A Data-Driven Driving Under the Influence (DUI) Detection, Notification and Prevention System Using Artificial Intelligence and Internet-Of-Things (IoT)

A Data-Driven Driving Under the Influence (DUI) Detection, Notification and Prevention System Using Artificial Intelligence and Internet-Of-Things (IoT)

---

연구 분야: Safety

---

---

학회: World Congress in Computer Science, Computer Engineering & Applied Computing

---

초록

Drunk driving represents a significant public health crisis not only in the United States but also globally [16]. Sober Guardian is developed to mitigate this issue by leveraging advanced technologies including Machine Learning, Computer Vision, and potentially chemical sensors in the future. This system employs facial recognition and machine learning algorithms to swiftly determine if a driver is impaired, with the aim of preventing them from starting the vehicle [17]. We used an IoT System to control the blocker and successfully prevent vehicle start with 100% rate. Additionally, Sober Guardian plans to incorporate hardware solutions such as sensors that can detect alcohol levels directly from the driver’s breath or skin, enhancing the accuracy and reliability of impairment detection. Throughout its development, the project has faced challenges such as image clarity, limited datasets, and ensuring user-friendliness. Preliminary experiments with both image and video inputs have demonstrated the system’s ability to predict sobriety with an accuracy between 86% and 87.5% [18]. However, video analysis for impaired individuals was hindered by the scarcity of appropriate datasets. Despite these challenges, Sober Guardian, in its nascent stages, has proven to be a promising solution for detecting driver impairment and preventing drunk driving, with ongoing enhancements expected to improve its functionality and deployment readiness.

---