Publication detail

Comparison of Multiple Reinforcement Learning and Deep Reinforcement Learning Methods for the Task Aimed at Achieving the Goal

PARÁK, R. MATOUŠEK, R.

English title

Comparison of Multiple Reinforcement Learning and Deep Reinforcement Learning Methods for the Task Aimed at Achieving the Goal

Type

journal article in Scopus

Language

en

Original abstract

Reinforcement Learning (RL) and Deep Reinforcement Learning (DRL) methods are a promising approach to solving complex tasks in the real world with physical robots. In this paper, we compare several reinforcement learning (Q-Learning, SARSA) and deep reinforcement learning (Deep Q-Network, Deep Sarsa) methods for a task aimed at achieving a specific goal using robotics arm UR3. The main optimization problem of this experiment is to find the best solution for each RL/DRL scenario and minimize the Euclidean distance accuracy error and smooth the resulting path by the Bézier spline method. The simulation and real word applications are controlled by the Robot Operating System (ROS). The learning environment is implemented using the OpenAI Gym library which uses the RVIZ simulation tool and the Gazebo 3D modeling tool for dynamics and kinematics.

English abstract

Reinforcement Learning (RL) and Deep Reinforcement Learning (DRL) methods are a promising approach to solving complex tasks in the real world with physical robots. In this paper, we compare several reinforcement learning (Q-Learning, SARSA) and deep reinforcement learning (Deep Q-Network, Deep Sarsa) methods for a task aimed at achieving a specific goal using robotics arm UR3. The main optimization problem of this experiment is to find the best solution for each RL/DRL scenario and minimize the Euclidean distance accuracy error and smooth the resulting path by the Bézier spline method. The simulation and real word applications are controlled by the Robot Operating System (ROS). The learning environment is implemented using the OpenAI Gym library which uses the RVIZ simulation tool and the Gazebo 3D modeling tool for dynamics and kinematics.

Keywords in English

Reinforcement Learning, Deep neural network, Motion planning, Bézier spline, Robotics, UR3

Released

21.06.2021

Publisher

Brno University of Technology

Location

Brno University of Technology

ISSN

1803-3814

Volume

27

Number

1

Pages from–to

1–8

Pages count

8

BIBTEX


@article{BUT172507,
  author="Roman {Parák} and Radomil {Matoušek},
  title="Comparison of Multiple Reinforcement Learning and Deep Reinforcement Learning Methods for the Task Aimed at Achieving the Goal",
  year="2021",
  volume="27",
  number="1",
  month="June",
  pages="1--8",
  publisher="Brno University of Technology",
  address="Brno University of Technology",
  issn="1803-3814"
}