Socially assistive robots

PDF

Download

Open document

Transcript (if available)

Content

SOCIALLY ASSISTIVE ROBOTS

by

Megan Elizabeth Chao

A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF ARTS
(BROADCAST JOURNALISM)

May 2008

Copyright 2008 Megan Elizabeth Chao
ii

Dedication

This project is dedicated to my parents, Sui and Michael Chao,
and my best friends, Hanna Alexander, Justin Patananan,
Cristian Coroian and Norton Kitagawa
for their continual support and encouragement
in my pursuit of higher education.
iii

Acknowledgements

I am indebted to my faculty mentor and professor of journalism, Daniel H. Birman, for
his friendship and professional collaboration throughout my two years at the Annenberg
School for Communication. His guidance and insight have meant a great deal to me.
Above all, he believed in me, and it is because of his encouragement and investment
into my ideas that I was able to take on this project, triumph over obstacles, and see it
through to fruition.

I am also grateful to Maja Matarić, Eric Wade and Nathan Koenig from the University of
Southern California’s Robotics Laboratory for taking the time out of their busy lives to sit
down and chat with me about their research, which in turn contributed a great deal to
the contents of this professional project. They were generous in allowing me access to
their materials, as well as movies and other helpful resources. Without them, this
project would not have been as successful.

I would also like to thank Timothy Miller and Lee T. Warner at the Annenberg School for
Communication, for providing the means to bring this project together into something
viewable for television. In addition, I would like to thank Joseph Saltzman, professor of
journalism, for imparting his knowledge and expertise in documentary production upon
me. I know his lessons will stay with me as I make my own imprint on the industry.

iv

Table of Contents

Dedication ii
Acknowledgements iii
Abstract v
Main Body 1
References 17
v

Abstract

As technology evolves and advances in the realm of healthcare, humans may soon be
interacting actively with robots for rehabilitation purposes. Called socially assistive
robots, three target groups have participated in experiments and pre‐clinical trials—
children with autism spectrum disorder, patients in post‐stroke rehabilitation, and the
elderly, in the prevention of mental and physical decline. Maja Matarić, professor of
computer science and neuroscience at the University of Southern California, her
students in the robotics laboratory, and their collaborators around the world have taken
on the creation and design of robots to promote social interaction and friendly physical
activity among such groups, while addressing the misconception that robots are harmful
entities. They also consider major ethical issues surrounding the incorporation of robots
into mainstream society, including medical privacy laws, insurance companies in their
recognition of robots as a form of treatment, and affordability.

1

Robotics
Producer: Megan E. Chao
Final Version

NATS
I2672 0:07:19
CU Bandit the Robot’s eyes and
eyebrows, zoom out, moving arms

(NATS: female voice emanating from
robot)

Please raise your arms straight up, like
this. (pause) Now extend your arm to the
side, like this. Look at how I do it.

VO Track 1
I2672
MS Bandit continuing to do demo
workout, zoom out to WS

MEET BANDIT.
NATPOP
I2672
WS Bandit continuing through demo
workout

(NATS: female voice emanating from
robot)

Now please extend your other arm, like
this. Good job.

VO Track 2
I2672
WS Bandit continuing through demo
workout
BANDIT IS A SOCIALLY ASSISTIVE ROBOT,
DESIGNED SPECIFICALLY FOR HUMAN
ROBOT INTERACTION. RIGHT NOW,
BANDIT IS GUIDING US THROUGH AN
EXERCISE ROUTINE.

NATPOP
I2672
Bandit lowers arms, finishes workout,
smiles

(NATS: female voice emanating from
robot)

You can lower your arms now. We are
finished. You did an excellent job.
2

SOT Maja Matarić, Professor
I2671 0:00:59‐0:01:10
Socially‐assistive robots are robots that
help people without actually touching the
people, through social interaction.

VO Track 3
Stock footage: autistic child speaking
with a robot, talking and touching the
robot (courtesy of USC Robotics
Laboratory)
WITH THE ABILITY TO ENGAGE THEIR
USERS IN CONVERSATION AND PHYSICAL
ACTIVITY, THESE ROBOTS ARE ABLE TO
HELP SOME HUMANS WITH SPECIAL
NEEDS.

