Deep reinforcement learning for bipedal locomotion

• Grant number: 17/21426-7
• Support Opportunities: Scholarships in Brazil - Master
• Start date: March 01, 2018
• End date: February 28, 2019
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Agreement: Coordination of Improvement of Higher Education Personnel (CAPES)
• Principal Investigator: ESTHER LUNA COLOMBINI
• Grantee: Yuri Corrêa Pinto Soares
• Host Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil

Abstract

Robotics and its service applications with biped robots have faced an upsurge lately as this category of robots is very suitable for the environments designed for humans where they are supposed to operate. However, bipedal locomotion has proven to be a challenge in theory and practice due to the high dimensionality of the problem once walking gaits typically involve precise real-time control of multiple actuators and sensors coupled with complex dynamical systems. Concomitantly, reinforcement learning (RL) and its deep version (DRL) are becoming a prominent approach in solving such challenging control problems due to their capacity to work on continuous and model-free processes. In this context, our goal in this work is to apply DRL for learning a stable walk of a simulated and a real version of a humanoid robot while assessing its effectiveness for robotic locomotion tasks and, in particular, bipedal locomotion. (AU)

Academic Publications

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

SOARES, Yuri Corrêa Pinto. Aprendizado por reforço profundo para locomoção bípede. 2020. Master's Dissertation - Universidade Estadual de Campinas (UNICAMP). Instituto de Computação Campinas, SP.