Long-term interaction for interactive behavior learning using deep reinforcement learning

• Grant number: 18/25782-5
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: November 01, 2019
• End date: September 30, 2022
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Roseli Aparecida Francelin Romero
• Grantee: José Pedro Ribeiro Belo
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

Social robotics represents a branch of human-robot interaction that aims to develop robots to work in unstructured environments in direct partnership with humans. Social robots are expected to have the ability to interact with humans by understanding social cues and responding appropriately in order to promote a natural and socially acceptable interaction between humans and robots. The aim of this thesis is the elaboration of a module for social robots capable of automatically learning the interactive behaviors of human beings and, from this, reacting in a natural and appropriate way to these behaviors. To achieve this goal, deep reinforcement learning techniques will be explored, as well as specific aspects of long-term interaction that will allow the robot to memorize past interactions. In addition, the system will be added to the Cognitive Model Development Environment (CMDE) architecture, developed at the Robot Learning Laboratory (LAR) of USP-São Carlos, aimed at modeling applications involving social robotics. In this way, this thesis contributes to the architecture by adding a learning module of interactive behaviors. In addition, the validation of the system will be done through an application that allows a long-term analysis, that is, long-term interaction with the robot, which will store interactions and actions performed, associated with human recognition. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

BELO, JOSE PEDRO R.; AZEVEDO, HELIO; RAMOS, JOSUE J. G.; ROMERO, ROSELI A. F.. Deep Q-network for social robotics using emotional social signals. FRONTIERS IN ROBOTICS AND AI, v. 9, p. 19-pg., 2022-09-26. (17/01687-0, 18/25782-5)

BELO, JOSE PEDRO R.; ROMERO, ROSELI A. F.; IEEE. A Social Human-Robot Interaction Simulator for Reinforcement Learning Systems. 2021 20TH INTERNATIONAL CONFERENCE ON ADVANCED ROBOTICS (ICAR), v. N/A, p. 6-pg., 2021-01-01. (18/25782-5, 17/01687-0)

Academic Publications

(References retrieved automatically from State of São Paulo Research Institutions)

BELO, José Pedro Ribeiro. Deep reinforcement learning for social robotics using social signals and facial emotions.. 2022. Doctoral Thesis - Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB) São Carlos.