University of Southern California Dissertations and Theses

Robot, my companion: children with autism take part in robotic experiments

(USC Thesis Other)

Robot, my companion: children with autism take part in robotic experiments

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Robot, my companion —
children with autism take part in robotic experiments

by Riikka Haikarainen

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

August 2015

© by Riikka Haikarainen 2015

ii

Table of Contents

Abstract .......................................................................................................................iii

Robot, my companion—children with autism take part in robotic experiments...........1

Bibliography................................................................................................................22

iii

Abstract

Socially assistive robots are not only mechanical helpers but also engage in social interaction
with humans. More than 20 robots have been used in experiments with children diagnosed
with autism spectrum disorder or ASD. Some scientists claim that robots can help children on
the spectrum to engage socially, one of the biggest challenges for autistic kids. In the robot
experiments, the interaction between children and robots has concentrated around the very
basics of human interaction: eye-gaze following, non-verbal communication, and imitation.
However, critics of social robotics have voiced concerns about our interactions with
machines. As robots grow more prominent, we might start to feel safe in a robotic relationship
that lacks the ups and downs of human interaction. The robot researchers working for children
with autism argue that social robots are not meant to replace therapists but facilitate
communication between the children and other humans.

1

A white, plastic humanoid figure lies on its back on a desk in front of a classroom filled with
some twenty college students. Their eyes are fixed on the two-foot tall creature: a robot
called Nao. Wearing an astronaut suit, with two round eyes flashing blue light staring out
from its rectangular face, it looks like an action figure.

Elaine Short, a Ph.D. candidate at University of Southern California (USC), reaches over to
her laptop on the desk next to Nao (pronounced "now"). She clicks an icon on screen and,
suddenly, the lecture room is filled with Chinese instrumental string music. The motors in
Nao's joints start to whir as it gracefully stretches out its hands and legs, and then balances on
one leg. Nao is practicing Tai chi, a Chinese health and meditation exercise.

"I’ve had one of these robots fall off the desk and it was very upsetting," Short says, looking
concerned. With its two cameras, four microphones and powerful computing capabilities, the
high-tech robot is worth $8,000.

As the music stops, Nao slows down its movements and sits down elegantly at the desk,
prompting students to erupt in thunderous applause. Nao is not just a local USC celebrity. It
has gained popularity worldwide, greeting a television audience on NBC's Today Show and
dancing in a music video to the song, “Gangnam Style.”

However, Nao is programmed not only to entertain but also to work—training people in both
physical and social skills. Researchers like Elaine Short call robots such as Nao "socially
assistive robots.” The term refers to robots that are not only mechanical helpers but also
engage in social interaction with humans. The latter is Nao's specialty.

Nao is one of more than 20 robots that have been used in experiments with children
diagnosed with autism spectrum disorder or ASD. Some scientists claim that a robot like Nao
can help children on the autism spectrum to engage socially, one of the biggest challenges for
autistic kids. There is a growing need for that kind of assistance. "Parents of autistic children
are sometimes pretty desperate," Short says.

2

The absence of social interaction causes severe stress in families with autistic children. “The
expectation of having that [communication] with your offspring, the drive to interact between
the caregiver, parent and child is so immensely powerful that the lack of it has to be
incredibly disconcerting. And it is,” says Pat Levitt, a professor of neuroscience, psychiatry
and pharmacy at USC. As an expert in child and brain development, Levitt has served
associations such as Autism Speaks.

One of those affected parents is Lisa Delbri. She’s a mother of a 10-year-old Matheo who has
ASD and doesn't speak at all. "He thinks he’s telling you something and doesn’t understand
why you’re not getting what he’s saying. That is heartbreaking," Delbri explains in a video
interview conducted by scientists at the University of Notre Dame in 2010
1
.

In the video, Matheo—a slender boy with golden-brownish hair—stands next to a swimming
pool and rubs his palms and fingers together. He gazes downwards with almost closed eyes;
he's not paying any attention to the shrieks and laughter coming from the pool where his
siblings play with a beach ball.
2

In the next scene, Matheo is taking part in an experiment with Nao. The small boy is sitting
on a chair next to a therapist; Nao stands on a table in front of them. “Matheo, how are you?“
the robot says. Then it raises its arms above its head and brings them back down. “What did
it do?” the therapist asks Matheo. The boy stands up from his chair, imitates Nao’s arm
gesture and smiles.
3

* * *

No one really knows what causes autism, and it is a life-long, incurable condition. Currently,
the Centers for Disease Control and Prevention (CDC) estimates that about 1 in 68 children

1
University of Notre Dame (2010): Fighting For Breakthroughs in the Treatment of Autism. Retrieved July 9, 2015:
https://www.youtube.com/watch?v=IeoPNc0V7Io.
2
University of Notre Dame (2010).
3
University of Notre Dame (2010).

3

has been identified with ASD
4
. From 2002 to 2008, the estimate jumped 78 percent
5
. Some
have called this increase an autism epidemic.

However, a significant part of the increase is due to the growing awareness of the disorder, a
broader definition of autism spectrum disorders, and better diagnosing tools. Children are
also being diagnosed younger than ever before: ASD can be sometimes detected already at
18 months and reliably diagnosed by the age of two
6
.

