Towards a Situation-Aware Cloud-Based Autonomous Driving System using Object Detection and Rule-Based Techniques

Autonomous driving has become a crucial aspect of modern transportation, enhancing efficiency, mobility, and safety. Essential to its success is situational awareness for navigating complex road environments. With significant growth expected in this sector by 2030, addressing the 1.19 million annual road traffic fatalities is critical. Comprehensive situational awareness can mitigate these tragedies but presents challenges like dynamic environment perception and sensor unreliability. This research aims to develop a fast cloud communication network between mobile vehicles and GPU servers, optimizing the YOLOv8 algorithm with custom datasets for improved accuracy. Objectives include detecting traffic elements and establishing a rule-based expert system over wireless communication for situation-based actions, ultimately advancing autonomous driving towards safer and more efficient systems.