sit-to-stand research uses
a one degree-of-freedom robot to assist individuals in the
process of getting up from a chair. The hope is that the results
of the research will aid hemiparetic stroke victims in using
both weak and strong sides to stand from a seated position.
Another area of research is investigating ways for robot-ics
to help people with neurological impairment, often result-ing
from stroke. Such individuals, Van der Loos notes, often
think they are reaching in a straight line when they are not.
“The goal of a lot of robotic therapy is getting people to
see what a straight line is,” he says. “What works best is to ac-centuate
the error by putting up a forcefield perpendicular to
the motion.”
Van der Loos is also overseeing research by his doctoral
student Bulmaro Valdés on physical compensation, specifi-cally
trunk compensation. The idea is to help stroke victims
reduce reliance on the non-damaged parts of the body, again
with the goal of speeding up rehabilitation.
While much of his work is designed to help people re-cover
from debilitating impairment, there is another area of
interest for Van der Loos—a somewhat knottier, even more
philosophical one: robo-ethics.
“For example, when is it appropriate?” Van der Loos asks,
“given the social context, for a robot to enter an elevator? How
does it know? What ends up being the most efficient, or least
error-prone set of conditions for the interaction between hu-man
and robot?”
For Van der Loos, and for robotic research in general, this
is one question that might never entirely be solved. But the
potential benefits of robots in rehabilitation are tangible, and
are just starting to be understood.
For more information, contact Machiel Van der Loos
at vdl@mech.ubc.ca
innovations magazine Fall/Winter 2014 9
state-of-the-art Alpha Lab and assemble a comprehensive team
of graduate students and research staff. He sees his industry
support combined with more traditional academic funding
as making for a healthy research environment. “It’s a marriage
between colleagues within the university and outside,” he says.
“It helps to attract good students, when you have a program
that integrates real company needs with advanced research. We
like getting engaged in research areas that are not an immediate
need for a company but may be implemented within a three- to
ten-year timeframe.”
Technical Challenges
Ordonez’s long-term goal is to establish a modular, flexible, con-trollable,
and scalable conversion architecture for different re-newable
energy systems. A number of complex problems need
to be resolved in realizing these architectures. At the scale of a
condominium building, for example, there may be a need to
extract the maximum power possible from photo-voltaic panels
on the roof or fuel cells in the basement; store it in lithium-ion
batteries using a battery management system tuned for efficient
charging; supply energy to and draw energy from electric vehi-cles
in the parking garage; or perhaps convert the entire build-ing
to DC for more efficient energy consumption and fewer
losses. The system must be kept flexible, modular, and scalable,
so additional components may be added as required, or for use
in other contexts using other energy sources.
Much attention has been paid in recent
years to renewable sources in the
push toward a more sustainable
energy supply. Technologies for efficiently converting power
produced by solar, wind, tidal, fuel cell and other sources into
a usable form, however, remain in their infancy. Electrical and
Computer Engineering professor Martin Ordonez has an ambi-tious
research program underway to address this deficit.
Ordonez, who is the Canada Research Chair in Power Convert-ers
for Renewable Energy Systems, moved to UBC from Simon
Fraser University in 2012 in part because of UBC’s emphasis
on sustainability. The “Campus as a Living Lab” initiative holds
particular appeal for him, as it provides a highly varied test-bed
for validating new power conversion techniques emerging
from his lab. Over the next few years, the initiative will allow
Ordonez to integrate novel technologies into the campus grid,
starting with a building-scale energy system controller for the
UBC Kaiser Building.
One of the reasons there hasn’t been more applied academic
research on renewable power conversion is the prohibitive cost
of outfitting a lab. The suite of necessary equipment, from bat-tery
backup systems to industrial-grade converters, is simply
too expensive for most researchers to purchase and keep up to
date. Ordonez’s strong industry connections, including Alpha
Technologies Ltd., Powertech Labs (BC Hydro), Delta-Q Tech-nologies,
and Corvus Energy have enabled him to equip UBC’s
Power
Conversion for
Renewable
Energy Sources
10 Fall/Winter 2014
Such a building-scale set-up would be beneficial to the utili-ties,
as it would contribute to the “smartness” of the power
grid, without relying on consumers to change their behaviour.
Peak demand periods could be flattened out by drawing on
local building energy sources, thus avoiding brownouts or
the need to switch on additional generators. Also, as Ordonez
points out, “the building would consume energy with very
low harmonic distortion and a good power factor, meaning a
saving for the utility on transmission and distribution losses.
The building could be made a very good citizen in terms of
energy consumption.”
Toward Distributed
Power Generation
Most renewable energy sources are inherently variable, due
to fluctuating winds, tidal currents, solar radiation, etc., while
consumers want instantaneous response. Storage systems are
therefore a necessary component of efficient renewable energy
systems. Also, as Ordonez points out, “Every source has some
benefit, but also poses unique challenges, with very special-ized
power conversion requirements.” Wind generation, for
instance, produces power with variable frequency and ampli-tude,
so it must be converted to DC to be usable.
“I would like to see modular
power conversion that is flex-ible
enough to be plugged in
almost anywhere.”
Extracting, converting, conditioning, storing, and integrating
power from a variety of different sources in real-time so that
it is accessible to users on demand is difficult enough on a
small scale. At the scale of the grid, the challenges become
that much more demanding. Yet Ordonez sees this as the
way of the future, as do his industry partners such as Alpha
Technologies. “We should be able to benefit from small power
generation,” he maintains, “harvesting everywhere, in a dis-tributed,
less centralized fashion. I would like to see modular
power conversion that is flexible enough to be plugged in al-most
anywhere.”
With less than 0.9% of worldwide energy consumption in 2011
derived from renewable sources, Ordonez’s flexible, modular
approach may be the way to go.
For more information, contact Martin Ordonez at
mordonez@ece.ubc.ca
innovations magazine Fall/Winter 2014 11
High dynamic range (HDR) video content is
currently attracting the attention of interna-tional
standardization bodies such as the Mo-tion
Picture Experts Group (MPEG), as tech-nologies
to capture and display HDR mature. The uncannily
lifelike appearance of HDR content—its wow factor—results
from its approximation of the range of contrast and colours,
or dynamic range, we can perceive in the real world. The
human visual system at the limits of perception can accom-modate
a dynamic range of ten orders of magnitude (1010),
with a functional range of about six orders of magnitude.
Conventional low dynamic range (LDR) displays. Including
current TVs and computer monitors, however, have a range
of roughly two orders of magnitude. It’s a limitation we’ve
come to accept as the norm, and so when we are exposed to
the level of detail that HDR content reveals to us, it seems
eerily lifelike, akin to the difference between black and white
and colour.
High dynamic range will be a familiar term for many smart-phone
users, as it is often an available camera option. When
selected, two images with different exposure times are taken,
and then merged to create a significantly higher range of
contrast than is available in a single image. Capturing HDR
video content is, however, much more difficult, since two
video streams at different exposure levels must be seamlessly
merged to create the desired effect. High-end commercially
available cameras, such as the RED camera, are capable of
capturing HDR video content, but are very expensive and
beyond the reach of the average consumer. HDR displays
are still in the prototype stage.