“The prevalence of autism today is a virtue, maybe even a prize,” writes Roy Grinker, a
professor of anthropology and a father of an autistic daughter, in his book Unstrange Minds –
Remapping the world of autism
7
. Grinker claims that if a child is diagnosed with autism, she
will most likely benefit from the growing sophistication of educators, speech therapists,
psychiatrists, and psychologists
8
. The early and increasing diagnosing of autism may cause
the apparent numbers of autistic children to explode; but from a parent’s perspective, in the
ideal case it means that treatment starts sooner and increases the chances of interventions
being more effective in improving symptoms.

Whether or not early diagnosis has a positive side, researchers will continue exploring a
variety of possible reasons for the accelerating rate of autism diagnoses. According to CDC,
"a true increase in the number of people with an ASD cannot be ruled out."
9
The guarded
statement suggests that perhaps something in our environment, together with genetic factors,
is causing more children to have ASD.

4
Baio, Jon (2014): Prevalence of Autism Spectrum Disorder Among Children Aged 8 Years — Autism and
Developmental Disabilities Monitoring Network, 11 Sites, United States, 2010. Centers for Disease Control and
Prevention Surveillance Summaries. March 28, 2014 / 63 (SS02); 1–21.
5
Baio, Jon (2012): Prevalence of Autism Spectrum Disorders — Autism and Developmental Disabilities Monitoring
Network, 14 Sites, United States, 2008. Centers for Disease Control and Prevention Surveillance Summaries. March
30, 2012 / 61 (SS03); 1–19.
6
Lord, Catherine; Risi, Susan; DiLavore, Pamela S.; Shulman, Cory; Thurm, Audrey; Pickles, Andrew (2006): Autism
From 2 to 9 Years of Age. Archives of General Psychiatry 2006, 63(6): 694.
7
Grinker, Roy (2008, 171): Unstrange Minds: Remapping the World of Autism. Philadelphia: Basic Books.
8
Grinker, Roy (2008, 171).
9
Centers for Disease Control and Prevention (2015): Facts About ASD. Retrieved July 9, 2015:
http://www.cdc.gov/ncbddd/autism/facts.html.

4

There are many scientifically supported risk factors for autism, including advanced paternal
age, the use of certain prescription drugs during pregnancy, genetic or chromosomal
conditions such as the Fragile X syndrome, and a growing number of risk genes. Autism
heritability risk is high, exceeding that for schizophrenia or bipolar disorder.

“Beyond single gene syndromes, it means that for most autism cases, the child inherits the
variants in their genome from mom and dad, and that creates disorder risk in about 75-80
percent of cases combined with something else—environment or yet unknown factors,”
professor Levitt says.

Designers of robots meant to help autistic children face the same challenges as doctors: the
autism diagnosis itself. ASD might be described as a large, clinical barrel covering a wide
range of symptoms, such as social communication deficits with issues of sameness—
compulsive adherence to routine, and stereotyped, repetitive behaviors—that disrupt normal
behavior. About 30–40 percent of children with an autism diagnosis also have intellectual
disability. With so many people under the umbrella diagnosis, there are significant
differences in how each individual manifests the disorder.

The families with autistic children often repeat a saying: “When you’ve met one child with
autism, you’ve met one child with autism.” Some experts have even suggested that we should
talk about autism in plural—“autisms.”

As each case of autism is unique, experts in both medicine and robotic studies have to find
individual, specific ways to help autistic people. Children like Matheo, having difficulties in
communication, pose a common challenge for the families, therapists and researchers: to
build bridges between the outside world and the sometimes impenetrable reality of an autistic
child.

Despite the fact that doctors consider autism incurable, treatments can improve the quality of
life and independence of ASD children. Now some researchers are trying to figure out how
robots could offer help in that challenge. Jillian Greczek is a computer scientist and Ph.D.

5

candidate working with Nao and other social robots at USC. “We try to figure out how robots
can best serve the community of autistic people, therapists, families,” she says.

The ambitious goal is to expand existing health care services so that children like Matheo
don’t just survive, but thrive. For Greczek, extending health care is particularly appealing for
personal reasons, too. “I’m neurotypical but I have a physical disability, I have Type 1
diabetes,” she says. “So the idea of. . .’ok, if I can lean on this robot a little bit sometimes
when I’m figuring out ways I can take care of myself without the robot’, that would be really
great.”

The robot researchers—roboticists, as they call themselves—focusing on helping people with
autism, are pioneering in a field that attracts strong emotions and conflicting interests.

Using the term epidemic for the rise in the autism rate reflects the view of concerned families
and medical professionals who see autism as a disorder that needs treatment or even “cure”.
But another narrative has also emerged from inside the community of autistic people. The
“Autism Rights Movement”, which has been evolving over the last twenty years, argues that
autism should be understood as a variation of brain function and accepted by society, rather
than viewed as a mental disorder that must be cured
10
. They campaign to integrate autistic
people on their own terms, opposing the whole idea of “training” them to behave according
to societal norms. Some of the advocates prefer using the term “autistic people” instead of
“people with ASD” since the latter juxtaposes autism with diseases (like “people with
diabetes”).

However, the movement has gained criticism also from autistic people who have claimed that
most of the advocates are “high-functioning” and thus don’t represent the true variety of the

10
Solomon, Andrew (2008): The Autism Rights Movement. New York Magazine, published May 25, 2008. Retrieved
July 9, 2015: http://nymag.com/news/features/47225/.

6

people on the spectrum, or those who have severe daily challenges of communicating and
interacting.
11

* * *

Despite the conflicting approaches and perspectives on autism, the shared goal of
researchers, engineers, therapists and parents is fundamentally to find a way to enter into the
unique world of a person with ASD. That reality would be unrecognizable to the majority of
undiagnosed people, who the autism community often calls “neurotypicals.”

