Learning manipulation skills using deep reinforcement learning

• Main Author: Tan, Herman Jin Xing
• Other Authors: Wang Dan Wei
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence; Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/149318
• Institution: Nanyang Technological University

In recent years, the growth of robotic arms working in the manufacturing line has been significant. Industrial robots are usually semi-supervised by an operator, and this is a very mundane task. However, the ability to teach robotic arms to become autonomous has been a challenge. Therefore, the purpose of this study is to remove the mundane tasks from the operator's work by teaching the robot to do pick-and-place tasks autonomously. The starting point proposed by this study is to teach the UR5 robotic arm to learn pick-and-place using Unity Technologies's simulation software and Machine Learning Toolkit called Unity Game Engine and ML-Agents, respectively. Besides, the report looks into incorporating inverse kinematics for joint rotation calculation to place the robotic arm's end effector to a destination.