Improving the performance of an artificial intelligence robot using deep reinforcement learning algorithms

Authors

  • Osama Ibrahim Damascus university
  • Samir Manhal Karaman Damascus university
  • Raouf Mhana Hamdan Damascus university

Keywords:

Artificial intelligence, Convolutional Neural network, Reinforcement learning

Abstract

The Deep reinforcement learning algorithms provided a solution for artificial intelligence robots to discover the best and fastest path to reach their goal through the ability to learn from their past experiences and gain knowledge and correct representation of their environments to reach a level close to human in complex real world environments. In this study, the performance of deep reinforcement learning algorithms such as the DQN algorithm and the Policy gradient algorithm was analyzed in order to help the AI ​​robot reach its target with the best and fastest path by adjusting the higher parameters of the deep convolutional neural network used, and by comparing the results it was shown that the policy gradient algorithm was superior to 50% Approximately.

Downloads

Download data is not yet available.

Author Biographies

  • Osama Ibrahim, Damascus university

    PhD student in Computer and Automation Engineering Department Damascus University

  • Samir Manhal Karaman, Damascus university

    . prof  in Computer and Automation Engineering Department, Damascus University

  • Raouf Mhana Hamdan, Damascus university

    . Lecturer in Computer and Automation Engineering Department, Damascus University

Downloads

Published

2024-12-22

How to Cite

Improving the performance of an artificial intelligence robot using deep reinforcement learning algorithms. (2024). Damascus University Journal for Engineering Sciences, 40(3). https://journal.damascusuniversity.edu.sy/index.php/engj/article/view/6104