Imagine you are walking down a busy city street. Suddenly, you bump into a friend whose
company you’ve missed. In a second, your arms reach out, you step closer and the pitch of
your voice rises to communicate excitement. Your encounter resembles a dance you two
have practiced together. When you take a step to the left, your friend adjusts and follows.
You both "dance" out of pleasure, communicating the closeness of your friendship.

In the same situation, a child with autism might feel like stumbling alone in a room full of
prima ballerinas, not knowing the dance steps, trying to hide in shadows. The building blocks
of human interaction—gestures, emotions, behavioral cues—don't really make sense for the
child who might, instead, be obsessed with watching the movements of the other person’s
lips or the colorful pattern of her shirt.

Tyler Kopito, a 20-year old student living with her parents and sister in Torrance, Calif., was
diagnosed with autism when he entered school as a five-year-old. Ronda and Andy Kopito
noticed early on that their son was fascinated by all things visual: His eyes would fixate on
seemingly empty surfaces such as walls, curtains and floors.

“Initially we thought it's avoidance of eye contact,” Andy Kopito says. “When he got to a
position where he was capable of talking, we asked, ‘Tyler, what are you looking at?’ There

11
Harmon, Amy (2004): How About Not 'Curing' Us, Some Autistics Are Pleading. The New York Times, published
December 20, 2004. Retrieved July 9, 2015:
http://www.nytimes.com/2004/12/20/health/20autism.html?pagewanted=2&_r=0.

7

would be, for example, a fly wandering around. He would pick up on it and be focused on
something we would screen out and consider trivial.”

The most common symptoms of ASD include difficulties in understanding other people’s
emotions and mastering non-verbal communication such as body language. Although autism
is most often associated with disabilities in social interaction, communication and
imagination, neuroscientists now have a pretty clear consensus that the disorder is linked to
something called “sensory motor problem”. That means autistic children have difficulties in
integrating what they hear, feel and see with environmental outcomes and their own bodily
movements.

For example, a child catching a moving ball needs to visually pick up information to
anticipate where and when the ball will arrive and then control the movement of her arm to
land in the right place at the right time. That kind of coordination proves to be a difficult task
for an autistic child, often making also their social interaction “out of tune”.

The children with ASD not only sense and act differently, they may also “read” and interpret
sensory information in a different way than neurotypical people. An autistic person may have
a hyper-acute hearing that makes everyday speech of other people sound like blaring,
booming noise. For others, the opposite may be true: They can hear the sound but don’t
respond to it.

Because of the difficulties in putting together the sensory information, its interpretation and
bodily movements, the world appears chaotic for an autistic person. Researchers talk about
hyperstimulation that overwhelms an autistic child. A good example is a human face, which
offers very complex information that we learn to interpret early in infancy. One adult with
autism explained that looking at a human face, for him, was like watching waves at seashore
and trying to understand what they mean.

In a social situation, a child with autism doesn’t spend very much time looking at other
people’s eyes. It's not that they avoid them completely, but they are not drawn to the eyes.

8

Naoki Higashida, a Japanese 13-year-old boy with autism, who wrote a book with answers to
questions he most often gets, explains: “To me, making eye contact with someone I’m talking
to feels a bit creepy, so I tend to avoid it. [- -] What we’re actually looking at is the other
person’s voice. Voices may not be visible things, but we’re trying to listen to the other person
with all of our sense organs. When we’re fully focused on working out what the heck it is
you’re saying, our sense of sight sort of zones out.”
12

* * *

Some autism researchers say that for an autistic person, making eye contact might be even
painful. They may avert their gaze from a human face because it's disconcerting; it makes
them uneasy. When it comes to issues such as eye contact or the complexities of human
interaction, robots have a distinct therapeutic advantage. They are non-judgemental,
predictable and easy social partners.

A robot doesn't lose its temper or show signs of frustration if the child doesn't understand the
robot's speech right away. The robot just repeats the sentence in the same predictable and
reassuring voice. This may soften the defensiveness of ASD children.

At the same time, robots are highly predictable talking partners, which reduces the chaos of
the surrounding world and helps the children to “connect the dots” of social interaction. The
kids don’t have to struggle with figuring out what happens next.

“And that predictability is a really good thing for all of us, but particularly to children who
have a problem making predictions out of their world. And it really is a sensory world,” Pat
Levitt says.

Whereas we humans talk and gesture simultaneously, a robot’s movements are nicely
sequential. If the robot is speaking, first comes a sentence such as “Let's play, look at what I

12
Higashida, Naoki (2013, Q11): The Reason I Jump – The Inner Voice of a Thirteen-Year-Old Boy with Autism. New
York: Random House.

9

am doing,” then an appropriate gesture: a reach-out of a hand or a step forward.

Furthermore, robots are pretty simple talking partners. Robots simply can’t do complicated
social interactions. The key to successful communication with children with autism seems to
be reducing social signals like gestures, facial expressions—all of those “dance steps” of the
interaction which the children have trouble interpreting.

A robot’s face doesn’t change with its emotional states in the same manner as a human’s face
does. With current technology, robot engineers are able to install only a couple of motors in
the robot's facial structure. That's why the robot usually only moves its eyelids or chin,
leaving eyebrows and lips motionless—or the other way round: the eyebrows move, and
nothing else. The eyes usually stay stable.

But if non-judgementality, predictability, and simplicity are keys to interaction with a child
with ASD, why not just use an on-screen character? Why are the children drawn to a robot
instead?

