Autonomous Car Parking System using Deep Reinforcement Learning

In recent years, technologies based on deep learning have been useful in various aspects of our daily lives. In the field of automated driving, which is attracting particular attention, image recognition technology is used to detect roads, white lines, and vehicles ahead. However, since automated vehicles are controlled by acquiring information about the vehicle's position and surrounding environment mainly from image sensors and cameras, the production cost is very high. The goal of this research is to develop autonomous driving technology using only an on-board visual camera, without any image sensors. Automatic parking is implemented using reinforcement learning in the virtual environment of Unity. Autonomous parking with high accuracy is achieved by using the input image as a segmentation image.