Overview / Usage

The aim of this project is to apply Deep Reinforcement Learning to Flight Simulators, with a standard set reward system to ensure maximum optimization, with the final objective of learning how to take off and stabilize mid-flight. This approach has huge potential for future implementations in the real world.

Methodology / Approach

The approach is to leave the AI with the controls to a basic flight simulator (an unknown environment), and reward it on the basis of it's actions with positive and negative reinforcement. The AI shall seek to gain the highest reward with it's actions, and learn the basics of piloting an aircraft.This is one of the most popular Unsupervised Learning Algorithms, and has shown high precision.

Technologies Used

Intel Distribution for Python, TensorFlow* Optimizations for the Intel Architecture