README.md

BipedalWalker

About

In this project, we utilized three reinforcement learning algorithms to teach our agent to walk which were Q-learning, Deep Q-Network (DQN), and Twin Delayed DDPG (TD3). The agent we used was from the OpenAI Gym environment called BipedalWalker-v3. The objective of the agent is to get a score of 300 or higher without falling. The score is defined as the cumulative rewards gathered for a particular episode. It is calculated as a function of the agents movements and if the agent were to fall, then it would get a score of -100. More information at https://gym.openai.com/envs/BipedalWalker-v2/

We evaluated the performances of our algorithms with a comprehensive academic report. This could be found at https://github.com/FranciscoHu17/BipedalWalker/blob/main/BipedalWalkerReport.pdf

Dependencies

Requires Python version 3.5 to 3.8

If on Windows:

• Install Anaconda at https://www.anaconda.com/products/individual
• Verify your installation in Anaconda Prompt with the command conda list which should display a list of installed packages
• Also verify that python is working by entering the command python
• Run conda install swig
• Make sure you have Microsoft Visual C++ 14.0 or greater. It can be installed at https://visualstudio.microsoft.com/visual-cpp-build-tools/
• Install gym by running

git clone https://github.com/openai/gym
cd gym
pip install -e .[box2d]

• Install pytorch by runnning conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch

If on MacOS (has not been tested, but should theoretically work):

• Install Anaconda at https://www.anaconda.com/products/individual
• Verify your installation in Anaconda Prompt with the command conda list which should display a list of installed packages
• Also verify that python is working by entering the command python
• Run conda install swig
• Install gym by running

git clone https://github.com/openai/gym
cd gym
pip install -e .[box2d]

• Install pytorch by runnning conda install pytorch torchvision torchaudio -c pytorch

If on Linux:

• Install gym by running

git clone https://github.com/openai/gym
cd gym
pip install -e .[box2d]

• Install pytorch by runnning pip3 install torch torchvision torchaudio

More information on installing OpenAI Gym API at https://github.com/openai/gym#installation

Running the Code

• Q-Learning

# From the root directory:
cd Q-Learning
python QLearningWalker.py

• DQN

# From the root directory:
cd DQN
python dqn_walker.py

• TD3:

# From the root directory:
cd TD3
python td3_walker.py