SOT Maja Matarić, Professor
I2671 0:02:19‐0:03:00
Stock footage: autistic child speaking
with a robot, talking and touching the
robot (courtesy of USC Robotics
Laboratory)
The idea behind socially‐assistive robotics
is that the robot serves a really kind of a
complicated role, whether as a buddy, as
a nurse, a companion, or a coach. So for
example, for a child with autism, a
socially‐assistive robot can engage the
child in interaction and play, to help train
the child to take turns, something that
children with autism have trouble with.

SOT Nathan Koenig, Graduate Student
I2370 0:10.27‐0:11:09

Children with autism tend to work better
with non‐human entities. They get a little
bit overwhelmed by the complex facial
expressions that go on with a human so
they tend to open up and interact more
with toys, with robots and things like that.
Things that are more stable, that they can
understand and are more regular. Having
a robot in an environment that they can
predict and understand and is not
overwhelming their sensory input, which
is preferable in children with autism.
3

VO TRACK 4
Stock footage: group of elderly patients
in a senior living complex in Silver Lake
exercising with the guidance of a robot,
some who have suffered from a stroke
(courtesy of USC Robotics Laboratory)

IN ADDITION, THE ROBOTS ARE DESIGNED
TO HELP PATIENTS WHO HAVE LOST LIMB
FUNCTION AFTER SUFFERING FROM A
STROKE. BANDIT CAN ACT AS A PHYSICAL
TRAINER, WORKING WITH PATIENTS TO
RESTORE MOVEMENT IN THOSE LIMBS.

SOT Maja Matarić, Professor
I2671 0:03:02‐0:03:42
Stock footage: group of elderly patients
in a senior living complex in Silver Lake
exercising with the guidance of a robot,
some who have suffered from a stroke
(courtesy of USC Robotics Laboratory)
A socially‐assistive robot that helps a
stroke patient is going to help that stroke
patient be motivated to do a lot of
exercises that are necessary, to recover
function after a stroke. Suppose someone
has had a stroke in their right arm, maybe
their dominant arm, and their dominant
arm is now disabled, maybe completely
paralyzed. Now that person should really
exercise their arm in doing everyday
things, like reaching for a newspaper,
reaching for the cereal, opening the door,
and that’s very hard to do because what
the brain does is that it compensates, and
uses the left arm instead. It’ll get the job
done, but not to recover. What the robot
does is it says, “No, no, no, come on now,
use the right arm. Oh, that’s really good.
Now try one more time. Good job.” So it
provides motivation, but also coaching.

SOT Maja Matarić, Professor
I2671 0:01:12‐0:01:33
Stock footage: focus in on certain people
in group of elderly patients in a senior
living complex in Silver Lake exercising
with the guidance of a robot, some who
have suffered from a stroke (courtesy of
USC Robotics Laboratory)
They have the potential to help a large set
of different populations including stroke
victims, patients who have suffered
traumatic brain injury such as in a car
accident or in a war, patients with
Alzheimer’s, or children with autism
spectrum disorder. They can serve just a
wide set of different disabilities.
4

VO TRACK 4
WS three Bandit robots, old to new

MAJA MATARIĆ’S ROBOTICS LAB AT THE
UNIVERSITY OF SOUTHERN CALIFORNIA IS
THE FIRST TO TEST ROBOTS AS SOCIAL
TOOLS FOR CHILDREN WITH AUTISM AND
AS REHABILITATION TOOLS FOR STROKE
PATIENTS. AND ACCORDING TO THEIR
STUDIES ON THESE GROUPS, PATIENTS
ENJOY SPENDING TIME WITH ROBOTS
AND INTERACT WITH THEM AS IF THEY’RE
INTERACTING WITH ANOTHER HUMAN.
THIS IS IN PART, BECAUSE OF THE
ROBOT’S DESIGN.

