Intelligent Parking Assistance Using Deep Reinforcement Learning and YOLOv8 in Simulated Environments

This study introduces a novel approach to enhancing vehicle parking accuracy by integrating the CARLA simulator and Deep Q-Learning, reinforced by cutting-edge technologies such as YOLOv8. The system is trained in a highly realistic simulation environment created using CARLA, which replicates diverse real-world parking scenarios. Leveraging reinforcement learning via Deep Q- Networks (DQN), the system learns parking strategies through trial and error, continuously improving its performance. Additionally, the research incorporates advanced features, including human-voice interaction, heatmap guidance for optimal parking using YOLOv8, and a stealth parking mode for security-focused scenarios. The voice interaction module facilitates human-system collaboration by providing real-time spoken feedback and accepting verbal commands. YOLOv8-based heatmaps visually highlight the most suitable parking spots, ensuring optimal decision-making. The stealth parking feature autonomously identifies inconspicuous spaces for enhanced privacy. These innovations aim to minimize common parking errors, reduce accidents, maximize parking lot efficiency, and provide an intuitive user experience. Rigorous evaluation through simulations and real-world testing ensures the system's practicality and reliability. This research paves the way for safer, more efficient, and user-friendly parking assistance technologies.