Childhood overweight and obesity represent an alarming problem worldwide. In Italy, approximately 21% of children are overweight or obese, with strong consequences both at health and social levels. The importance of the problem rises the need for policies aimed at preventing pediatric overweight and obesity. Game-based nutritional learning is an effective approach to enhance children's knowledge, behavior, and healthy dietary habits. Therefore, playful, and creative methodologies should be included in nutritional education programs for children. Among novel technologies, humanoid robots have a special appeal for educational purposes. In this paper, we present our first prototype in which the social robot Alpha Mini is used as a personal persuasive technology in the context of nutrition. In particular, in the context of the HERO project, we are developing an application with the main aim of using a combination of social robotics and serious games to support nutrition education and selfawareness trying to change wrong attitudes in children. To investigate the appropriateness of the approach a Wizard of Oz study was performed, showing interesting results in terms of engagement, motivational strength, trust, and recall of relevant information.
Childhood obesity is a growing problem in the world. More than 35 million children under the age of 5 and about 340 million children and adolescents aged 5-19 are overweight or obese. In the context of the HERO (Healthy Eating RObot) project, we want to investigate the effect of personal social robotics as a persuasive technology.
Persuasive Technologies (PT) are intentionally designed to change people’s attitudes or behaviors without coercion or deception, acting therefore upon users’ beliefs always in an atmosphere of free choice. According to their functional roles [1], PT can be categorized into tools, media, or social actors – or as multiple categories at once. As tools, PT can increase people’s ability to perform a target behavior by making it easier. As media, PT can use both interactivity and narrative to create persuasive experiences that support behavior change or attitude. PT can also function as social actors, indeed when perceived as social actors, computer products can leverage principles of social influence to motivate and persuade people [1]. In this context, we expect human-like social robots to have a positive persuasive effect by exploiting both social and conversational capabilities.
Social robots are physically embodied,
autonomous agents that communicate and interact
with humans on a social and emotional level. Due
to their ability to enable natural interaction, social
robots have a great potential for helping people in
their daily activities by providing assistance and
information possibly personalizing their behavior
to the user [
To this aim, we performed a preliminary study
on the appropriateness of the approach based on
the Wizard of Oz (WOZ) technique [
The HERO robot behavior will be
personalized to the child’s user profile and the
level of healthy eating acting as a motivating
technology. To achieve this aim, we are
developing the following functionalities (see
Figure 1 for an overview):
• Data gathering from different and
heterogeneous sources: besides wearables,
smartphones, and other sensors, the robot
itself, through its camera, microphones, and
sensors, will be used as a non-intrusive tool
to collect data about nutrition habits,
moodrelated behavior, and so on.
• Health behavior tracking: the child's
lifestyle will be monitored through dialog
and collected data. HERO will stimulate the
child to talk about the experience with food
using techniques typical of narrative
medicine. HERO will integrate connections
to sensors or will use its own camera for
food detection and intake estimation and
for tracking the physical activity
automatically. Moreover, sleeping will be
also monitored through sensors, since the
quality of sleep is related to obesity and
food intake.
• Engaging psychophysical coaching: HERO
will use appropriate personalized strategies
to increase awareness and motivation in
children through dialog and serious games.
• Emotion and Mood analysis: It is well
known that there is a strong relationship
between food and emotions. In particular,
emotional eating, defined as ‘eating in
response to emotions’[
To achieve these goals, HERO has to be
embodied in a robotic body that allows to
implement the above-mentioned functionalities
and, in addition, it should be not too expensive.
For this reason, we have decided to investigate the
potentiality of Ubtech Alpha Mini in this
application context [
The Alpha Mini robot is a humanoid small
portable robot (about 24 cm tall), that can capture
speech through its microphones, talk through its
speakers, move, and recognize faces, which
allows him to smoothly follow the person he’s
currently interacting with, and some objects. He is
fully programmable using the Android SDK. It
can be also programmed, at high level, using a
Scratch-type block interface with a mobile app
(Figure 2b-c) [
Besides the available functionalities, we are implementing the coaching module that starting from the analysis of user behavior and affective state (facial expressions, voice prosody, verbal component of the communication, gestures, and posture), will use a dialogue management module to handle the personalized motivational dialogue with the user.