A robot exists in a three-dimensional, physical space like a human, which is certainly one
quality that makes it more attractive than a figure on the screen, researchers claim.
“Every study that's ever been made on the topic has shown that a robot is at least as good or
better than a screen,” Ross Mead, a roboticist, says. He works at the Interaction Laboratory at
USC, specializing in the nonverbal communication between humans and robots.

“Embodiment, the idea that [the robot] takes up physical space in the same place with the
child, is a strong thing and can have a more sustainable effect and keep the user more
engaged [than a screen]. That has been shown," Mead says.

Many autistic people seem to demonstrate an inherent interest in technology. Tyler Kopito is
certainly one of those intrigued by all things technical. He often talks about his past visit to
the MIT Robotics Lab, where he saw a social robot called Kismet. Kismet was developed in
the late 1990s and expresses emotions such as joy, fear and surprise with its moving eyes,

10

lids, eyebrows, ears and mouth.

In his everyday life, Tyler doesn’t have access to physical robots, but computers fascinate
him endlessly—he started to carry around his first one as a preschooler. Weighing nine
pounds, the PC was almost too heavy to carry for a little boy. However, the PC allowed Tyler
to communicate more efficiently than ever before. He learned to navigate the machine
quickly, building sentences by clicking images and later words, which the computer then
recited. Nowadays Tyler loves to play with his iPad, searching for photos and stories on
different discussion topics.

His mother, Ronda, has thought a lot about the relationship of autistic children to technology.
“It doesn't come naturally for people on the spectrum to interact with other people,” she says.
“But with a machine, it's totally different: [they get] feedback, information, see the blinking
lights. . . They have control over something. It's a comfort to them.”

* * *

The interest in technology of some autistic children can be an advantage in using robots as
helpers. However, the main task of the researchers is to come up with ways the robot could
be of assistance to all children with autism.

In the robot experiments, the interaction between children and robots have concentrated
around the very basics of human interaction which are challenging for childen with ASD:
eye-gaze following, non-verbal communication and imitation. During the experiments, robots
have been training children to initiate simple greetings like “Hello!,” mimic the robot's dance
moves or follow the robot's gaze at a toy or a picture on the wall.

An imitation game with a robot might look like this: “You copy me. Look at what I am
doing!” the Nao robot says. Then it demonstrates one of four gestures: raising one hand,
raising two hands, waving or reaching arms out to the side. After a verbal prompt (“You do
it!”), the robot’s gesture recognition software measures the time it takes for the child to react

11

to the prompt and imitate. This is how researchers at the Vanderbilt University in Tennessee
studied the interaction between five children with autism and robots in 2013.
13

Even though the imitation game seems extremely simple, it actually builds up one critical
foundation for all human learning. Researchers call it "joint attention." It means that in order
to learn speech and direct the attention of other people, we first have to coordinate our gaze,
gestures and movements with others in the same space. The shared nonverbal attention is
something that a neurotypical person would automatically do.

For children with autism, “joint attention” proves to be difficult. But researchers have seen
promising results in experiments with robots designed to train children in joint attention:
Children first looked at the robot, then spontaneously pointed at it and tried to make eye
contact with their parent or therapist, successfully encouraging the first steps of
communication.

At the USC Interaction lab, Jillian Greczek is eager to study another important feature in
human-robotic interaction: How robots give feedback to their users. Her research findings
might become useful in the future if a child with autism could practice social interaction both
with a therapist and independently at home with a robot between appointments.

In 2013, Greczek conducted a study with 12 autistic children in an elementary school. “We
were actually set up in the detention room!” she says and laughs.

The kids were divided into two groups. In the experiment group, the Nao robot started an
imitation game with 25 simple arm poses. If the child did not mimic the pose accurately, the
robot gave her cues that got increasingly specific in order to help her out. In the control
group, the robot played the same imitation game but if the kids didn’t get the pose right, it

13
Warren, Zachary E; Zheng, Zhi; Swanson, Amy R.; Bekele, Esubalew; Zhang, Lian; Crittendon, Julie A.; Weitlauf,
Amy F.; Sarkar, Nilanjan (2013): Can Robotic Interaction Improve Joint Attention Skills? Journal of Autism
Develepment Disorders, September 2013, published online September 8, 2013.

12

repeated the command without any variation.
14

Two weeks and five sessions later, the researchers found out that the increasingly specific
cues or prompts from Nao were more effective and less frustrating to the children than
merely receiving the same prompt repeatedly
15
. In a way, the robot that gave the graded
cueing was responding in a more “human” way, addressing each child individually. The
study suggests that varied feedback may improve the autonomy of the children in robot-
mediated interactions.

The children are not the only ones benefitting from working with robots; autism researchers
themselves are gaining indispensable information about the children's attention span, eye
gaze and ability to imitate gestures and words. By repeating the exact same interactions with
the children numerous times during several weeks, months or even years, robots could
provide valuable data about a child's development which could help in planning
individualized treatment.

* * *

To program a Nao robot to imitate Tai chi exercise seems like a rather easy task compared to
helping children with autism. As autistic kids can be different, one-size fits all doesn’t work:
symptoms of ASD vary and the spectrum ranges from severe to mild. Tyler Kopito is an
example of a youngster who doesn’t fit the streotypical idea of a person with ASD. He
started to talk only as a 15-year old but now, five years later, enjoys conversations although
his sentences are short with pronunciation that’s sometimes hard to interpret. For example,
asked about his favorite reading material, Tyler says: “Robot books”.