SOT Eric Wade, Postdoctoral Fellow
I2670 0:13:30‐0:14:19
Naked Bandit robot
The robot is made up of three main
components. The most noticeable of all is
the humanoid torso. As you can see, he
can do some limited facial expressions
like smiling, he can move his eyebrows,
his head is actuated, his arms are also
actuated, as are his hands. He has less
degrees of freedom, obviously, when
comparing him to a normal human, but
the point is that he has some sort of
humanoid characteristics.

SOT Eric Wade, Postdoctoral Fellow
I2670 0:14:20‐0:15:18
Tour of Bandit’s backside circuitry with
Wade

On/off switch, stop button
USB serial converter

So if you come around to the back, you
can see a couple of things… first I’ll
unbutton his shirt, if you will, and you can
see his on/off switch, we have a stop
switch. And the mechanism with which
he communicates with the computer is
with this USB serial converter, right here.

5

SOT Eric Wade, Postdoctoral Fellow
I2670 0:14:20‐0:15:18
Tour of Bandit’s backside circuitry with
Wade

Motherboard
Hard drive
Speaker
So that brings us to the computer, the
control computer that we’re using.
As you can see here, you can see a small
profile computer and the board is
something that you would see similarly in
a desktop PC. It hooks up to a hard drive,
here. And what we’re doing here is
controlling all of Bandit’s actions.
Anything that Bandit says, if you will,
which will come out of this speaker, is
going to be triggered in this board. Any of
his physical motions, behaviors, the
different behavior models. The different
personalities that we have for the model
are controlled using this computer.

VO TRACK 5
Wide, open space in lab, Bandit robot
sitting in the middle, looking around
THE WAY THAT THE ROBOT WILL KNOW
HOW TO RESPOND TO ITS USER IS BY
WATCHING AND LISTENING TO THE
PATIENT.

SOT Maja Matarić, Professor
I2671 0:04:57‐ 0:05:30
CU robot’s camera eyes
CU robot’s inputs for microphones
Wireless band for patients to wear

What the robot does is that it has a series
of different sensors. It may have cameras
to see what’s going on in the
environment. It may have microphones
to hear what’s going on in the
environment. And it may have access to
the participant’s physiological responses
like heart rate, blood pressure, body
temperature, skin conductivity, all sorts of
inputs that it takes in as it’s behaving, as
it’s responding to the person so that it
really is engaged in a real social
interaction, like see and hear what you’re
doing and it can respond to that. It’s
really engaged with you, in real time.

6

SOT Nathan Koenig, Graduate Student
I2670 0:03:53‐0:04:30
Koenig with Bandit

We’ve been developing a bit more
computational power on the back side of
the robot, in order to process sensory input
properly and in a real‐time manner. So
we’ve been developing some more inputs
to the robot from the stereocamera into
the computer itself to do vision processing
and even attaching microphones in here.

VO TRACK 6
Mobile base detached from humanoid
torso
THIS MEANS THAT THE ROBOT CAN
QUICKLY ANALYZE WHAT THE PATIENT
SAYS AND DOES, AND REACT. AND
BANDIT, WHILE IT HAS A HUMAN‐LIKE
TORSO, HAS WHEELS FOR LEGS.

SOT Nathan Koenig, Graduate Student
I2670 0:05:08‐0:05:42
Stock footage: mobile base following a
person around a room based on sensory
inputs (courtesy of USC Robotics
Laboratory)

The reason why we use a mobile base such
as this is because it’s easy to drive around
in environments. Having a robot being
stable on two legs is difficult and is still
under active research. We’re more in the
field of tackling human‐robot interaction
and not really the dynamics of legged
motion.

SOT Eric Wade, Postdoctoral Fellow
I2670 0:15:20‐0:15:56
Tour of mobile base with Wade

It’s plugged in and being charged right now,
but in practice it would not be plugged in
and would be fully mobile. It is sort of a
standardized base. You can see it says
“Pioneer” here. We have a number of
these “Pioneers” here in the lab, there are
several models and essentially what they
are, are just mobile bases with which to
move around. Those would be the sonar
sensors here, there are DC motors used to
drive.
7

SOT Eric Wade, Postdoctoral Fellow
I2670 0:15:20‐0:15:56
Tour of mobile base with Wade

This is what will be used to drive the robot
around the user, in all of our experiments.