Investigators in the ICICS Digital Multimedia Lab are ad-dressing
the many technological hurdles that must be over-come
in order to deliver high dynamic range video content
to consumers, from capturing, processing, and compress-ing/
transmitting, to displaying. There is a general effort
internationally to develop techniques for displaying HDR
video on existing low dynamic range displays (e.g., TVs and
computer monitors). This involves selecting those compo-nents
from the HDR video that can best be reproduced on
a conventional display, so that the viewer benefits from an
HDR-like rich viewing experience. High dynamic range im-ages
prepared for printing (as seen in the photo on page 12)
roughly approximate the appearance (and advantages) of
HDR content shown on an LDR display using these tech-niques.
It is envisioned that the necessary processing would hap-pen
in a set-top box, or perhaps a chip in new televisions
or displays. If these technologies prove sound, we may be
viewing a version of high dynamic range content on our TV
sets within the next five years.
For more information, contact Mahsa Pourazad at
pourazad@icics.ubc.ca
High Dynamic
Range Video
Delivering the “Wow” Factor
12 Fall/Winter 2014
LDR image HDR image
“It’s a limitation we’ve come to accept as the norm, and so when
we are exposed to the level of detail that HDR content reveals to
us, it seems eerily lifelike, akin to the difference between black
and white and colour.”
innovations magazine Fall/Winter 2014 13
Sports Player
Tracking,
Identification,
and Action
Recognition
14 Fall/Winter 2014
In the 1970s, Toronto Maple Leafs head
coach Roger Neilson became known as
“Captain Video” for his extensive use of
videotape as a coaching tool. Techno-logical
advances since then have made
video analysis less labour intensive, with major contribu-tions
from computer vision expert Jim Little and his col-leagues
in UBC’s Laboratory for Computational Intelligence
(LCI), in automated player tracking, action recognition, and
identification. Once fully implemented, advances such as
these will make it possible to search and select video seg-ments
featuring aspects of the game that coaches may want
to emphasize in a practice. Video assistants would no longer
have to pore over footage following a game to manually or-ganize
it.
Automated tracking and identification of sports players, as
well as recognizing their actions, presents unique research
problems. Tracking pedestrians in a mall, for example, is
less difficult since their movements are more predictable.
The monitoring cameras are usually
stationary, making it easier to isolate
moving persons in the foreground.
Sports videos, however, are mostly
captured by panning, zooming cam-eras,
making it harder to distinguish
them from a background that also ap-pears
to be moving. In addition, players are often bunched
up (as in hockey), their faces obscured, their jersey numbers
hidden, and their movements very rapid, causing blurring
in the video.
Semi-supervised Learning
Sports does, however, have a “narrow semantic space,” as
Little points out. Players’ activities are largely confined to
those you might expect in a game. Also, teams have a given
player line-up that remains fairly consistent throughout a
season, making it possible to prepare training data offline.
Little does so by having his graduate students draw “bound-ing
boxes” around players and tagging them with the player’s
name in a training video, which becomes the template for
tracking the players in a test video. Since the players’ shapes
change throughout the game according to their actions, the
templates must be continually updated, using machine-learning
techniques. Thousands of player features, such as
small patches of texture and colour, are automatically gener-ated
as the player is tracked. When combined and related to
the offline tagging, they form a classifier that enables auto-mated
identification of players regardless of whether or not
their faces or jersey numbers can be seen.
This combination of manual and automated learning is
known in the field of artificial intelligence as “semi-super-vised
learning”. “Training data is one of the key problems in
model learning,” Little says. “We start with a small nugget
of labelled data and see how far we can extend it.” Little in-corporates
play-by-play information to narrow the variable
space, so the system knows which players are on the ice/
court/field at a given time.
Estimating Pose
Currently, Little is focusing on pose estimation as a means
of understanding a player’s actions. Many state-of-the-art
systems identify pose in individual frames, by fitting a skele-ton
to the player. Little is incorporat-ing
movement by adding pose infor-mation
from subsequent frames to
confirm the initial pose prediction.
His work in this area is also finding
application in other areas, including
the ICICS People and Planet Friendly
Home sustainability initiative (see Innovations, Fall/Winter
2012). Little is looking at monitoring the pose of occupants
in a non-intrusive way (where the video is not stored), to
detect anomalies such as falls. This work is complicated by
the fact that the home is a far more varied environment
than the sports arena, with a much wider range of activities.
The setting must be taken into account when assessing the
meaning of an individual’s pose, as well as their health pro-file.
A certain pose would be expected, for example, when a
person is loading a dishwasher, but an alert should be trig-gered
if they appear to fall while doing so, particularly if
they are elderly.
Other applications of Little’s work include mobile robotics,
where the robot needs to understand its environment and
the actions of nearby individuals in order to navigate safely.
For more information, contact Jim Little at
little@cs.ubc.ca
“Training data is one
of the key problems in
model learning.”
innovations magazine Fall/Winter 2014 15
MAGIC:
25 Years Later
by Siobhán McElduff
16 Fall/Winter 2014
2015 will see the twenty-fifth anniver-sary
of UBC’s Media and Graphics
Interdisciplinary Centre (MAG-IC).
Founded in 1990 under the directorship of Computer
Science professor Kellogg Booth, its debut at UBC came one
year before CERN (the European Organization for Nuclear
Research) hosted the world’s first website, and two years be-fore
the first photograph was posted to the Internet (also by
CERN). MAGIC’s mission was—and is—to foster research
in media technology in support of economic development
in British Columbia, and it has adhered to this even as what
is meant by media technology is no longer recognizable as
the entity it was in 1990.
What has made and makes MAGIC special is its interdis-ciplinary,
innovative approach, which brings together re-searchers
from computer science, engineering, medicine,
music, psychology, education, archival science, history, ar-chaeology,
and more—in other words, from the entire gamut
of UBC’s faculties and departments. (Recently, MAGIC has
been under the interim directorship of Siobhán McElduff,
Associate Professor in Classical, Near Eastern and Religious
Studies. MAGIC’s Associate Director is Brian Fisher, Associ-ate
Professor in the School of Interactive Arts and Technol-ogy,
and in the Cognitive Science Program, both at Simon
Fraser University—a testament to MAGIC’s diverse interests
and reach.)
Current projects reflect MAGIC’s interdisciplinary and in-novative
ideals: the Canadian Environmental Health Atlas;
the BC Child Injury Research and Prevention Network and
a range of other visual analytics projects under the supervi-sion
of Brian Fisher; the first wearable technologies group at
UBC, created by Junia Anacleto, Visiting Peter Wall Scholar
and Associate Professor in Human–Computer Interaction
at the University of San Carlo, Brazil; the laptop orchestra;
an augmented reality app for a Bronze Age site in Cyprus
in partnership with NGRAIN Corp., Vancouver; a multi-lingual
tool for creating and assessing philosophical argu-ments;
the Global Legal Identity Watch; and the Digital Sa-lon,
a group of researchers in the Arts who use and develop
computational tools.
Additionally, as part of its mandate to encourage innovation
at UBC, MAGIC recently launched the I3 (idea, innovation
and inaugurate) Challenge, in conjunction with Mark Salo-pek
of the GRAND (Graphics, Animation and New Media)
national research network. This challenge, aimed at gradu-ate
students, is now in its second round, with members of
the first intake already in discussions with investors about
their projects.
