Portfolio Details

Auto Dino

Auto Dino is a project designed to automates Google Chrome's Dino Game and try to continuously improve its high score. This automating program uses DeepMind's deep learning model for reinforcement learning.

Launch this Project

The inspiration for this project came when there was no internet for almost 12 hours. During that time, there was a competition held in my lab to score the highest till the internet is back. This resulted me to think that how to automate that process and get the highest score without even playing it.

A 2013 publication by DeepMind titled ‘Playing Atari with Deep Reinforcement Learning’ introduced a new deep learning model for reinforcement learning, and demonstrated its ability to master difficult control policies for Atari 2600 computer games, using only raw pixels as input.

The algorithm that was discussed in the above paper is Q-learning, a technique of reinforcement learning, where we try to approximate a special function which drives the action-selection policy for any sequence of environment states. Q-learning is a model-less implementation of Reinforcement Learning where a table of Q values is maintained against each state, action taken and the resulting reward. Deep Neural Networks help to choose an action with highest predicted Q-value via regression. For this project I decided to use Keras with TensorFlow backend.

After 15 hours of training on my laptop, auto dino was able to cross 1000. Even though this project didn't win the competition, it surely helped me make better use of my time!