Robot writing via reinforcement learning
Loading...
Date
2023-10-04
Authors
Καλέργης, Αλέξανδρος
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Recently, with the rapid development in the field of artificial intelligence, attempts have been made to apply this technology to robots in the service area. So, several studies have been conducted to use robots to execute human tasks in daily life.
In this thesis we aim to solve a problem of a robotic hand trying to learn how to write alphabetical characters and/or other shapes. We created a physical 2 DOF robotic system that represents a human arm that only operates in x-axis and y-axis. The methodology we used is an adaptation of reinforcement learning and specifically the Q-Learning algorithm as an easier to understand, use and have the desired results with less time consuming and complexity. The robot writing though is not just a physical approach, it has also impact in humanity. Robot writing is a task that could be used for helping purposes in human with disabilities or/and in early education. further studies about reinforcement learning in robotic systems could also bring even better impact in humanity.
Description
Keywords
Artificial intelligence, Machine learning, Reinforcement learning, Robotics, Q-learning, Deep Q-learning, Neural networks