For more information, contact Siobhán McElduff at
siobhan.mcelduff@ubc.ca
innovations magazine Fall/Winter 2014 17
President and CEO Nima Ziraknejad, who founded the
company in 2009, believes that consumers could benefit
more from sensing technologies, particularly machine vi-sion,
than they have. As an industrial partner in a research
project sponsored by the AUTO21 Network of Centres of
Excellence, NZTech is working to commercialize an adaptive
and automatic head restraint, or headrest, system that helps
prevent whiplash injuries. Ziraknejad was heavily involved
in the underlying research during his PhD studies, develop-ing
a hybrid sensor array that adapts the head restraint posi-tion
to the location and orientation of the driver’s head, in
accordance with guidelines from the Insurance Institute for
Highway Safety.
Whiplash is a common, debilitating occurrence in ve-hicular
accidents, most often occurring as a result of rear or
side impacts where the occupant’s head is thrown backwards
and sideways, causing soft tissue damage in the neck or back.
Whiplash is extremely painful, and can take months or even
NZ TECHNOLOGIES INC. (NZTECH) IS AN ICICS-INCUBATED START-UP COMPANY THAT DEVELOPS
INTELLIGENT DEVICES FOR SAFETY AND INTUITIVE HUMAN-MACHINE INTERACTIONS IN THE AU-TOMOTIVE,
RESIDENTIAL, AND CLINICAL SECTORS.
Bringing Machine
Vision to the
Consumer
18 Fall/Winter 2014
years to heal, and sometimes never does. Head restraints
can help mitigate the resulting injury, but only if correctly
positioned.
A self-contained, motorized head restraint prototype
system has been developed, with a customized capacitive
sensor array integrated in the
front of the device. Using electric
field sensing, the array measures
the distance between the back of
the occupant’s head and the front
of the restraint, and a 3D time-of-
flight camera installed below
the rear-view mirror determines
the orientation of the occupant’s
head. This latter information is
important, since whiplash severity is greater when the vehi-cle
occupant’s head is turned—the restraint needs to be that
much closer to limit the injury.
“With the fused sensor data,” Ziraknejad says, “we can
control motors embedded in the head restraint to raise
and lower it, and move it forward and backward.” Ongoing
research will determine the optimal position for mitigat-ing
whiplash injuries, given the occupant’s head position
and orientation. Appropriate intervals for re-positioning
the headrest in the course of normal driving will be deter-mined
as well. Knowledge of the driver’s head orientation
also lends itself to possible driver distraction/fatigue detec-tion
applications.
Machine Vision at Your Fingertips
On the residential side, NZTech is part of a six-member
industrial consortium led by TELUS
that is supporting the ICICS “People and Plan-et
Friendly Home” sustainability initiative (see
Innovations, Fall/Winter 2012). In this project, 25 research-ers
from 7 different UBC departments are developing novel
integrated technologies for residential spaces that promote
both sustainability and quality of life.
In the research strand supported by NZTech, machine
vision and cloud computing are being harnessed to assist
in cooking and baking. The researchers and NZTech have
developed a small, portable device equipped with 3D cam-eras,
a thermometer, and an ARM processor-based embed-ded
system, that attaches to the outside of the window in
the oven door. As an item is cooking, it selectively senses for
designated states, such as a certain colour, that will indicate
that the item is done. This information is correlated with
the item’s internal temperature determined by a cooking
thermometer, and an alert together with live video frames of
the food are sent to the resident on a mobile device.
Selective sensing does not capture the entire object of in-terest,
therefore requires less bandwidth to transmit the live
images. In addition, less onboard
preprocessing is required, reduc-ing
power consumption. Addi-tional
processing, if necessary, is
done on a cloud-based server.
“There is a trade-off,”
Ziraknejad states. “You might be
able to do all of the processing
onboard, but at the cost of higher
power consumption. By leaving
the heavy processing for the cloud, we can also keep the sys-tem
working in real-time, which is critical.”
Real-time, portable machine vision and processing for
a variety of tasks in the home is NZTech’s ultimate goal for
the device, having developed a proof-of-concept prototype
for the oven application. The company is now working on a
pan-and-tilt mount, so the device can be placed anywhere
in the home and targeted as required. They are also inves-tigating
the feasibility of wireless power transmission to the
device.
“We hope to have a suite of image-processing modules
for different applications,” Ziraknejad says, “including drug
dosage reminders and fall detection. We want to provide
machine vision at your fingertips.” He means this last point
both figuratively and literally. The company is developing
a gesture and voice-based interface, associated with a light
projector that can project results on any surface. Internet-based
information such as weather or traffic patterns could
also be requested by a voice command, then projected on a
designated surface with a certain gesture.
“We are aiming at an affordable, general platform that
can do many things for you,” Ziraknejad states. With a work-ing
prototype in hand and ICICS research heft behind him,
NZ Technologies is well on the way.
For more information,
contact NZ Technologies at info@nztech.ca
“We can control motors
embedded in the head
restraint to raise and
lower it, and move it
forward and backward.”
innovations magazine Fall/Winter 2014 19
UBC’s Institute for Computing, Information and Cognitive Systems (ICICS) is an
umbrella organization that promotes collaboration among researchers from the faculties of
Applied Science, Arts, Commerce, Education, Forestry, Medicine, and Science, and with
industry. ICICS facilitates the collaborative multidisciplinary research of approximately
200 faculty members and 1,000 graduate students in these faculties.
Our members attract approximately $18 million annually in grants and contracts. Their
work strengthens Canada’s strategic Science and Technology research priority areas,
benefiting all Canadians.
PUBLICATIONS MAIL AGREEMENT NO. 40049168
RETURN UNDELIVERABLE CANADIAN ADDRESSES TO:
ICICS, University of British Columbia
289-2366 Main Mall
Vancouver, BC V6T 1Z4
info@icics.ubc.ca
Client TELUS TEL255_InnMag_8.5x7.67 Created Nov. 28, 2014
Account Jennifer Creative Joe Mac Artist Jay Producer Claire
Trim 8.5” x 7.67” | Bleed: 8.75” x 7.92” | Live: 8.25” x 7.5” Insertion Date(s) Dec. 1, 2014
Colours CYANI MAGENTAI YELLOWI BLACKI Ad Number TEL225_InnMag_8_5x7_67
Publications Innovations Magazine
Info Final fi le is PDFX1A
All colours are printed as process match unless indicated otherwise. Please check before use. In spite of our careful checking, errors infrequently occur and we request that you check this proof for accuracy. The&Partnership’s liability is limited
to replacing or correcting the disc from which this proof was generated. We cannot be responsible for your time, fi lm, proofs, stock, or printing loss due to error.
APPROVAL
Creative Team Proofreader Producer Studio Client/Account Manager
TELUS, the TELUS logo, the future is friendly and telus.com are trademarks of TELUS Corporation, used under licence. All other trademarks are the property of their respective owners. © 2014 TELUS.
Want to get the most out of your
smartphone or tablet? Book a free
one-to-one learning session.
It’s in our nature to care.
We care about
your questions big and small.
Visit telus.com/learningcentre
to learn more.
