The target of the project is to develop an autopilot machine using only visual feedback from a front facing camera. The primary objective is to develop the autopilot on various racing games, and extend the application further to real life systems.

• Part 1 Image Processing and Control

• Part 2 Learning on Images

• Part 3 Control using Neural Network

• Part 4 Reinforcement Learning Control

The first stage of this project is to capture any game window running on the machine and import those to opencv as screenshots. This will be the primary input to any control/learning algorithm used for the entire project. The second step is to send basic keyboard commands to the game for controlling the vehicle.

For more details on this section, head on to the Part 1 Directory