AIoT Platform of Accident Prevention in Alley Environments

With the recent advancements in technology, the spread of smartphones has increased rapidly. Consequently, most individuals now carry smartphones and using them while walking has become a common practice. As a result, accidents involving pedestrians who are distracted by their smartphones and fail to notice approaching vehicles have become a significant concern. In this study, we propose an AI-based alleyway accident prevention platform to prevent such accidents. The proposed system integrates AIoT(Artificial Intelligence of Things) functionalities with CCTV(Closed-Circuit TeleVision)s installed in alleyways to detect vehicles and provide real-time notifications of their approach to pedestrian smartphone applications. In this study, YOLOv8(You Only Look Once version 8) was utilized as the AI(Artificial Intelligence) technology for object recognition, with training conducted using both daytime and nighttime data to enhance the model's robustness across different environments. When a vehicle enters the alley and is detected, an LED(Light Emitting Diode) installed on the ground is activated. Subsequently, when an integrated module attached to the bottom of the vehicle detects the LED, it transmits a beacon signal, which is then relayed to the pedestrian smartphone application to issue vehicle approach notifications. The results of the experiment demonstrated that vehicles were detected with an accuracy exceeding 90%, and the functionality of each component was verified. Based on these results, the proposed system is expected to effectively reduce accidents between pedestrians and vehicles.