TEL255_InnMag_8.5x7.67.indd 1 12/4/14 11:37 AM
www.icics.ubc.ca ICICS
CONNECTING KNOWLEDGE
magazine
111 l*Vt- \ iWiV
Fall/Winter 2014
eHEALTH
LEVERAGING
14:59^
PROMOTE H
Also
REHABILITATION ROBOTICS
POWER CONVERSION FOR RENEWABLE ENERGfl SYSTEMS
HIGH DYNAMIC RANGE VIDEO
SPORTS PLAYER TRACKING
and more...
|UBC|      a place of mini
THE UNIVERSITY OF BRITISH COLUMBIA 6-7
ADAPTIVE USER INTERFACES
Customizations according to
user needs
4-5
e HEALTH
Leveraging new technologies to promote
health
innovations
fall/winter 2014
Production
Editor
Writer
Design
Office
Email
Sharon Cavalier
ICICS Administrator
Craig Wilson
ICICS Communication Writer
Industry Design
www.industrydesign.ca
ICICS
University of British Columbia
289-2366 Main Mall
Vancouver, BC, Canada V6T 1Z4
Tel: 604-822-6894
Fax:604-822-9013
info@icics.ubc.ca
8-9
REHABILITATION ROBOTICS
FOR STROKE VICTIMS
Robotic therapies that speed recovery
director's desk
2    Fall/Winter 2014
18-19
BRINGING MACHINE VISION
TO THE CONSUMER
ICICS start-up company
develops devices for consumer
safety and convenience 10-11
POWER CONVERSION FOR RENEWABLE
ENERGY SYSTEMS
Providing the missing pieces
12-13
HIGH DYNAMIC RANGE VIDEO
Delivering the "wow" factor
Welcome to the latest edition of ICICS Innovations
magazine. I would like to take this opportunity to
introduce you to a new model for collaborating with
industry developed by ICICS that takes advantage of
our unique pan-university status.
Under this model, a blanket intellectual-property agreement is negotiated
between the company and UBC, freeing company managers of the need to
negotiate one-on-one agreements with individual researchers. ICICS works
with the company to define one or more broad-based research projects
in support of their R&D goals. We then assemble research teams from
our 160-strong cross-faculty membership, pursue additional government
funding, and manage the projects. Tangible results are realized for companies,
with direct benefits for their innovation strategies. Researchers and students
gain support, industry exposure, and perspective.
This new approach has led to the ICICS/TELUS People and Planet
Friendly Home sustainability initiative, which we profiled in the Fall/
Winter 2012 issue. This broad-based endeavor involves 25 professors and
37 students from 4 different faculties, and is supported by a 6-member
industry consortium led by TELUS. The project continues to grow, attracting
additional industry and government funding. A site has been selected in
the central campus to build a prototype home and broaden the scope of the
integrated research efforts.
We look forward to additional meaningful collaborations with industry
emerging from this new collaboration model. In the meantime, I hope you
enjoy the sampling of UBC research achievements we provide here.
Panos Nasiopoulos, ICICS Director
14-15
SPORTS PLAYER TRACKING
Advanced computer vision
techniques for tracking
sports players
16-17
MAGIC: 25 Years Later
Human-computer interaction and
graphics centre at ICICS
innovations magazine
Fall/Winter 2014    3 eHEALTH:
LEVERAGING NEW
TECHNOLOGIES TO
PROMOTE HEALTH
BY SHAWN CONNER
WITH THE FAST PACE OF TECHNOLOGY DEVELOPMENTS IN THE
AREA OF HEALTH CARE SERVICES DELIVERY,HOW CAN
STUDENTS, PROFESSIONALS AND PATIENTS KEEP UP?
This is one of a number of questions facing Dr.
Kendall Ho. A professor in the Department of
Emergency Medicine, Ho is also a Director of the
eHealth Strategy Office in the Faculty of Medicine
at UBC. As Director, Dr. Ho leads an interdisciplinary team of
researchers conducting research into eHealth and technology-
enabled knowledge translation. They are focusing on how to
best harness modern technologies to enhance health services
and accelerate the incorporation of the latest health evidence
into routine practices.
"We are very interested in understanding how modern
information technologies, like mobile phones, tablets and
computers, can be used optimally for healthy pursuits," Dr.
Ho says. "The motto of the strategy office is 'Health Interconnected: Connecting People to Better Health.' We use the word
'better' as a verb."
The research helps policy-makers like the B.C. Ministry
of Health's Patients as Partners branch develop strategies to
engage the general and multicultural public in using technology for their own health. "And then we carry out education for
health professionals, including medical students," Dr. Ho says.
The eHealth Strategy Office opened in September 2008. At
that time, the Dean of UBC's Faculty of Medicine appointed
Dr. Ho as Founding Director. But Dr. Ho's work in eHealth
4    Fall/Winter 2014 had begun before that, in 2003, with the use of videoconferencing for Emergency Health Services.
Of course, we've come a long way since then. "Technologies have come down significantly in costs and gone up in
terms of ease of use," he says. "The general public has increasing understanding and expectation of using technologies in
health. As technologies have become more mainstream, governments and health organizations have been increasingly
adopting eHealth programs. Health professionals are becoming much more adept at using technology."
eHealth Advantages
Using technology to promote health offers advantages across
the sector. For health care professionals, technology offers the
opportunity to measure, quantify, analyze, and disseminate research and knowledge in unprecedented numbers and specifics. Meanwhile, health consumers are finding useful information to support their own health, and have become better able
to connect both with other patients with the same diseases,
and with health professionals. And government can use solid
health data in terms of understanding disease patterns and
health trends and where to put its resources. As an example,
Dr. Ho cites the rising number of people with diabetes, and
the subsequent rise in costs of care. "How do you dedicate resources appropriately to the type of services needed?" he asks.
Knowledge from eHealth initiatives can help policy-makers
make these types of decisions.
Technology also offers challenges, however. "Is the information accurate?" Dr. Ho says. "How do you deal with inaccurate information? How do you make sure the information
one obtains is appropriate for that particular individual? How
do we, in this new environment, form this health professional-
patient relationship, so that we can support each other?"
Diabetes Management
Currently, Dr. Ho is spearheading several eHealth-related
projects that look at these and other eHealth challenges. One
project, on diabetes, is supported by the Lawson Foundation,
a Canadian philanthropic organization. The project helps people with diabetes and caregivers manage the condition using
sensors, social media and text messaging. The project's participants use the sensors to measure their weight, blood pressure,
and glucose level. Social media links them to their fellow patients and family members to discuss the best ways to manage
the disease. Supportive tips are provided by health professionals via text messaging.
This research has found that patients who follow the
program are better able to control their blood pressure, and
improve their A1C levels, an indication of how well they are
able to control their glucose levels over time. "They actually
have some desirable weight reduction," Dr. Ho says. "They
found that the system is very supportive to their health. At the
eHealth Strategy Office, we don't invent the sensors ourselves,
we don't invent social media, but we apply these different
kinds of technology in the health context to support patients."
Engaging Communities
Dr. Ho is also working with the Ministry of Health's Patients as
Partners initiative on two programs. One is the Intercultural
Online Health Network (ICON). It uses multimedia channels
such as live forums, Web 2.0, YouTube, and social media to
engage the Chinese and South Asian general public, and to
present material in their native language to support them in
chronic disease management. "We found that seven out often
people who attend one or more of our different media channels want to make a big change in their lives to improve their
health," Dr. Ho says. "Three out of four people want to make
at least a small change to live a healthier life." Another sign of
success: some temples in South Asian communities have begun changing their menus to reflect diabetes-related concerns.
