Mobile robot tracking control based on deep reinforcement learning

• Main Author: Toh, Yeong Jian
• Other Authors: -
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Electrical and electronic engineering::Control and instrumentation::Robotics; Engineering::Electrical and electronic engineering::Control and instrumentation::Control engineering
• Online Access: https://hdl.handle.net/10356/149297
• Institution: Nanyang Technological University

The research and applications of artificial intelligence in machines, specifically in the field of robotics have achieved exponential growth in the recent years due to the outburst in volume of readily accessible data online and the production of powerful graphic processing units for coding, computation and even visualisation purposes. The integration of artificial intelligence in robots yield a significant improvement in performance of their taskings, allowing adaptational ability to changing environment autonomously and continuous self-evaluation and improvement to their ability. Deep reinforcement learning (DRL), in particular, uses neural network in conjunction with machine learning (ML) techniques to enhance mobile robot control in different situations. This paper explores two different methodologies under deep reinforcement learning category to evaluate and compare their performances. The two said methodologies are Deep Q Network (DQN) and Deep Deterministic Policy Gradient (DDPG) algorithms. In both methods, training process was conducted to the individual agents to achieve their predetermined goals of desired trajectories, then results were extracted for evaluation and comparison.