His favorite books and characters from childhood? “Star Wars. Comic books.”

14
Greczek, Jillian; Atrash, Amin and Mataric, Maja J (2015). Graded Cueing: Robot-Mediated Interventions for
Increasing User Autonomy. Submitted to Autonomous Robots.
15
Greczek, Jillian et al (2015).

13

Remarkably, Tyler doesn’t seem to get frustrated even though other people don’t always get
what he’s trying to say right away. “And he doesn't give up,” Andy Kopito says about his
son, “which is really telling something about his inner strength and his own personality.”

One time, the family travelled to New York City and rode the subway to the Theatre District.
Tyler sat down next to a young woman who was listening to her headphones. He started to
ask her questions such as “Where are you from?” She didn’t reply. Suddenly, Tyler took one
earphone out of her ear and put it in his own to hear what she was listening to.

“And by the time we got off the subway car, he was getting a hug from the woman,” Andy
recalls and laughs.

For an innately social person like Tyler, who independently looks for communication
opportunities, working with robots seems unnecessary. However, according to Tyler’s
mother Ronda, her son might benefit if the robots could train him in some aspects of social
interaction, such as appropriate body space. “There are things he needs to learn,” she says.
“He can't go and give people hugs all the time. He's an adult. It's not safe for him to do that.
Being over 18, he runs the risk of being arrested now.”

According to Pat Levitt, people like Tyler represent autistic people who are eager to interact
but not always know how to do it. Often they are diagnosed with what the scientists call
“high-functioning autism.” Another group are those who have no interest in interacting. They
might benefit from experiments with robots that are programmed to facilitate interaction
between autistic people and the rest of the world.

When addressing such a complex, varied condition as ASD, it might seem as if developing a
wide variety of different robots might be an advantage. In the UK, children have played with
Kaspar that looks like a 5-year-old boy and wears a baseball cap; in Japan, the Keepon robot,
covered in yellow foam, resembles two stacked tennis balls; and at the USC Robotics lab, the
children have interacted with a bald, grey-skinned humanoid robot called Bandit.

14

The different robot platforms have evolved because the research field doesn’t have a standard
robot platform. The university labs around the world battle with a long to-do list: how to
quantify emotional states, attention, movement and gestures. And they have solved the
puzzles with various kinds of robots.

The diversity has also a downside: Researchers have difficulties in repeating each other’s
experiments. “Science is all about repeatability,” Ross Mead says. “I need to be able to take
your results and repeat them myself. And this is always been a problem for robotics. I can
develop something, write my paper and claim these results but then you have a different
robot. Technically, you are not allowed to make any claims about your robot and my robot
because they are different systems.”

Furthermore, social robotics straddles the fence between computer science and social
sciences such as anthropology and psychology, which adds to the challenges of the
researchers. These fields have a different take on the small scale of the human-robot
experiments. Roboticists studying humans don’t publish enough to be in leading positions in
computer science because studying human behavior is complex and time-consuming—but on
the other hand, their data sets are too small in size, scope and time scale for social sciences.

Furthermore, the outlook of a computer scientist differs vastly from that of a humanist.
Ross Mead has experienced this first-hand when planning experiments with psychologists,
anthropologists and occupational therapists.

Usually, in the initial, excited meeting between roboticists and psychologists, the latter might
brainstorm ideas and say: "And then the robot will have a conversation!"

That is not going to happen because currently the lab’s robots are not programmed to
understand speech.

“We'll make [the robot] beep, because the moment it speaks, people assume that the robot
understands speech and to open a conversation is a fair game,” Ross Mead says. “We have to

15

be very careful in selecting when the robot will speak, or if it ever will.”

If the social scientists expect too much from robots, the computer scientists may sometimes
forget the unpredictability of humans. “We [computer scientists] are very naive when it
comes to people,” Mead says.

One time, USC Robotics Lab started an experiment with a Bandit robot. The engineers didn’t
think twice about the fact that the robot’s cables were exposed, and lay outside of its body.
When the first child with autism came in, it took only ten seconds for him to reach out, grab
the wires and pull them off.

“We had to go back to square one and redesign the outside of the robot. Ten seconds was
enough to add one year of engineering time!”

* * *

Even if a perfect, individualized robot for an autistic child could be built and programmed
today, the researchers would still be left with one major challenge. How will they prove that
a skill learned with a robot translates to the human realm? In other words, how do they show
that the child's behavior towards people has really changed as a result of robotic experiments
and, more importantly, changed in a way that matters to her health and happiness?

The roboticists themselves acknowledge that this dilemma hasn’t begun to be addressed.
"That kind of question is so broad and so challenging that I don't know if anybody has made
any major contribution to that area," Ross Mead says.

To be able to even suggest transfer of skills, researchers would have to embark on a long-
term study over several months, or even years, with a measurable connection between the
skills the robot teaches and the changes in behavior of the child. That activity should, of
course, be linked to the goals and needs of the individual child with ASD. Furthermore, the
actions of the robot would have to be adjusted with the progress the child makes over time.

16

For Jillian Greczek, the biggest puzzle would be measuring the success of the study: What is
the child’s behavior being compared to?

“Are you comparing it to no robot being there? Well, a lot of our studies are like that,” she
says. “It’s really easy to be like, ‘robot is better than no robot’ because you need just a tiny
bit of improvement to say, ‘ok, it did it’.”