VO TRACK 7
Slow zoom out from tour of mobile base
to show whole Bandit robot, and Eric
Wade standing next to Bandit

A MOBILE BASE, SAYS MATARIĆ, KEEPS THE
ROBOT FROM LOOKING INTIMIDATING. AT
MOST, IT WILL BE THREE‐FOURTHS THE
SIZE OF A FULLY‐GROWN HUMAN AND
WILL NOT BE PROGRAMMED TO GET
PHYSICAL WITH ITS USERS. AND THE MOST
VALUABLE FEATURE ‐‐ ITS
CUSTOMIZABILITY.

SOT Maja Matarić, Professor
I2671 0:07:33‐0:08:30

The idea is to make these so affordable
that people can do so, they can go out and
get one customized by their health care
professional. And the robots of course, can
also work with therapists. It’s not just like
a toy; it can do serious things like
rehabilitation. It’s really medically related
and medically justified.

SOT Nathan Koenig, Graduate Student
I2670 0:08:59‐0:10:22
WS USC Robotics laboratory

We have done limited studies on this and
the hope is that it will interact with people
in a more natural setting. We have a range
of experiments that we conduct‐‐ we have
students here working on emotional
studies, interpreting how people react to
different emotional states and incorporate
it with the robot, and also using the robot
to learn what kind of behavioral styles a
person prefers, whether the robot should
behave more coach‐like or should be more
of a nurturing, motherly type of persona.
8

SOT Nathan Koenig, Graduate Student
I2670 0:08:59‐0:10:22
WS USC Robotics laboratory

And ideally we can express these types of
emotional states not only through speech
and voice, but also through face and
gestures.

SOT Maja Matarić, Professor
I2671 0:06:28‐0:07:25
Stock footage: Robot follows patient
walking around room, patient stops, robot
rolls up to patient (keeps a 4 ft distance)
(courtesy of USC Robotics Laboratory)

One of the main benefits of having the
robot interact with the person is the
robot’s ability to detect if the person is
getting frustrated, if they’re happy, if
they’re doing well at a task or maybe if
their performance is declining. So how
does the robot do that? Well, first of all,
the robot watches for example, if you have
to stack books, it watches you, it’s
counting, and it’ll say, “Great, you can do
two more. Let’s do one more. Yesterday
you did five.” and so on and so forth.

NATPOP
I2672 0:09:28‐0:10.03
Stock footage: Patient begins putting
books on book shelf, robot uses human
voice, body motion and is persistent in
encouraging patient (courtesy of USC
Robotics Laboratory)
(NATS: Robot and patient)

Robot: Hi there, how are you?
Patient: I’m fine. (laughs)
Robot: How about we do a fun activity?
You can play librarian, and put the books on
the bookshelf.
Patient: Okay. (begins putting magazines
into shelf)
Robot: Great job, keep it up.

SOT Maja Matarić, Professor
I2671 0:06:28‐0:07:25
Stock footage: Patient begins putting
books on book shelf, robot uses human
voice, body motion and is persistent in
encouraging patient (courtesy of USC
Robotics Laboratory)

So it can keep history, but it can also watch
what you’re doing. But it also has access to
your heart rate, your skin conductivity, so it
can tell, “Hey, maybe you’re getting
frustrated,” and so it knows that and can
say something like, “Well maybe we should
do something else” or “How about a little
break?”

9

SOT Maja Matarić, Professor
I2671 0:06:28‐0:07:25
Stock footage: Patient begins putting
books on book shelf, robot uses human
voice, body motion and is persistent in
encouraging patient (courtesy of USC
Robotics Laboratory)

So the robot really has a lot of access,
hopefully, to what you’re doing, how
you’re doing, how you’re performing, and
in turn provide that encouragement, that
motivation, and really be as helpful as
possible.