Another collaboration with the Ministry of Health looks
at methods to introduce eHealth technology to the general
public. In B.C., patients can now access their own lab results,
on the same day they take the tests. More and more of these
tools, including commercial, health-related apps, are coming.
"Downstream, there will be more opportunities for patients to access their own information," Dr. Ho says. "In the
meantime, how do we make the general public aware of these
tools, and how do we make sure we work with them so they
use these tools appropriately?"
Although eHealth has been around for a long time, in
some ways we're just beginning to understand how to best utilize it. "It's not only about reading the information, but knowing if it's helpful," Dr. Ho says. "How do we work with the general public in raising eHealth literacy?"
For more information, contact Kendall Ho
at kendall.ho@ubc.ca
innovations magazine
Fall/Winter 2014    5 AS THE INFORMATION USOCIETY HURTLES FORWARD,
COMPUTER USERS OFTEN FEEL OVERWHELMED. COMPUTER
SCIENCE PROFESSOR CRISTINA CONATI IS WORKING TO
MAKE COMPLEX INTERACTIVE SYSTEMS MORE EFFECTIVE
BY ADAPTING THEM TO USER NEEDS.
Winter 2014 the field of user-adaptive interfaces, there are two
I -^^^    main areas of inquiry, spanning disciplines that
include artificial intelligence, cognitive science,
.^LA» ^L and human-computer interaction (HCI). In the
first, a user model has to be established, based on the user's interactions with the system—the machine needs to learn something about who it is dealing with, using artificial-intelligence
techniques. In the second, the system needs to adapt to the user's
perceived needs by, for example, changing how the information
is presented, customizing interface functionalities, or providing
explicit suggestions that might help the user. Conati believes that
these user-adaptive interactions must strike a balance between
automated adaptations and those controlled by the user. "You can
see it as a continuum," she says. "At one extreme, the machine understands what is going on with the user and is in complete control of how to adapt to it. At the other, the user does all the work
and has good tools to customize the interface as they like. Neither
approach works perfectly, so we aim for a 'mixed-initiative interaction, or a balance between the two."
Adapting Visual Information
to the User
The exponential increase in the amount of widely available digital
information made possible by technological advances has created
a need for new tools to facilitate its interpretation, such as visual
analytics. Conati has been exploring user-adaptive information-
visualization scenarios in some of her recent work, where the
user interprets data that are presented visually. Since different users have different visualization preferences and abilities, it can be
useful to alter the visualization accordingly. In developing these
visualizations, Conati collaborates closely with fellow UBC Computer Science professor Giuseppe Carenini. UBC cognitive psychologist James Enns has lent his expertise on user study design,
and the selection of user parameters to guide the adaptations—
Conati is currently focusing on perceptual speed, visual working
memory, and verbal working memory, and may consider additional features in future work. Through extensive lab studies,
these features are used to develop user models, which help guide
real-time adaptations as the user interacts with a visualization.
Gaze data obtained by an eye tracker, for example, may indicate
that a user has low perceptual speed, since they refer frequently
to additional information—such as a graph's legend—to interpret
patterns in visual data. In combination with their interactions
with the system, the user can then be "classified" and the visual
information modified accordingly.
While real-time user-adaptive interactions are the ideal, Conati
concedes that again a balance must be struck, in this case between
"making an early prediction based on developed user models in
order to help the user as soon as possible, and acquiring enough
data to make it fairly accurate. The ultimate user model would
know as many things as possible about the user: emotional reactions, perceptual abilities, the task they are working on, and their
expected performance, in order to adapt and personalize the interaction."
Urban Planning Applications
Conati collaborates with Vancouver-based company Metroquest,
which has developed a community-engagement platform for
planning projects that is used by municipalities, as well as architectural and engineering firms, across North America. The visually based platform is designed to solicit public input on planning
options, such as regional transportation plans, both online and
in public kiosks. The company is providing system access and
technical support to Conati, to help her develop adaptations that
will keep their target audiences engaged, limiting complication as
necessary and responding to user preferences. As a testbed, Conati is using a customized Metroquest system to canvass student
opinion regarding public transit improvements to UBC.
Potential real-world applications of Conati's work are growing
daily. "We are moving toward the big data era," she observes,
"where we are going to be surrounded by data that need to be interpreted by a wide range of individuals. Health-tracking data, for
example, will need to be personalized in such a way that anyone
can benefit from it, whatever their skill level."
For more information, contact Cristina Conati
at conati@cs.ubc.ca
innovations magazine
Fall/Winter 2014    7 Rehabilitation
Robotics for
Stroke Victims
i
by Shawn Conner
Mechanical engineering
professor Machiel Van der
Loos is using robotics to
help people recover from
debilitating impairments.
How do you make rehabilitation involving, and
entertaining? "Rehabilitation is really hard
work," Van der Loos says. "We need to give people reasons to keep at it. There's only so much
you can do intrinsically—we can encourage them with, 'You
have to get better, stronger.'"
Van der Loos and his team in the RREACH lab, which
is part of the CARIS (Collaborative Advanced Robotics and
Intelligent Systems) lab at UBC, have come up with a novel
solution: the FEATHERS (Functional Engagement in Assisted
Therapy through Exercise Robotics) project, which uses video
games and social media to facilitate and encourage physical
rehabilitation.
By altering the hardware of store-bought controllers and
the software "between the measurement of the position and
movement of the cursor," Van der Loos and his colleagues
have made it necessary for the player to manipulate two controllers simultaneously, thus ensuring that both arms are employed equally.
"The software takes bimanual motions and maps those
to the movement of one cursor," Van der Loos says. "If you
have one weak and one strong arm, and you're asked to do
bimanual motions, you're likely to do smaller-range, less precise motions with your weak side. The cursor gets mapped to
the hand that moves the least to promote making your hands
move in an even and symmetrical way to get the cursor to
move."
The target beneficiaries of the project are survivors of
stroke and teenagers with cerebral palsy. As of publication,
the project is in the midst of its final testing phase, a six-
month design study.
Another important component of FEATHERS is the so-
8    Fall/Winter 2014 cial media aspect. Players are able to log in and share results
not just with their physiotherapists but also other players.
"Having some extrinsic motivators like games and social
media are really powerful tools to keep people engaged," Van
der Loos says.
To change the shape of the controller itself, FEATHERS
brought in an Emily Carr design team to create a new shell
"more appropriate to the kinds of movements used in physical therapy. The shape is easier to put into a hand that may be
perennially clenched due to flexors stronger than extensors [a
result of stroke]."
RREACH (which stands for Robotics for Rehabilitation, Exercise and Assessment in Collaborative Healthcare) is
also collaborating with other UBC labs and departments on
FEATHERS, including the Department of Kinesiology and
the Department of Physical Therapy.
"Many people who live with an impairment don't have a
therapist," he says. "So they use whatever they can glean off
the Internet, or they depend on however hard their parent or
spouse pushes them to get better. But the Canadian healthcare
system has nothing dialled in for the kinds of repetitions that
are needed to effect some change. You need thousands and
thousands of repetitions. In gaming, you get thousands and
thousands of repetitions."
