Yet Another Reinforcement Learning Tutorial

This repo contains self-contained RL implementations, including:

• Basic Plot Usage
• Basic OpenAI Gym Usage
• Value Iteration
• Policy Iteration
• Monte Carlo Learning
• SARSA
• Q Learning
• DQN
• Generalized Advantage Estimate

Custom RL environment using MuJoCo with Snapbot:

The objective is to make a custom MuJoCo environment for training an agent (Snapbot) with RL.

• Parse Snapbot MJCF
• Make Snapbot environment
• Train Snapbot to walk using soft actor-critic (SAC)
• Evaluate SAC-trained Snapbot
• Train Snapbot to walk using augmented random search (ARS)

Lecture slides:

• Introduction to RL
• Model-based methods in RL
• Model-free methods in RL
• Policy-based methods in RL
• Population-based methods in RL
• Summary