GuillermoVR92 / Deep-RL-Q_Learning_Taxi_v2

Implementing a Q Learning agent that solves the taxi v2 environment from Open Gym.

OpenAI Gym Taxi v2

Guillermo del Valle Reboul

Project details

The goal of the agent is to collect a client at point A and then move to point B while avoiding obstacles.

OpenAI Gym: Taxi v2 Vector Observation space type: discrete Vector Action space type: discrete Vector Action space size (per agent): 6 (up, down, right, left, pick client, drop client) Vector Action descriptions: , , ,

state vector = grid actions = 6 discrete actions (up, down, right, left, pick client, drop client) The environment is considered solved when agents reaches average score of 9.7.

Algorithm Used: Q Learning (RL)

A Q Learning agent was used for this project. The policy in use is epsilon greedy. Hyperparameters:

• Num episodes = 20000
• GAMMA (discount factor) = 0.77
• Alpha = 0.25

Instructions

Download OpenAI Gym Taxi v2 and execute main.py.