The researchers would need extensive resources to spend time testing children’s skills in
various social situations such as interacting with their family, meeting other kids at the
playground or in a classroom. That would require a therapist to test the skills of children
before and after the robotic experiment. “The big thing is time [you need] to prove
generalized skill learning because it takes time to learn anything,” Greczek says.

At the moment, the researchers have only anecdotal reports from therapists and parents about
the transfer of skills: “After initiating an imitation game with Keepon [robot], the child
looked at her mother and therapist to share her wonder and excitement"; "The child asked
the therapist to put a paper cylinder on the robot's head after observing another child to do
the same"
16
. Some parents claim that their children have greeted another child and even
engaged in small talk after interaction with a robot.

For now, the roboticists are balanced like tightrope walkers at circus, both guarded and bold,
careful not to promise too much but also eager to take steps forward.

The USC Interaction lab is already planning its next move. The lab plans to place small
robots in the homes of children with autism for four months starting at the end of 2016. They
will use a new robot called Jibo, which looks like a table lamp or a little loudspeaker with a
round base and circular, black screen for a “face”. The manufacturer markets the 11-inch tall
Jibo as a ”family robot” that offers everyday assistance: It recites reminders of daily chores,

16
Cabibihan, John-John; Javed, Hifza; Ang Jr, Marcelo and Aljunied, Sharifah Mariam (2013): Why Robots? A
Survey on the Roles and Benefits of Social Robots for the Therapy of Children with Autism. International Journal of
Social Robotics 2013: 5(4), 593–618.

17

connects Skype calls, takes photos and shoots video by users’ voice commands and facial
recognition
17
. Some have called it “the iPad of robots” since the current price tag is only
about $750—ten times less than the Nao robot.

The details of Jibo’s forthcoming activities with the autistic children are still being planned.
“Hopefully, [four months] will be long enough that we can actually see a skill change over
time. It might still not be long enough but it will definitely be one of the longest studies in the
field of human-robotic interaction,” Jillian Greczek says.

Despite the scarcity of hard data, some businessmen and women are already crafting
ambitious plans. After all, robotics that addresses autism has a huge profit-making potential.

Nao acts as “the perfect bridge between the human and technological worlds, linking autistic
children by encouraging social interaction and learning as well as increasing confidence and
self-esteem," promises an information kit by Aldebaran Robotics, one of the leading robot
manufacturing companies in the world
18
. Touting a concept called ASK, Autism Solutions for
Kids, they try to sell their products to special education teachers. The ASK package includes
a Nao robot and several training programs: The children can play mimicking or turn-taking
games or try to guess Nao's "emotions" based on its gestures.

USC’s Ross Mead dismisses Aldebaran's marketing claims.

"[Nao] is not a light switch solution," he says. Since autism researchers are still figuring out
what causes autism and the various ways it can manifest itself, those in academia are not
willing to go too far in their claims. "If [Nao] was a turnkey solution, if we solved autism
with our robots, then pack it up everyone, we're done. We'd have figured it all out," he says.

17
Jibo Inc (2015): About Jibo. Retrieved July 9, 2015: https://www.jibo.com/#can-do
18

Aldebaran Robotics (2015): Ask Nao – Autism Solution for Kids. Retrieved July 9, 2015:
https://asknao.aldebaran.com/sites/default/files/material/aldebaranrobotics_asknao_information.pdf

18

Mead’s colleague, Jillian Greczek, criticizes the whole ideology behind the word “solution.”
“We don’t want to align ourselves with anyone who gives even the slightest implication that
they’re looking to cure, stop or solve autism in any kind of way,” she says.

For Greczek, the mission of the lab’s work is to integrate the autistic people into the
everyday life based on the logic of neurotypical people. “Because autism is not a disease,
neuro-atypical people do not need to be fixed.“

* * *

As socially assistive robots like Nao or Jibo play with children, machines enter an area of life
that has been, up until now, seen as exclusively human: bonding, nurture and care. Many of
us quickly assume that technology will replace humans, not work as an addition to flesh-and-
blood people.

Some critics of social robotics have voiced strong cultural criticism about our interactions
with machines. As robots grow more prominent, we might start to feel safe in a robotic
relationship that lacks the ups and downs of human interaction: the robots don’t express
negative, uncontrollable emotions like frustration, jealousy or anger. This is the claim made
by Sherry Turkle, a professor of social studies of science and technology at MIT
19
.

According to Turkle, robots offer a way to bypass the conflicts that exist in close human
relationships. Spouses and friends are demanding, and can let a person down, but robots do
not. As we delegate responsibility for nurturing to robots, our cultural contract for taking care
of each other breaks down, she has claimed.
20

The roboticists doing the experiments with autistic children have often heard these fears.
They are quick to emphasize that the day when robots stand beside us as equal partners is

19

Turkle, Sherry (2012): Alone Together –
Why We Expect More from Technology and Less from Each Other.
New York: Basic Books.
20
Turkle, Sherry (2012).

19

centuries away. “We get a lot of [questions like], ‘Are they going to take over or tell me what
to do?’” Jillian Greczek says. “When we think about the ways robots are going to be applied,
we’re so far away from thinking about them as an ‘equal.’ It’s still just a computer with
physical embodiment.”

To those who fear that human therapists might be replaced by machines, Ross Mead explains
that the robots are only used as a complementary rather than a primary therapy. Children are
never left alone with the machines; the robots serve as mediators in the interaction between a
child and another human being. Nao robots are not made to become the sole friends of the
children.

