This paper describes the use of UX Personas for defining a stable character for the Pepper robot. This approach generated from the first analysis of the social interactions that occurred between Pepper and a group of children with autism working with the robot in a therapeutic laboratory on autonomy.
The advent of the IoT offers the opportunity
for everybody and everything to be connected,
receiving and processing information in real time.
This generates a shift in production and
consumption patterns, changing the relations
between all the agents in the economic system.
According to a report of the World Economic
forum, it is one of the largest enablers for
responsible digital transformation and it will add
$14 trillion of economic value to the global
economy by 2030 [
In this smart space the user becomes the center
of informative systems made of ambient and
personal sensors, including social robots,
communicative tools, mobile and ubiquitous
computing devices, and assumes responsibilities
traditionally intended for developers. In this
context, social and humanoid robots will play an
important role: they can be seen as integrated sets
of sensors and actuators, thus IoT platforms will
make it easier to monitor and control them [
Personas are fictional characters, which the designer creates based upon her research in order to represent the different user types that might use service, product, site, or brand in a similar way. Creating personas helps the designer to understand users’ needs, experiences, behaviors and goals.
This paper describes the use of UX Personas for defining a stable character for the Pepper robot. This approach generated from the first analysis of the social interactions that occurred between Pepper and a group of children with autism working with the robot in a therapeutic laboratory on autonomy.
This paper is organized as follows: Section 2 describes the related work in the field, while Section 3 introduces our proposal, and Section 4 concludes the paper.
Personas, also known as “characters”, are an
archetype of the users a project is aimed at. It is a
useful tool for identifying the actions, behaviors,
needs, aspirations and difficulties of the users of a
particular service and are the result of the union of
various information collected during the study of
the target. Personas were conceived by Alan
Cooper in 1998 [
In parallel with this trend, inspiration for the
design of robotic characters has come from drama
studies. Since the seminal work of Laurel [
The development of personas in our project acknowledges the relevance of dramatic elements in the design of robot characters, combined with the use of HCI techniques and conducted in cooperation with a multi-disciplinary team which includes therapists and experts.
The Sugar, Salt & Pepper - Humanoid robotics for autism - research project focuses on the use of the Pepper robot in a therapeutic laboratory on autonomy that aims to promote functional acquisitions in highly functioning (Asperger) children with autism.
During the laboratory, started at the end of February 2021 in an apartment of the Paideia Foundation, we wanted to test the exchanges and interactions of children in rehabilitation contexts with the robot helping the operators. Another goal of the project was to provide young participants with a space for increasing their skills of mutual communication and socialization and strengthen the acquisition of strategies related to daily activities, such as the preparation of a snack or the management of homework, with the help of Pepper, configured as a highly motivating and engaging tool.
The weekly meetings lasted one hour and were led by a therapist. Each meeting had this structure: welcome in the apartment managed by the robot and the therapists; dialogue session with the robot on a predetermined topic (e.g., music, video games, etc.); moment of snack preparation; moment of post-snack dialogue and final feedback.
During the first sessions that were carried out, the
robot was totally devoid of a personality and
character of its own. The robot limited itself to
listening and answering the questions that were
asked, through pre-recorded sentences (and which
were updated from session to session) on its
program. From an HCI point of view, this rather
cumbersome method turned out to be totally
inefficient and counter productive; the young
participants immediately realized that the
questions that were asked were not the work of an
AI operation, but of manual input by the
operators. Most of them, having understood the
game, enjoyed asking goliardic questions, or to
which the robot was unable to answer, thus ending
up getting funny and senseless answers; in some
cases they even managed to send it on tilt.
For this reason, we decided to develop a
personality that could be given to the robot: this
was possible thanks to the analysis of the main
questions and interests that emerged during the
sessions with the children, as described in [
We believe that personality is one of the most important aspects in a conversation, because it identifies the character of the interlocutor, thus making it possible to choose which phrases to use to make it more effective and which ones not. In the following tables, it is possible to observe the implementation of the information necessary for the construction of the robot personality, approached in a Personas style. These are divided into: • Background information: physical appearance of the robot, its biography, personality and family ties (see Table 1); • Interests: knowledge, skills and tastes (Tables 2-3-4-5-6); • Goals: short and long term goals (Tables 7-8). It must obviously be taken into account that this personality has been made as simple as possible given the target users to which it interfaced.
As described in the analysis reported in [
At the current stage, the software architecture of the robot in our project encompasses a specialized module for casual conversation with autistic children, developed based on a set of experimental interactions with the children in real settings, which supports the believability of the robot character as an expert in a set of restricted domains through external knowledge sources. The current architecture does not include a reinforcement learning component, but this element may be integrated in the robot to gain advantage from the monitoring of the responses of the children during the use of the robot on the field.
As future work, we are also working on the adaptive mechanisms with which Pepper will be enriched. The robot will show more and more intelligence and reasoning skills, which will allow it to adapt to the user's needs and customize the interaction in both adaptive and adaptable form. Thus, the robot will become an adaptive robot with respect to the user, able to adapt its behavior based on the characteristics of the subject with whom it is interacting and to customize the interaction in an adaptive form, that is, configurable by the staff expert in therapy. However, from our initial investigation we believe that the interaction should be "robotguided" in order to give smarter answers to the (big variety) of questions collected.
Another future work concerns the personalization
of the natural language generated messages.
Indeed, previous work showed the possibility to
guide the user toward a virtuous behaviour by
using some specific word category (e.g.
adjectives, adverbs) and some specific sentence
coordination strategy [
The multidisciplinary project is carried out by the University of Turin, the Paideia Foundation, Intesa Sanpaolo Innovation Center and Jumple, and was funded by Banca Intesa Sanpaolo, Banca dei Territori Division.