A Literature Survey of Drunk Driving Detection Approaches

Abstract

Drunk and distracted driving have been prime reasons for road accidents. An increase in population in urban cities leads to the risk of increasing deceased cases due to road accidents. Metropolitan cities needed a preventive & scalable system to prevent the loss of life due to accidents. Nevertheless, each method has limitations, such as usability, complexity, scalability, and burdensome implementation. This work describes the various approaches used to date in drunk driving systems with their pros and cons. Techniques are also grouped based on the methodology adopted by the researcher as follows; Alcohol sensor-based, IOT or Videos Based, Ignition Control using Hardware-based, Touch-based technology, Using Machine Learning or Neural Networks, and Hybrid approaches. This work also lists the accuracy of machine learning algorithms like Linear Discriminant Analysis, Support Vector Machine, Ada Boost, and Random Forest acclimated to the drunk-driving system.