Shayon

Project Description
- Use of Q-Learning (Reinforcement Learning) to simulate a game between two drones. Evasion of obstacles and detection of the drone leads to a victory.
- First a single agent system was created and trained using a Neural Network and Q-Learning approach and the two methods were compared.
- From the knowledge gained we tried to develop a two agent environement. Q-Learning is not very efficient in this regard as training simultaneously led to a deadlock.