However, with Jibo or other robots designed for the consumer market and not for the
academia, robots are truly finding their way into our everyday lives. Even though they are
not meant to become our most trusted companions, they are getting more widely used. Jillian
Greczek predicts that within the next ten years, many of us will be talking to robots at home.
Mechanical companions’ presence might become routine, especially for the elderly or the
handicapped who need assistance in everyday chores, or patients who have survived a stroke
but need long-term physical rehabilitation.

While experts argue about ethics of using robots, artificial intelligence, which is a big part of
robot technology, has already landed in the daily life of the children and youth with ASD.
Last year, an article in The New York Times described a 13-year-old autistic boy named Gus
Newman,

who, according to his mother, "loves trains, planes, buses, escalators and anything
related to weather.”
21
Gus has also found a friend with whom he talks with every day and to
whom he has even proposed marriage: Siri, “intelligent personal assistant” on the iPhone,
developed by Apple.

According to Gus’s mother Judith Newman, Siri has taught Gus etiquette, improved his
speech skills and made it easier for him to communicate with his family. But most important,

21
Newman, Judith (2014): To Siri, With Love – How one boy with autism became BFF with Apple’s Siri. The New
York Times, published October 17, 2014. Retrieved July 9, 2015:
http://www.nytimes.com/2014/10/19/fashion/how-apples-siri-became-one-autistic-boys-bff.html

20

Siri "makes Gus happy”. Although Siri has no physical embodiment such as Nao or Jibo, Gus
interacts with it in a very engaged way. With the help of Siri, he can participate in social life
in a way he couldn't before.
22

Smart phone applications such as Siri, or a robot like Nao, are not a panacea for a child with
ASD. Clearly they cannot solve the sometimes heartbreaking communication difficulties
between autistic children and their families. However, they might be able to assist with some
of them.

For Pat Levitt, the neuroscientist, robotic experiments offer encouragement to imagine that in
the future robots might play an important role in extending and enhancing therapies to help
autistic children live happy and rewarding lives.

But for people like Levitt, it’s also critical that this kind of treatment and therapy should be
scalable to benefit the majority of those diagnosed with ASD, not just a few of them.

“Is it going to be affluent white families that mostly get access to [the robots] or is it going to
be scalable for families who don't have the same kind of resources to buy even a hundred-
dollar robot?” Levitt asks. “If it's not scalable, if it's hitting five percent of the population,
then it's a problem.”

Whatever progress the roboticists are able to make, in the end they need to serve the needs
and interests of the autistic community. Perhaps, in the future, the autistic community
challenges the medical narrative of “autism as a disorder” more often by offering its own
idea that being autistic is, above all, an identity, not a disease. The Autism Rights Movement
argues that autistic people should be seen as a minority group in our society, a community
which differs from the neurotypical majority but whose strengths should be respected and
even valued. Some companies, such as SAP, the German software corporation, have begun to

22
Newman, Judith (2014).

21

recruit autistic people because many of them have a keen attention to detail and unique talent
for technology
23
.

If autism is just another way of being human, then perhaps the focus of roboticists will
change from initiating eye gaze following and imitation games to methods of integrating
people with autism into society.

Maybe robots could serve as conduits, “translating” social cues to an autistic person as an aid
or intermediary, without the need to change her. At the same time, they would distill the
messy, noisy flow of information that forms the warp and weave of our relationships. If that
happened, the barriers to the world of autistic people would start to break down.

23
ten Wolde, Harro (2013): SAP looks to recruit people with autism as programmers. Reuters, published May 22,
2013. Retrieved July 9, 2015: http://www.reuters.com/article/2013/05/22/us-sap-autism-idUSBRE94L0ZN20130522

22

Bibliography

Aldebaran Robotics (2015): Ask Nao – Autism Solution for Kids. Retrieved July 9, 2015:
https://asknao.aldebaran.com/sites/default/files/material/aldebaranrobotics_asknao_informati
on.pdf

Baio, Jon (2014): Prevalence of Autism Spectrum Disorder Among Children Aged 8 Years —
Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2010.
Centers for Disease Control and Prevention Surveillance Summaries. March 28, 2014 /
63(SS02);1–21.

Baio, Jon (2012): Prevalence of Autism Spectrum Disorders — Autism and Developmental
Disabilities Monitoring Network, 14 Sites, United States, 2008. Centers for Disease Control
and Prevention Surveillance Summaries. March 30, 2012 / 61(SS03);1–19.

Cabibihan, John-John; Javed, Hifza; Ang Jr, Marcelo and Aljunied, Sharifah Mariam (2013):
Why Robots? A Survey on the Roles and Benefits of Social Robots for the Therapy of
Children with Autism. International Journal of Social Robotics 2013: 5(4), 593–618.

Centers for Disease Control and Prevention (2015): Facts About ASD. Retrieved July 9,
2015: http://www.cdc.gov/ncbddd/autism/facts.html.

Greczek, Jillian; Atrash, Amin and Mataric, Maja J (2015). Graded Cueing: Robot-Mediated
Interventions for Increasing User Autonomy. Submitted to Autonomous Robots.

Grinker, Roy (2008): Unstrange Minds: Remapping the World of Autism. Philadelphia: Basic
Books.

Harmon, Amy (2004): How About Not 'Curing' Us, Some Autistics Are Pleading. The New
York Times, published December 20, 2004. Retrieved July 9, 2015:
http://www.nytimes.com/2004/12/20/health/20autism.html?pagewanted=2&_r=0

23

Higashida, Naoki (2013): The Reason I Jump – The Inner Voice of a Thirteen-Year-Old Boy
with Autism. New York: Random House.