VO TRACK 8
Stock footage: patient stacking different
sized blocks on top of one another,
robot watching patient (courtesy of
USC Robotics Laboratory)
BANDIT KEEPS A LOG OF THE PATIENT’S
MOVEMENTS, VITAL SIGNS AND VOICE
RECORDINGS – INFORMATION IT USES TO
CREATE INTERACTION. BUT THIS STORAGE
OF INFORMATION CAN POTENTIALLY
VIOLATE HIPAA, A FEDERAL MEDICAL
PRIVACY LAW BECAUSE IT MIGHT BE ABLE
TO TRANSMIT THE INFORMATION TO
ANOTHER PERSON LIKE A DOCTOR, NURSE
OR PHYSICAL THERAPIST. HOWEVER,
STEPS ARE BEING TAKEN TO ENSURE THAT
ONLY CERTIFIED PERSONNEL WILL HAVE
ACCESS TO THIS INFORMATION.

SOT Maja Matarić, Professor
I2671 0:09:40‐0:10:48
Robot built without webcam eyes,
plastic eyes in place
Other variations on robot physical
features, shows customizability of
device

As we think about these technologies in
people's lives though, to closely interact
with people, it's very important to think
about privacy issues, ethical issues, safety,
let’s say. Safety is something we worry
about. We want to make sure the robot is
safe. And privacy, one of the things for
example, we worry about is to make sure
that the data that the robot has access to
such let's say blood pressure.

10

SOT Maja Matarić, Professor
I2671 0:09:40‐0:10:48
Robot built without webcam eyes,
plastic eyes in place
Other variations on robot physical
features, shows customizability of
device

The user is completely aware of what kinds
of data is coming into the robot and the
user can choose not to use the data. You
can have a robot that interacts with you
but never uses cameras. So there's no
video of what you're doing. There may be
other inputs into the robot, like it knows
whether your heart rate is going up or
down, but because it has no cameras,
people are more comfortable with that.
Privacy is definitely an issue that we have
to worry about, and it's an issue that cuts
across many technologies in robotics.

VO TRACK 9
Clip from movie I, Robot: driving scene
where enemy robots come jumping out
of two US Robotics trucks and onto
Detective Spooner’s Audi RSQ Concept,
car spins 360 degrees in a tunnel as
robots attack

AND WHILE PRIVACY REMAINS ONE OF THE
LAB’S MAIN CONCERNS WITH
INCORPORATING ROBOTS INTO THE
MAINSTREAM, THEY ARE NOT WORRIED
ABOUT ROBOTS TAKING OVER THE WORLD
OR HARMING HUMANS.

NATPOP
Clip from movie I, Robot: driving scene
where enemy robots come jumping out
of two US Robotics trucks and onto
Detective Spooner’s Audi RSQ Concept,
car spins 360 degrees in a tunnel as
robots attack

(NATS: scene from the movie, I, Robot)

SOT Maja Matarić, Professor
I2671 0:13:28‐0:14:00
Clip from movie I, Robot: driving scene
where enemy robots come jumping out
of two US Robotics trucks and onto
Detective Spooner’s Audi RSQ Concept
Most people are familiar with robots from
the movies. And for some reason, robots
are always evil. There are a few
exceptions, of course, with nice robots, but
in the movies they’re always usually the
evil killer robots. It’s just terrible.
11

Car spins 360 degrees in a tunnel as
robots attack

Montage of clips from movie
Bicentennial Man: robot coming out
of the box, helping family around the
house, doing chores, tucking children
into bed

It’s all wrong! So we’re trying to fight that
misconception that robots are evil and that
they’re dangerous. Robots can only be as
evil or dangerous as people create them. I
mean, robots are made by people. And
yes, they can be made by people to be
really bad, or they can be really wonderful.

NATPOP
Montage of clips from movie
Bicentennial Man: robot coming out
of the box, helping family around the
house, doing chores, tucking children
into bed

(NATS: scene from movie, Bicentennial
Man)
VO TRACK 10
I2672 0:02:25‐0:03:08
Wade making adjustments on robot
WHILE THE TECHNOLOGY BEHIND THESE
ROBOTS IS COMPLEX, THE ROBOTS WON’T
BE TOO EXPENSIVE.

