Reinforcement Learning Projects

This repository hosts diffrent reinforcement learning implementations across diverse environments and algorithms.

Projects

1. OpenAI Gym Dino DQN

Trains a Deep Q-Network (DQN) agent to play a Pygame-based Chrome Dinosaur game. The agent learns to jump obstacles using visual input and reward feedback. Uses gymnasium a fork of openAI Gym framework.

• Dependencies: numpy, torch, pygame, gymnasium, matplotlib.

2. MineRL PPO Agent

Implements a Proximal Policy Optimization (PPO) agent in Minecraft (MineRL environment) that performs gathering resources and lighting campfires tasks.

• Dependencies: numpy, torch,minerl, gym, tqdm.

3. ML-Agents Unity

Trains a 2D agent in a Unity ML-Agents environment to navigate to targets.

• Dependencies: numpy, torch,mlagents, pyyaml, Unity Editor.

---

Prerequisites

• Python 3.8+
• pip for package installation.