Abstract

During last years, evolving of robotics encouraged a gradual moving of robots from laboratories and industrial environments into people daily lives. In these environments, robots are increasingly becoming able to perform tasks and services autonomously or in cooperation with the occupants who share the working space. 

This project name is TERESA  (Therapeutic Educational Robot Enhancing Social interActions) and it is a collaboration of Dipartimento di Medicina Sperimentale, Dipartimento di Ingegneria dell'Informazione, Informatica and Statistica and Siet (Società Italiana Educazione Terapeutica). In the project TERESA they use the robot Pepper, it is a humanoid developed by SoftBank Robotics.

In this project Pepper is named TERESA, and it is used in order to prove that the presence of this social robot could be an help in medical therapies. In particular the evaluation takes place through the organization of a series of therapeutic education meetings with obese patients, using the didactic contents already extensively tested by doctors in previous experiences, but adding the presence of the social robot as an assistant to the doctor. During the therapy sessions, another member of the medical staff is responsible for the high-level control of the robot, using a simple web interface with a few buttons. According to the doctors one of the main problem during these meetings is the fact that the robot does not have the ability to interact with a specific patient, but addresses everyone, looking into space. So the interaction need to be personalized. 

Starting from this problem of the TERESA project I can now move on to the aims of my thesis: the fact that the doctors believed that an interaction in which the robot addresses the specific person might be more comfortable for the patient give me the possibility to establish the first point of my work, the personalization of robot-patient interaction.

The second step of my work starts from this TERESA’s project to be extended to an autonomous work regarding the movement of the robot through tags. The robot has to acquire images of the tags, which will be found on the floor, to understand how to move, where to look and whether to rotate towards one person or another.

At the end I studied the robustness of the detection system, I did a lot of experiments to show how the performance of this system increases or decreases in presence of occlusions and light changes.