Since the benefits reside in using the controllers and not
in the games themselves, anything from popular titles like
Candy Crush Saga and Farmville to more complicated fare,
like first-person-shooter-type games, can be played using the
modified system. "We're saying, 'Play whatever you want. Get
a handful of games you like, and include the types of movements that are going to further your rehabilitation so there's
carryover to your daily tasks'"
Standing Balance and
Sit-to-Stand Research
Besides FEATHERS, Van der Loos is engaged in a number of
other robotics-and-rehab projects. One is STARR (Standing-
balance Training and Assessment using RISER), which uses a
robot to develop new therapies to achieve standing balance.
That robot, RISER (Robot for Interactive Sensory Engagement and Rehabilitation), measures and can control standing
balance.
"It's a way of exploring the human sensorimotor systems,
and how all of these different inputs are integrated to perform
what we consider to be really mundane tasks," Van der Loos
says.
The research can be used to help people who have trouble
with balance, including stroke victims, diabetics, or people
who have other neuropathies. Van der Loos is working with
colleagues from the Department of Physical Therapy, including a standing-balance group run by Professor Jayne Garland.
Van der Loos is also involved in sit-to-stand asymmetry
research. Still in its early stages, the sit-to-stand research uses
a one degree-of-freedom robot to assist individuals in the
process of getting up from a chair. The hope is that the results
of the research will aid hemiparetic stroke victims in using
both weak and strong sides to stand from a seated position.
Another area of research is investigating ways for robotics to help people with neurological impairment, often resulting from stroke. Such individuals, Van der Loos notes, often
think they are reaching in a straight line when they are not.
"The goal of a lot of robotic therapy is getting people to
see what a straight line is," he says. "What works best is to accentuate the error by putting up a forcefield perpendicular to
the motion."
Van der Loos is also overseeing research by his doctoral
student Bulmaro Valdes on physical compensation, specifically trunk compensation. The idea is to help stroke victims
reduce reliance on the non-damaged parts of the body, again
with the goal of speeding up rehabilitation.
While much of his work is designed to help people recover from debilitating impairment, there is another area of
interest for Van der Loos—a somewhat knottier, even more
philosophical one: robo-ethics.
"For example, when is it appropriate?" Van der Loos asks,
"given the social context, for a robot to enter an elevator? How
does it know? What ends up being the most efficient, or least
error-prone set of conditions for the interaction between human and robot?"
For Van der Loos, and for robotic research in general, this
is one question that might never entirely be solved. But the
potential benefits of robots in rehabilitation are tangible, and
are just starting to be understood.
For more information, contact Machiel Van der Loos
at vdl@mech.ubc.ca
innovations magazine
Fall/Winter 2014    9 Much:
attention has been paid in recent
years to renewable sources in the
.push toward a more sustainable
energy supply. Technologies for efficiently converting power
produced by solar, wind, tidal, fuel cell and other sources into
a usable form, however, remain in their infancy. Electrical and
Computer Engineering professor Martin Ordonez has an ambitious research program underway to address this deficit.
Ordonez, who is the Canada Research Chair in Power Converters for Renewable Energy Systems, moved to UBC from Simon
Fraser University in 2012 in part because of UBC's emphasis
on sustainability. The "Campus as a Living Lab" initiative holds
particular appeal for him, as it provides a highly varied test-
bed for validating new power conversion techniques emerging
from his lab. Over the next few years, the initiative will allow
Ordonez to integrate novel technologies into the campus grid,
starting with a building-scale energy system controller for the
UBC Kaiser Building.
One of the reasons there hasn't been more applied academic
research on renewable power conversion is the prohibitive cost
of outfitting a lab. The suite of necessary equipment, from battery backup systems to industrial-grade converters, is simply
too expensive for most researchers to purchase and keep up to
date. Ordonez's strong industry connections, including Alpha
Technologies Ltd., Powertech Labs (BC Hydro), Delta-Q Technologies, and Corvus Energy have enabled him to equip UBC's
state-of-the-art Alpha Lab and assemble a comprehensive team
of graduate students and research staff. He sees his industry
support combined with more traditional academic funding
as making for a healthy research environment. "It's a marriage
between colleagues within the university and outside," he says.
"It helps to attract good students, when you have a program
that integrates real company needs with advanced research. We
like getting engaged in research areas that are not an immediate
need for a company but may be implemented within a three- to
ten-year timeframe."
Technical Challenges
Ordonez's long-term goal is to establish a modular, flexible, controllable, and scalable conversion architecture for different renewable energy systems. A number of complex problems need
to be resolved in realizing these architectures. At the scale of a
condominium building, for example, there may be a need to
extract the maximum power possible from photo-voltaic panels
on the roof or fuel cells in the basement; store it in lithium-ion
batteries using a battery management system tuned for efficient
charging; supply energy to and draw energy from electric vehicles in the parking garage; or perhaps convert the entire building to DC for more efficient energy consumption and fewer
losses. The system must be kept flexible, modular, and scalable,
so additional components may be added as required, or for use
in other contexts using other energy sources.
10    Fall/Winter 2014 , -    .      >.
Such a building-scale set-up would be beneficial to the utilities, as it would contribute to the "smartness" of the power
grid, without relying on consumers to change their behaviour.
Peak demand periods could be flattened out by drawing on
local building energy sources, thus avoiding brownouts or
the need to switch on additional generators. Also, as Ordonez
points out, "the building would consume energy with very
low harmonic distortion and a good power factor, meaning a
saving for the utility on transmission and distribution losses.
The building could be made a very good citizen in terms of
energy consumption."
Toward Distributed
Power Generation
Most renewable energy sources are inherently variable, due
to fluctuating winds, tidal currents, solar radiation, etc., while
consumers want instantaneous response. Storage systems are
therefore a necessary component of efficient renewable energy
systems. Also, as Ordonez points out, "Every source has some
benefit, but also poses unique challenges, with very specialized power conversion requirements." Wind generation, for
instance, produces power with variable frequency and amplitude, so it must be converted to DC to be usable.
"I would like to see modular
power conversion that is flexible enough to be plugged in
almost anywhere."
Extracting, converting, conditioning, storing, and integrating
power from a variety of different sources in real-time so that
it is accessible to users on demand is difficult enough on a
small scale. At the scale of the grid, the challenges become
that much more demanding. Yet Ordonez sees this as the
way of the future, as do his industry partners such as Alpha
Technologies. "We should be able to benefit from small power
generation," he maintains, "harvesting everywhere, in a distributed, less centralized fashion. I would like to see modular
power conversion that is flexible enough to be plugged in almost anywhere."
With less than 0.9% of worldwide energy consumption in 2011
derived from renewable sources, Ordonez's flexible, modular
approach may be the way to go.
For more information, contact Martin Ordonez at
mordonez@ece.ubc.ca
innovations magazine
Fall/Winter 2014   11 High Dynamic
Range Video
Delivering the "Wow" Factor
High dynamic range (HDR) video content is
currently attracting the attention of international standardization bodies such as the Motion Picture Experts Group (MPEG), as technologies to capture and display HDR mature. The uncannily
lifelike appearance of HDR content—its wow factor—results
from its approximation of the range of contrast and colours,
or dynamic range, we can perceive in the real world. The
human visual system at the limits of perception can accommodate a dynamic range often orders of magnitude (1010),
with a functional range of about six orders of magnitude.
