An Intelligent In-Vehicle Drunk Driving Prediction System Based on E-Nose Assisted Multi-signal Fusion Technology

Abstract

Alcohol-impaired driving remains a critical threat to public safety, while traditional detection methods, such as breathalyzers, are often limited by the need for driver cooperation and their inability to provide real-time results. In response to these challenges, an intelligent in-vehicle system has been developed for the active detection of impaired driving. This system features a robust YSZ-based electronic nose (e-nose) using a SnO₂/Co₃O₄ composite, which demonstrates exceptional selectivity and sensitivity to breath alcohol, with a detection limit as low as 10 ppb and acceptable response/recovery (7 s/60 s) time. In addition, a millimeter-wave (MmWave) sensor and a mini camera are seamlessly integrated with the e-nose to monitor breath alcohol levels, heart rate, and facial expressions simultaneously. By employing multi-signal fusion technology, the integrated system achieves high accuracy in identifying alcohol-impaired driving, with a sensitivity of 97.14% and an accuracy of 98.67%. This innovative multi-signal synchronized monitoring system offers a promising strategy for enhancing road safety through real-time, highly accurate detection of drunk driving within vehicles.