Virtual Autonomous Parking System Using Reinforcement Learning [2017]

I constructed the system, which takes only pixels from the viewpoints of the car agent as input, evaluates the reward and updates a model in the virtual reality environment. Also, I implemented the model using reinforcement learning algorithms including Deep Q-Network (DQN), Deep Deterministic Policy Gradient (DDPG). I verified that the car agent in VR could park accurately and rapidly within three seconds after 300 learning episodes.