/VizDoom-ReinforcementLearning

This project implements an agent for playing the VizDoom game on various levels using the Proximal Policy Optimization (PPO) algorithm from the stablebaselines3 library. The agent is trained to learn the optimal actions to take at each step in the game in order to complete the level and maximize the score.

Primary LanguageJupyter Notebook

VizDoom-ReinforcementLearning

Creating agent that can play VizDoom using stablebaselines3.

This repository consists of 4 jupyter notebooks where agents are trained to kill enemies in different environments:

  • VizDoom-Basic.ipynb
  • VizDoom-DefendTheCenter.ipynb
  • VizDoom-DefendTheLine.ipynb
  • VizDoom-DeadlyCorridor.ipynb

Environments as well as results of rl model are shown below

Results:

Basic.mp4
DefendTheCenter.mp4
DefendTheLine.mp4
DeadlyCorridor.mp4