Simulation of Pilot Behavior in Air to Ground Combat using Deep Reinforcement Learning

With the advancement of technology, the level of insight in air-ground battle is expanding, and the role of computerized wise dynamic frameworks is getting more demanding. The project aims to simulate pilot behavior in the air to ground combat using deep reinforcement learning. We designed an aircraft and a SAM launcher for simulation in unity with the help of an ML agent. The aim was to make the aircraft (pilot) learn from each run-in learning behavior so that it can survive for a longer period and bomb the SAM launcher more accurately.