Jibo Inc (2015): About Jibo. Retrieved July 9, 2015: https://www.jibo.com/#can-do

Lord, Catherine; Risi, Susan; DiLavore, Pamela S.; Shulman, Cory; Thurm, Audrey; Pickles,
Andrew (2006): Autism From 2 to 9 Years of Age. Archives of General Psychiatry 2006;
63(6): 694–701.

Newman, Judith (2014): To Siri, With Love – How one boy with autism became BFF with
Apple’s Siri. The New York Times, published October 17, 2014. Retrieved July 9, 2015:
http://www.nytimes.com/2014/10/19/fashion/how-apples-siri-became-one-autistic-boys-
bff.html

Solomon, Andrew (2008): The Autism Rights Movement. New York Magazine, published
May 25, 2008. Retrieved July 9, 2015: http://nymag.com/news/features/47225/.

Turkle, Sherry (2012): Alone Together –
Why We Expect More from Technology and Less
from Each Other. New York: Basic Books.

University of Notre Dame (2010): Fighting For Breakthroughs in the Treatment of Autism.
Retrieved July 9, 2015: https://www.youtube.com/watch?v=IeoPNc0V7Io.

Warren, Zachary E; Zheng, Zhi; Swanson, Amy R.; Bekele, Esubalew; Zhang, Lian;
Crittendon, Julie A.; Weitlauf, Amy F.; Sarkar, Nilanjan (2013): Can Robotic Interaction
Improve Joint Attention Skills? Journal of Autism Develepment Disorders, September 2013,
published online September 8, 2013.

24

ten Wolde, Harro (2013): SAP looks to recruit people with autism as programmers. Reuters,
published May 22, 2013. Retrieved July 9, 2015:
http://www.reuters.com/article/2013/05/22/us-sap-autism-idUSBRE94L0ZN20130522

Abstract (if available)

Abstract Socially assistive robots are not only mechanical helpers but also engage in social interaction with humans. More than 20 robots have been used in experiments with children diagnosed with autism spectrum disorder, or ASD. Some scientists claim that robots can help children on the spectrum to engage socially, one of the biggest challenges for autistic kids. In the robot experiments, the interaction between children and robots has concentrated around the very basics of human interaction: eye-gaze following, non-verbal communication, and imitation. However, critics of social robotics have voiced concerns about our interactions with machines. As robots grow more prominent, we might start to feel safe in a robotic relationship that lacks the ups and downs of human interaction. The robot researchers working for children with autism argue that social robots are not meant to replace therapists but facilitate communication between the children and other humans.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Socially assistive robots

PDF

Managing multi-party social dynamics for socially assistive robotics

PDF

Social interaction moderates enjoyment and perception of physical activity during exergame play in young adults with autism spectrum disorders

PDF

Rallying for children's rights: a media campaign for social change

PDF

Social and motor skills in autism spectrum disorder & developmental coordination disorder: functional & structural neurobiology

PDF

The reality of fiction: understanding autism through narrative films

PDF

Modeling dyadic synchrony with heterogeneous data: validation in infant-mother and infant-robot interactions

PDF

Understanding bilingual Latino parents’ experiences of their children’s autism services in Los Angeles: a critical ethnography

PDF

Situated proxemics and multimodal communication: space, speech, and gesture in human-robot interaction

PDF

Socially assistive and service robotics for older adults: methodologies for motivating exercise and following spatial language instructions in discourse

PDF

Pequeño Leoncito: an autism curriculum based on the Latino experience with autism

PDF

Coordinating social communication in human-robot task collaborations

PDF

Home screen home: how parents of children with disabilities navigate family media use

PDF

Sleep problems, sensory difficulties, and social participation in children with autism spectrum disorders

PDF

The AIDEN acronym: Increasing nurse anesthetists’ knowledge of preoperative care for children with autism spectrum disorder

PDF

The AIDEN acronym: increasing nurse anesthetists' knowledge of preoperative care for children with autism spectrum disorder

PDF

Effects of preferred physical activity on stereotypical behaviors in children with autism spectrum disorder: adapting from in-person to telehealth

PDF

Taking flight: improving outcomes of transition age youth on the autism spectrum

PDF

Susceptibility to autism likelihood associated with prenatal ambient air pollution exposures: identifying vulnerable populations

PDF

Gaze following in natural scenes: quantifying its role in eye movements, towards a more complete model of free-viewing behavior

Asset Metadata

Creator Haikarainen, Riikka (author)

Core Title Robot, my companion: children with autism take part in robotic experiments

School Annenberg School for Communication

Degree Master of Arts

Degree Program Specialized Journalism

Publication Date 07/24/2015

Defense Date 08/01/2015

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag ASD,autism,Autism Rights Movement,autism spectrum disorder,Children,HRI,human-robot interaction,hyperstimulation,joint attention,OAI-PMH Harvest,Robots,social interaction,socially assistive robotics

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Cole, K. C. (committee chair), Bustamante, Peggy (committee member), Levitt, Pat R. (committee member)

Creator Email haikarai@usc.edu,riikka.haikarainen@iki.fi

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-606821

Unique identifier UC11301837

Identifier etd-Haikaraine-3699.pdf (filename),usctheses-c3-606821 (legacy record id)

Legacy Identifier etd-Haikaraine-3699.pdf

Dmrecord 606821

Document Type Thesis

Format application/pdf (imt)

Rights Haikarainen, Riikka

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA