ARS is a random search method for training linear policies for continuous control problems. The paper it is based on: "Simple random search provides a competitive approach to reinforcement learning."

This repository is a part of the Udemy course titled "Artificial Intelligence 2018: Build the Most Powerful AI".

For Mac and Linux users, use a terminal. For Windows users, use Anaconda Prompt. Type the following commands line by line to install the PyBullet and OpenAI gym environments:

• pip install gym==0.10.5
• pip install pybullet==2.0.8
• conda install -c conda-forge ffmpeg

The above lines will install all the environment files that you would need to run the code.

After successful installation run the code in Spyder/Ipython notebook according to your preference. The output (in the exp folder) will be saved in the form of videos that can be compared according to the rewards obtained in the respective steps.