To generate the personalized persuasion and
motivational strategy, we will adapt and use the
PORTIA model [
In previous work, we performed a study in the
healthy eating scenario, with the Pepper robot
providing healthy eating advice in two different
settings non-personalized vs. personalized
condition [
Before implementing all the functionalities to
handle interaction in the wild with children, we
decided to investigate whether the designed
approach was accepted by children. To this aim,
we simulated HERO behaviors using a Wizard of
OZ (WOZ) approach [
In order to help the wizard follow a systematic
interaction path, we created some scripts based on
Scratch [
(a) (b) (c)
The dialog was simple, and it was previously designed by a human nutrition coach: after the presentation and greeting phase, the robot explained the benefits of eating fruits and vegetables on health.
The study involved a group of 5 children (3 males and two females) from 6 to 10 years old. Each of them came in our lab in which the organizational issues followed the governmental/institutional regulations for public health to prevent the spread of COVID19.
After receiving the consent of the children’s parents, and a short explanation describing the purpose of the experiment, every child interacted with HERO embodied in the AlphaMini. Using the WOZ interface, the wizard, the same expert in nutrition who designed the dialog, previously trained on the use of the app, handled the dialog with the children. At the end of the interaction, each child filled out a simple post-test questionnaire aiming at assessing, besides an overall evaluation of the user experience with the robot (Q1-3), the easiness of understanding the content of the dialog (Q4), the motivational strength of the messages (Q5), the level of trust in the robot (Q6-7) on a scale from 1 to 5. Moreover, we tested the level of recall about information provided through the dialog (R1-2) where the child had to indicate whether the statement was true or false.
The statements in the questionnaire were the following:
Q1. I was able to interact with the robot Q2. Interacting with the robot was engaging Q3. The system had an adequate response time Q4. It was easy to understand what the robot was saying Q5. What the robot said to me motivated me to eat more fruit and vegetables during the day Q6. I think that what the robot was saying was true Q7. The information that the robot gave to you was reliable Recall Questions:
R1. According to what the robot told you, eating fruit and vegetable can: 1. boost immunity and improve the body's ability to fight diseases [True][False]. 2. aid in the proper function of the digestive system and prevent constipation [True][False]. 3. help in preventing cardiovascular diseases [True][False].
R2. Fruits and vegetables contain:
Fiber [True][False].
Vitamins [True][False].
Proteins [True][False].
We analyzed the results of the post-test
questionnaire data and, even if the number of
participants was small, the interaction with the
robot received an overall good evaluation in terms
of user experience (4 on average). Results in
terms of understanding the content of the dialog
were also good (4 on average). An interesting
result is the one related to the motivational
strength that was 4.6 on average maybe due also
to the novelty aspects related to the use of the
robot [
Our long-term research goal is to develop a Personal Persuasive Social Robot in domains such as nutrition, wellbeing, and health for helping people in changing their wrong attitudes and habits. In the HERO project, we are focusing on childhood obesity. To this aim, we are developing a nutrition coach to be embodied in the Alpha Mini Social Robot. As a preliminary investigation, we run a simple experiment in our lab based on the Wizard of Oz technique. Even if performed on a small number of children, results suggest that the robot was perceived positively in terms of user experience and motivational strength. In particular, all children believed what the robot was telling them. As far as the retention of the learned information is concerned, the children remembered most of the information conveyed during the dialog.
In a future experiment, we will better explore these results with a larger sample and more ad hoc measures, also by considering other behavioral measures. These preliminary results are encouraging us in continuing investigating on using social robots in this domain by comparing this technology to other media and also to humans (i.e. parents, teachers).
[1] B.J. Fogg. Computers as Persuasive Social Actors, Ubiquity, 10.1016/B978155860643- 2/50007-X. (2011).