SOT Maja Matarić, Professor
I2671 0:07:33‐0:08:30

Right now, we’re considering technologies
that would be affordable, for people to
buy. Another way to look at this would be
maybe a set of robots that a health care
provider could loan out to people so that
they could have it when they need it most,
say for the first three months after a
stroke.

12

SOT Maja Matarić, Professor
I2671 0:12:54‐0:13:23

No one should have to go through special
training, or have a special level of
education or income to have access to this
technology. If socially assistive robotics is
really going to help people, it needs to be
engaging and ready to use out of the box.

SOT Maja Matarić, Professor
I2671 0:08:33‐0:09:24
Roomba vacuum cleaner roaming around
a living room floor

We’re looking at a range of about $500 to
$1000, not tens of thousands. If you look
at the Roomba vacuum cleaner that you
can buy for $200 now, we’re trying to make
it comparable. It’s important to keep in
mind though, that if people want a robot
that looks like a human, has a humanoid
form, you know, two arms, a head, moves,
arms can move, that’s a little more
expensive. The whole philosophy is to
make the robots small and cheap and
affordable and not very physically capable,
but friendly and engaging and smart
enough to really help you.

VO TRACK 11
Food and Drug Administration building
and website
THE NEXT STEP IS TO GET INSURANCE
COMPANIES TO RECOGNIZE THIS AS A
FORM OF TREATMENT. MATARIĆ SAYS
THIS WON’T HAPPEN THOUGH, UNTIL THE
ROBOTS GO THROUGH CLINICAL TRIALS
AND ARE APPROVED BY THE FOOD AND
DRUG ADMINISTRATION, A PROCESS THAT
WILL TAKE AT LEAST THREE YEARS.

13

SOT Maja Matarić, Professor
I2671 0:09:28‐0:09:40

The goal in the not too distant future is
actually to demonstrate to the health care
providers and insurance companies that
this technology can really improve the
quality of life, and we want to make it
recoverable through insurance costs and
whatnot to make it affordable, to make it
available to people.

VO TRACK 12
Bandit the Robot rolling around the
laboratory, graduate students making
adjustments
BANDIT WILL SOON BE ON ITS WAY TO THE
MAINSTREAM PUBLIC. IT IS READY TO
ENTER THE CLINICAL TRIALS PROCESS,
FOLLOWING A FEW MORE EXPERIMENTS,
TOUCHUPS ON ITS DESIGN AND
SOFTWARE. IT HAS ONLY TAKEN THE U‐S‐C
ROBOTICS LABORATORY A FEW YEARS
FROM INITIAL DEVELOPMENT TO THE
CURRENT VERSION OF BANDIT, IN
COLLABORATION WITH OUTSIDE
CONTRACTORS LIKE BLUE SKY ROBOTICS.

SOT Nathan Koenig, Graduate Student
I2670 0:05:59‐0:07:00
Koenig with Bandit robot
The idea for the humanoid torso came
about 3 years ago. We had a student come
in and develop the very first iteration of
this type of robot. The problem was that it
was rapid prototyped so it wasn’t very
robust. So since then, we have contracted
out to Blue Sky Robotics to develop a much
more dynamic, fully‐featured robot good
for use in both a school situation and in the
research lab.

14

SOT Nathan Koenig, Graduate Student
I2560 0:00:34‐ 0:01:30
Naked Bandit robot, original prototype

Blue Sky Robotics has done pretty good
work for us so far. There have been some
stages of development here, this is a very
crude stage of this robot, very prototype.

SOT Nathan Koenig, Graduate Student
I2560 0:01:34‐0:01:52
Pan across to next stage of Bandit robot

We’ve moved on to a more improved,
more human‐like robot. Even so, we’re still
missing his hands. And the final version
has a more humanoid structure.

SOT Nathan Koenig, Graduate Student
I2670 0:01:53‐0:02:33
Pan across to final, current stage of
Bandit robot

This looks a little more human‐like. It has
some more facial features. It can move.
The eyes change position, it has simple
hands to grip objects, and we’ve even put it
on a mobile base so we can move it around
to interact with people and the
environment.

