A TinyML-based Alcohol Impairment Detection System For Vehicle Accident Prevention

Abstract

Driving under the influence of alcohol is one of the most severe and critical problems in every country throughout the world. Driving is a difficult endeavor that demands a high degree of concentration and great visual processing. A system based on the Internet of Things can be utilized to measure drivers’ alcohol level and restrict their operation of motor vehicles. This technology is affordable but has a number of disadvantages, including the requirement for an internet connection, the transfer of data to other organizations, bandwidth and latency constraints, and security concerns. TinyML is an emerging technology that can overcome the aforementioned challenges by performing machine learning models locally and delivering real-time intelligence. In this work, the possibility of developing a TinyML-based system that can detect alcohol and alert the driver was investigated. The experimental findings demonstrate a high degree of accuracy, indicating that the technology under consideration may be utilized to develop compact, intelligent, and inexpensive devices capable of detecting alcohol and alerting the driver in real-time.