• What are social robots?
• Robots that evoke a social response from people
• Robots that correctly respond to social cues
• Social robots are not service robots
• They will not make you a cup of tea or clean the house
• Most social robots are not intelligent (yet)
• Artificial Intelligence is not mature enough to make
Social robots and healthcare
• Variety of potential healthcare
• Provision of health education
• Supporting communication
between patients and
• Compliance and persuasion
• Providing entertainment for
Physical versus on-screen robot
• Study in which children play a quiz.
• Two conditions: physical robot or on-screen robot.
• Attention and number of fixations significantly higher.
Robot Quiz screen Other
On screen robot
Looking duration in 10 min. session
(Belpaeme et al., 2013, Journal of HRI)
• Adult participants solve a puzzle and get lessons on how to:
when the lesson is provided by a social robot, people
become twice as fast in solving the puzzle (Leyzberg et al. 2014)
Leyzberg, D., Spaulding, S. and Scassellati, B. (2014) Personalizing robot tutors. HRI2014.
Robot tutoring - children
• Robots can also teach children new knowledge
• No difference in performance between an on-screen
robot and a real robot
Kennedy, J., Baxter, P., Belpaeme, T. (submitted) Children InteractingWith a Robot Tutor Overcome Learning Bias.
• Children are more accepting of the robot’s “flaws”, when the robot
personalizes its responses.
• Trend towards more learning by the child.
• Children will align with the robot (use same language, same posture).
• Children quickly settle in a turn taking pattern with the robot.
Robot results in more compliance
• Keeping a diabetes diary is important for diabetes self-
• Learn about diabetes.
• Recognize patterns, (day, weekend, season).
• Link own behavior with glucose levels (apply knowledge).
• Personal robot stimulates use of diary and medical
• Children meet robot in person at hospital or at home
• They keep a diary for two weeks
• Robot helps filling in the diary through a video conferencing
• 6 children with diabetes (9-12 yrs old)
• Within subject design
• Survey of how do children experience filling in a diary with the robot
• Log of how do the children use the diary
• Observations of how do the children behave with the robot
• time spent on filling in diary (seconds)
• amount of feedback given (number of characters)
Filling in diary Characters in activity log (sig. p = 0.09)
• Case study: 8 y. old, cerebral stroke
• Lost mobility in right leg, arm and partial mobility in
• Patient at the Neurology Department
• No specific paediatric facilities
• Rehabilitation long and frustrating process for patient
• Medical staff suggested use of robot
• Robot used for motivation and rehabilitation exercises.
• Used for 6 days, several times per day, with bespoke
activities running on the robot
• Spectrum of developmental disabilities:
Autism Spectrum Disorder (ASD)
• Deficits in social communication and interaction
• Presence of restricted, repetitive patterns of behaviour
• Varies substantially in the severity and nature of
• Affects 1 in 88 children in the United States
• 1% of adult population in the UK
• No cure, but early intervention can result in a
positive long-term outcome
Cent. Dis. Control Prev. (CDC). 2011. Autism spectrum disorders.
Therapy for ASD
• Behavioral and
- Computers (gaming and
virtual reality) and
- Social robots
Robots and autism
• Social robots have been used for autism therapy for
about a decade
• Robots seem to improve engagement and elicit
novel social behaviours from people (especially
children and teenagers) with autism
Scassellati et al. (2012) Robots for Use in Autism Research. Annual Review of Biomedical Engineering: 14: 275-294
• Minimally expressive
complexity of social
• Tele-operated by
• University of
• Huggable robot
• Green anteater: children
have no or very little
• Tele-operated by a hidden
• Vrije Universiteit Brussel
Why a robot?