VO TRACK 12
Pan across to final, current stage of
Bandit robot

THEY MAY BE MECHANICAL, YET THESE
ROBOTS SEEM TO BLEND RIGHT IN WITH
HUMANS, WHICH MATARIĆ SAYS OFFERS
PROMISE FOR THE FUTURE.

SOT Maja Matarić, Professor
I2671 0:03:51‐ 0:04:54

We have done extensive experiments with
robots interacting with stroke patients,
with children with autism, with Alzheimer’s
patients, and we’re really very happy with
the report that we’ve never had any of the
patients report any disagreement or dislike
for the robot. It’s really very interesting to
see the different, many different ways that
people interact with the robot.

15

SOT Maja Matarić, Professor
I2671 0:03:51‐ 0:04:54

Some people really start to play with the
robot and that may not make them listen
to the robot… they’re sort of having more
fun and not being coached as they should
be.

NATPOP
Stock footage: Child mocking robot
(courtesy of Robotics Lab)

(NATS: child laughing at robot)

SOT Maja Matarić, Professor
I2671 0:03:51‐ 0:04:54
Bandit imitating a person’s movements,
arms up, arms to the side, arms down,
head turning side to side
And other people are very responsive to
what the robot has to say. For example if
the robot says, “Raise your arm” they do
that just perfectly, but sometimes they may
not necessarily seem like they’re having as
much fun as we’d like. So there’s still a lot
of research to be done in terms of how you
manage what the robot says and does, how
it interacts and behaves, and how it
motivates people in order to be effective.
But we’re happy to say that already, we’re
not having any adverse effects.

SOT Eric Wade, Postdoctoral Fellow
I2670 0:16:21‐0:16:40
Bandit imitating a person’s movements,
arms up, arms to the side, arms down,
head turning side to side

The idea is that we just want it to talk and
interact with the user sort of in a normal
way that you would want to interact with a
physical therapist.

SOT Maja Matarić, Professor
I2671 0:12:54‐0:13:23
Bandit robot waving goodbye

If we design these robots right, then they
will be friendly and accessible.

16

NATPOP
Bandit robot waving goodbye
(NATS: sound of motors in robot’s arm
actuating)

17

References

1. Nathan Koenig, Graduate Student
Interaction Lab, Department of Computer Science
Viterbi School of Engineering
University of Southern California
3650 McClintock Avenue, OHE 200
Los Angeles, CA 90089‐1450
Tel: (213) 740‐6245, Fax: (213) 821‐5696
nkoenig@usc.edu

2. Maja Matarić, Professor of Computer Science and Neuroscience
Founding Director, USC Center for Robotics and Embedded Systems
Director, USC Robotics Research Lab
Senior Associate Dean for Research, Viterbi School of Engineering
University of Southern California
3650 McClintock Avenue, OHE 200
Los Angeles, CA 90089‐1450
Tel: (213) 740‐4520, Fax: (213) 821‐5696
mataric@usc.edu

3. Eric Wade, Postdoctoral Fellow
Interaction Lab, Department of Computer Science
Viterbi School of Engineering
University of Southern California
3650 McClintock Avenue, OHE 200
Los Angeles, CA 90089‐1450
Tel: (213) 740‐6245, Fax: (213) 821‐5696
ericwade@usc.edu

Abstract (if available)

Abstract As technology evolves and advances in the realm of healthcare, humans may soon be interacting actively with robots for rehabilitation purposes. Called socially assistive robots, three target groups have participated in experiments and pre-clinical trials -- children with autism spectrum disorder, patients in post-stroke rehabilitation, and the elderly, in the prevention of mental and physical decline. Maja Matarić, professor of computer science and neuroscience at the University of Southern California, her students in the robotics laboratory, and their collaborators around the world have taken on the creation and design of robots to promote social interaction and friendly physical activity among such groups, while addressing the misconception that robots are harmful entities. They also consider major ethical issues surrounding the incorporation of robots into mainstream society, including medical privacy laws, insurance companies in their recognition of robots as a form of treatment, and affordability.

Linked assets