Reinforcemnent Learning

Using various Reinforcement Learning algorithms to solve the following 3 problems. Full details can be found in Project Report.pdf.

1. Maze Navigation

How to get from S to G in the Maze, and collect the most number of flags F along the way.

2. Open AI Acrobot (See video result)

How to swing the bot’s arm to reach a certain height.

3. Open AI Mountain Car (See video result)

How to push & pull the car to make it reach the mountain’s top.

How to run

Run these commands in your Terminal:

1. Maze Navigation problem:

1.1 Value iteration algorithm:

python value_iteration.py

1.2 Q-learning algorithm:

python qlearning_maze.py

2. Open AI Acrobot problem:

2.1 Deep reinforcement learning - REINFORCE algorithm with Neural network:

python reinforce_bot.py

2.2 Q-learning algorithm:

python qlearning_gym.py bot

3. Open AI Mountain Car problem:

3.1 Deep reinforcement learning - REINFORCE algorithm with Neural network:

python reinforce_car.py

3.2 Q-learning algorithm:

python qlearning_gym.py car