Fall Detection of Riders using Inertial Sensors: A Smart Helmet

A helmet is a protective headgear used by riders to make driving safer and avoid injuries. In this paper, a smart helmet is designed which is capable of detecting fall of riders and then transmitting a message through a global system of mobile (GSM) module along with the location of fall using global positioning system (GPS) avoiding delays to rescue. Inertial sensors, i.e., gyroscope and accelerometer sensors with a three- axis sensing unit, are placed within the helmet and controlled via ARDUINO based microcontroller for fall detection. The designed helmet is used for data acquisition in 28 trials, and twenty-two statistical features are extracted from the recorded data. A subset of discriminating features are selected using the wrapper method and four different classifiers, i.e., Naïve Bayes, k-nearest neighbor, random forest, and support vector machine are applied on selected features to classify the two states i.e., fall and non-fall. It is evident from results that the Naïve Bayes classifies the fall of riders with the highest accuracy of 98.21%. The proposed methodology outperforms the existing state-of-the-art fall detection techniques in terms of classification accuracy.