Conventional low dynamic range (LDR) displays. Including
current TVs and computer monitors, however, have a range
of roughly two orders of magnitude. It's a limitation we've
come to accept as the norm, and so when we are exposed to
the level of detail that HDR content reveals to us, it seems
eerily lifelike, akin to the difference between black and white
and colour.
High dynamic range will be a familiar term for many smart-
phone users, as it is often an available camera option. When
selected, two images with different exposure times are taken,
and then merged to create a significantly higher range of
contrast than is available in a single image. Capturing HDR
video content is, however, much more difficult, since two
video streams at different exposure levels must be seamlessly
merged to create the desired effect. High-end commercially
available cameras, such as the RED camera, are capable of
capturing HDR video content, but are very expensive and
beyond the reach of the average consumer. HDR displays
are still in the prototype stage.
Investigators in the ICICS Digital Multimedia Lab are addressing the many technological hurdles that must be overcome in order to deliver high dynamic range video content
to consumers, from capturing, processing, and compressing/transmitting, to displaying. There is a general effort
internationally to develop techniques for displaying HDR
video on existing low dynamic range displays (e.g., TVs and
computer monitors). This involves selecting those components from the HDR video that can best be reproduced on
a conventional display, so that the viewer benefits from an
HDR-like rich viewing experience. High dynamic range images prepared for printing (as seen in the photo on page 12)
roughly approximate the appearance (and advantages) of
HDR content shown on an LDR display using these techniques.
It is envisioned that the necessary processing would happen in a set-top box, or perhaps a chip in new televisions
or displays. If these technologies prove sound, we may be
viewing a version of high dynamic range content on our TV
sets within the next five years.
For more information, contact Mahsa Pourazad at
pourazad@icics.ubc.ca
12    Fall/Winter 2014 11   ^
m
\   ■
T
■kLf
■¥]■■
1    H   ■
'1
iFWi
{m
IS1'
■I
i   L
»r
[U
*■ - V8
I
LDR image -
"Its a limitation weve come to accept as the norm, and so when
we are exposed to the level of detail that HDR content reveals to
us, it seems eerily lifelike, akin to the difference between black
and white and colour
innovations magazine Identificatio
and Action
»
14    Fall/Winter 2014 ^^F" the 1970s, Toronto Maple Leafs head
I ^ ^^^ coach Roger Neilson became known as
I I^ "Captain Video" for his extensive use of
III videotape as a coaching tool. Techno-
^^^ ^L ^L logical advances since then have made
video analysis less labour intensive, with major contributions from computer vision expert Jim Little and his colleagues in UBC's Laboratory for Computational Intelligence
(LCI), in automated player tracking, action recognition, and
identification. Once fully implemented, advances such as
these will make it possible to search and select video segments featuring aspects of the game that coaches may want
to emphasize in a practice. Video assistants would no longer
have to pore over footage following a game to manually organize it.
Automated tracking and identification of sports players, as
well as recognizing their actions, presents unique research
problems. Tracking pedestrians in a mall, for example, is
less difficult since their movements are more predictable.
The monitoring cameras are usually
stationary, making it easier to isolate
moving persons in the foreground.
Sports videos, however, are mostly
captured by panning, zooming cameras, making it harder to distinguish
them from a background that also appears to be moving. In addition, players are often bunched
up (as in hockey), their faces obscured, their jersey numbers
hidden, and their movements very rapid, causing blurring
in the video.
Semi-supervised Learning
Sports does, however, have a "narrow semantic space," as
Little points out. Players' activities are largely confined to
those you might expect in a game. Also, teams have a given
player line-up that remains fairly consistent throughout a
season, making it possible to prepare training data offline.
Little does so by having his graduate students draw "bounding boxes" around players and tagging them with the player's
name in a training video, which becomes the template for
tracking the players in a test video. Since the players' shapes
change throughout the game according to their actions, the
templates must be continually updated, using machine-
learning techniques. Thousands of player features, such as
small patches of texture and colour, are automatically gener
ated as the player is tracked. When combined and related to
the offline tagging, they form a classifier that enables automated identification of players regardless of whether or not
their faces or jersey numbers can be seen.
This combination of manual and automated learning is
known in the field of artificial intelligence as "semi-supervised learning". "Training data is one of the key problems in
model learning," Little says. "We start with a small nugget
of labelled data and see how far we can extend it." Little incorporates play-by-play information to narrow the variable
space, so the system knows which players are on the ice/
court/field at a given time.
Estimating Pose
"Training data is one
of the key problems in
model learning."
Currently, Little is focusing on pose estimation as a means
of understanding a player's actions. Many state-of-the-art
systems identify pose in individual frames, by fitting a skeleton to the player. Little is incorporating movement by adding pose information from subsequent frames to
confirm the initial pose prediction.
His work in this area is also finding
application in other areas, including
the ICICS People and Planet Friendly
Home sustainability initiative (see Innovations, Fall/Winter
2012). Little is looking at monitoring the pose of occupants
in a non-intrusive way (where the video is not stored), to
detect anomalies such as falls. This work is complicated by
the fact that the home is a far more varied environment
than the sports arena, with a much wider range of activities.
The setting must be taken into account when assessing the
meaning of an individual's pose, as well as their health profile. A certain pose would be expected, for example, when a
person is loading a dishwasher, but an alert should be triggered if they appear to fall while doing so, particularly if
they are elderly.
Other applications of Little's work include mobile robotics,
where the robot needs to understand its environment and
the actions of nearby individuals in order to navigate safely.
For more information, contact Jim Little at
little@cs.ubc.ca
innovations magazine
Fall/Winter 2014   15 25 Years Later
by Siobhan McElduff
• 9
; i
16    Fall/Winter 2014
%
•     • m
will see the twenty-fifth anniver
sary of UBC's Media and Graphics
Interdisciplinary Centre (MAGIC). Founded in 1990 under the directorship of Computer
Science professor Kellogg Booth, its debut at UBC came one
year before CERN (the European Organization for Nuclear
Research) hosted the world's first website, and two years before the first photograph was posted to the Internet (also by
CERN). MAGIC's mission was—and is—to foster research
in media technology in support of economic development
in British Columbia, and it has adhered to this even as what
is meant by media technology is no longer recognizable as
the entity it was in 1990.
What has made and makes MAGIC special is its interdisciplinary, innovative approach, which brings together researchers from computer science, engineering, medicine,
music, psychology, education, archival science, history, archaeology, and more—in other words, from the entire gamut
of UBC's faculties and departments. (Recently, MAGIC has
been under the interim directorship of Siobhan McElduff,
Associate Professor in Classical, Near Eastern and Religious
Studies. MAGIC's Associate Director is Brian Fisher, Associate Professor in the School of Interactive Arts and Technology, and in the Cognitive Science Program, both at Simon
Fraser University—a testament to MAGIC's diverse interests
and reach.) _
Current projects reflect MAGIC's interdisciplinary and innovative ideals: the Canadian Environmental Health Atlas;
the BC Child Injury Research and Prevention Network and
a range of other visual analytics projects under the supervision of Brian Fisher; the first wearable technologies group at
UBC, created by Junia Anacleto, Visiting Peter Wall Scholar
and Associate Professor in Human-Computer Interaction
at the University of San Carlo, Brazil; the laptop orchestra;
an augmented reality app for a Bronze Age site in Cyprus
in partnership with NGRAIN Corp., Vancouver; a multilingual tool for creating and assessing philosophical arguments; the Global Legal Identity Watch; and the Digital Salon, a group of researchers in the Arts who use and develop
computational tools.
