The Study of Drunken Abnormal Human Gait Recognition using Accelerometer and Gyroscope Sensors in Mobile Application

Abstract

International Classification of Diseases (ICD-10) defined that alcohol is a contributing factor that caused catastrophic health problems and the environment. Even 0.05% of Blood Alcohol Concentration (BAC) in the human body could affect the gait of a person. Thus, this research aims to develop a recognition technique to identify drunken humans by their gait using an accelerometer and the gyroscope sensors. The abnormal gait detection of a drunk person is important for the early recognition of a drunk person to avoid bad incidents to occur. For the collection of the data, the sensor was placed on the right leg of the volunteer. The gait data were recorded before and after the volunteer consumes alcohol. The Deep Neural Network algorithms were implemented in this model to predict either the person drunk or normal. The objectives of this research were successfully achieved by capturing the people's gait walking patterns. The collection of data and identification of the best algorithms used in detecting the behavior of the person also achieved. From this, we can conclude that this study using the Deep Neural Network algorithm fits well on the test data by achieving approximately 79% of the accuracy.