Deep Q Network Agent for Solving 2D Robot Arm Reacher

Overview

This is an example of Deep Q learning algorithm for solving reaching target tasks on the 2D robot arm. The environment is available through this link https://github.com/ekorudiawan/gym-robot-arm

Example of agent performance is shown in this animation

Result

Here is an example result of DQN agent performance that was trained after 2000 episodes.

Dependencies

• OpenAI Gym
• PyGame
• PyTorch 1.4.0