Additionally, as part of its mandate to encourage innovation
at UBC, MAGIC recently launched the 13 (idea, innovation
and inaugurate) Challenge, in conjunction with Mark Salo-
pek of the GRAND (Graphics, Animation and New Media)
national research network. This challenge, aimed at graduate students, is now in its second round, with members of
the first intake already in discussions with investors about
their projects.
For more information, contact Siobhan McElduff at
siobhan.mcelduff@ubc.ca
• •
,y
WUU'
Fall/Winter 2014   17 ng Machine
ion to the
lonsumer
NZ TECHNOLOGIES INC. (NZTECH) IS AN ICICS-INCUBATED START-UP COMPANY THAT DEVELOPS
INTELLIGENT DEVICES FOR SAFETY AND INTUITIVE HUMAN-MACHINE INTERACTIONS IN THE AUTOMOTIVE, RESIDENTIAL, AND CLINICAL SECTORS.
President and CEO Nima Ziraknejad, who founded the
company in 2009, believes that consumers could benefit
more from sensing technologies, particularly machine vision, than they have. As an industrial partner in a research
project sponsored by the AUT021 Network of Centres of
Excellence, NZTech is working to commercialize an adaptive
and automatic head restraint, or headrest, system that helps
prevent whiplash injuries. Ziraknejad was heavily involved
in the underlying research during his PhD studies, develop
ing a hybrid sensor array that adapts the head restraint position to the location and orientation of the driver's head, in
accordance with guidelines from the Insurance Institute for
Highway Safety.
Whiplash is a common, debilitating occurrence in vehicular accidents, most often occurring as a result of rear or
side impacts where the occupant's head is thrown backwards
and sideways, causing soft tissue damage in the neck or back.
Whiplash is extremely painful, and can take months or even
18    Fall/Winter 2014 years to heal, and sometimes never does. Head restraints
can help mitigate the resulting injury, but only if correctly
positioned.
A self-contained, motorized head restraint prototype
system has been developed, with a customized capacitive
sensor array integrated in the
front of the device. Using electric
field sensing, the array measures
the distance between the back of
the occupant's head and the front
of the restraint, and a 3D time-
of-flight camera installed below
the rear-view mirror determines
the orientation of the occupant's
head. This latter information is
important, since whiplash severity is greater when the vehicle occupant's head is turned—the restraint needs to be that
much closer to limit the injury.
"With the fused sensor data," Ziraknejad says, "we can
control motors embedded in the head restraint to raise
and lower it, and move it forward and backward." Ongoing
research will determine the optimal position for mitigating whiplash injuries, given the occupant's head position
and orientation. Appropriate intervals for re-positioning
the headrest in the course of normal driving will be determined as well. Knowledge of the driver's head orientation
also lends itself to possible driver distraction/fatigue detection applications.
Machine Vision at Your Fingertips
On the residential side, NZTech is part of a six-
member industrial consortium led by TELUS
that is supporting the ICICS "People and Planet Friendly Home" sustainability initiative (see
Innovations, Fall/Winter 2012). In this project, 25 researchers from 7 different UBC departments are developing novel
integrated technologies for residential spaces that promote
both sustainability and quality of life.
In the research strand supported by NZTech, machine
vision and cloud computing are being harnessed to assist
in cooking and baking. The researchers and NZTech have
developed a small, portable device equipped with 3D cameras, a thermometer, and an ARM processor-based embedded system, that attaches to the outside of the window in
the oven door. As an item is cooking, it selectively senses for
designated states, such as a certain colour, that will indicate
that the item is done. This information is correlated with
"We can control motors
embedded in the head
restraint to raise and
lower it, and move it
forward and backward."
the item's internal temperature determined by a cooking
thermometer, and an alert together with live video frames of
the food are sent to the resident on a mobile device.
Selective sensing does not capture the entire object of interest, therefore requires less bandwidth to transmit the live
images. In addition, less onboard
preprocessing is required, reducing power consumption. Additional processing, if necessary, is
done on a cloud-based server.
"There is a trade-off,"
Ziraknejad states. "You might be
able to do all of the processing
onboard, but at the cost of higher
power consumption. By leaving
the heavy processing for the cloud, we can also keep the system working in real-time, which is critical."
Real-time, portable machine vision and processing for
a variety of tasks in the home is NZTech's ultimate goal for
the device, having developed a proof-of-concept prototype
for the oven application. The company is now working on a
pan-and-tilt mount, so the device can be placed anywhere
in the home and targeted as required. They are also investigating the feasibility of wireless power transmission to the
device.
"We hope to have a suite of image-processing modules
for different applications," Ziraknejad says, "including drug
dosage reminders and fall detection. We want to provide
machine vision at your fingertips." He means this last point
both figuratively and literally. The company is developing
a gesture and voice-based interface, associated with a light
projector that can project results on any surface. Internet-
based information such as weather or traffic patterns could
also be requested by a voice command, then projected on a
designated surface with a certain gesture.
"We are aiming at an affordable, general platform that
can do many things for you," Ziraknejad states. With a working prototype in hand and ICICS research heft behind him,
NZ Technologies is well on the way.
For more information,
contact NZ Technologies at info@nztech.ca
innovations magazine
Fall/Winter 2014   19 We care about
your questions big and small.
TELUS   "*'
Learning
Centre8 w
Want to get the most out of your
smartphone or tablet? Book a free
one-to-one learning session.
It's in our nature to care.
Visit telus.com/learningcentre
to learn more.
^Ttelus
the future is friendly"
TELUS, the TELUS logo, the future is friendly and telus com are trademarks of TELUS Corporation, used under licence All other trademarks are the property of their respective owners ©2014TELUS
UBC's Institute for Computing, Information and Cognitive Systems (ICICS) is an
umbrella organization that promotes collaboration among researchers from the faculties of
Applied Science, Arts, Commerce, Education, Forestry, Medicine, and Science, and with
industry. ICICS facilitates the collaborative multidisciplinary research of approximately
200 faculty members and 1,000 graduate students in these faculties.
Our members attract approximately $18 million annually in grants and contracts. Their
work strengthens Canada's strategic Science and Technology research priority areas,
benefiting all Canadians.
fAUUESTiu** vuAiuitncr ^^^^^
a place of mind
THE UNIVERSITY OF BRITISH COLUMBIA
CONNECTING KN0WLEDC
PUBLICATIONS MAILAGREEMENT NO. 40049168
RETURN UNDELIVERABLE CANADIAN ADDRESSES TO:
ICICS, University of British Columbia
289-2366 Main Mall
Vancouver, BC V6T 1Z4
info@icics.ubc.ca
www.icics.ubc.ca

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/cdm.focus.1-0225871/manifest

Comment

Related Items