• Controllable social responses
• Social responsiveness of the robot can be controlled,
e.g. exaggerated or switched off to avoid
• Robots behave in a consistent and predictable manner,
which seems reassuring for ASD children
• Robots can interact and play out social interaction
• Robots are consistent and potential to be a safe intervention
• Typical behavioural therapy requires min. 15h/week
• Human resources are not available
• Estimated to cost 3.2 million US$ over lifetime
• ASD children often have a high
level of motivation and involvement
when interacting with robots
• The robot acts as social mediator
Robot as catalyst
• n=17 + 11 TD controls, Ages 9-12 high functioning
(full‐scale IQ >= 70)
• Task: encourage a robot that is afraid of water to
Robot as catalyst
• Both groups show high
engagement and positive
• ASD group post play time
with therapist is longer, and
free play with robot is longer
• More face-to-face orientation
to the interviewer
• More relevant statements to
interviewer’s personal story
Kim, Berkovits, Bernier, Leyzberg, Shic, Paul and Scassellati (2012) Social Robots as Embedded Reinforcers of Social Behavior in Children
with Autism. Journal of Autism and Developmental Disorders.
Robot as tutor
• Learn to identify situation based emotion
• A = baseline phase
B = intervention phase
• A: no facial and
• B: social robot,
expressive face and
Robot walks and stumbles over a rock.
Someone takes lollipop off robot
Robot falls and breaks his toy.
Robot goes to shop to buy a lollipop.
Robot visits the zoo.
Robot plays games on the computer.
Robot as tutor
• Social stories
• 10 animations for each of the two basic emotions (clip fo
15 to 30 seconds)
• animations were played on robot's belly screen
• after each animation the child is asked of the robot
would be happy or sad, by pointing out smiling or
10 years of ASD/robot research
• ASD children are more responsive to feedback, even
social feedback, when administered via technology
rather than a human (Ozonoff, 1995)
• Robots are well-suited for play therapy (Francois et al., 2009,
Kim et al., 2012)
• Robots elicit interactions (Robins et al., 2009)
• More intrinsically interested in therapy when it involves
electronic or robotic components (Robins, 2006)
• Robot act as a social crutch, assisting children in the
transfer process(Robins et al., 2009, Scassellati, 2011)
• Facilitates joint attention episodes between a child and
an adult (Robins et al., 2004, Meltzoff et al., 2011)
… however, the story is subtle
• Increase but also decrease in social communication
when interacting with a robotic interlocutor in
comparison to a human interlocutor (Pioggia et al., 2008)
• Greater interest in a robot therapeutic partner than
a human, but better verbal and nonverbal imitation
performance in response to the human partner
(Duquette et al., 2010)
The limit of our knowledge
• Almost all evidence is N=1, or to a lesser extent anecdotal or
driven by commercial interest
• Given the “spectrum” nature of autism, it is unclear if
observations carry over to other children
• Robot-assisted therapy for autism needs larger scale
• Does it last, or does the response to robots wear off?
• What is the long-term beneficial effect of interacting
• Why do ASD children respond well to robots?
• The DREAM project
• Development of Robot-Enhanced therapy for children
with AutisM spectrum disorders
• Robot-enhanced therapy
• Assessment/data collection using robotic technology
• Supervised autonomy for therapeutic robots
• Building a cognitive model of the child (Theory of Mind)
• Real-world deployment
• Ethics of robots in therapy
• Address shortcomings of current robot-assisted
therapy research, i.e.
• Low participant numbers, lack of control conditions
• Unclear diagnostic and inclusion criteria
• Lack of quantitative measures, overreliance on anecdotal
• Move away from
• Autonomous robots instead
of human operated
• ALIZ-E and DREAM teams, esp. Ramona Simut and Bram Vanderborght
• Funded by FP7 ALIZ-E and FP7 DREAM projects
• Further reading
• Scassellati, Admoni, Mataric (2012) Robots for use in autism research. Annual Review
of Biomedical Engineering: 14:275-294
• Thill, Pop, Belpaeme, Ziemke and Vanderborght (2013) Robot-assisted therapy with
(partially) autonomous control: challenges and outlook. PALADYN Journal of
Behavioral Robotics: 3(4): 209-217
• Robins, Dautenhahn, Dickerson (2009) From isolation to communication: A case
study evaluation of robot assisted play for children with autism with a minimally
expressive humanoid robot. ACHI2009, Cancun, Mexico.