Machine Learning Based Condition Recognition System for Bikers

This paper describes an efficient way of detecting the behavior and monitoring the safety of the two-wheeler driver. The Extreme Learning Machine Algorithm based Driver Condition Recognition (ELMA-DCR) is proposed to detect three parameters such as alcohol taken by the driver, accident detection, and overweight detection (more than two persons). The alcohol sensor is used to identify the drunken person, an accelerometer is used to determine the accident, and a strain gauge is used to find the overloading of the vehicle. A communication device is interfaced with the controller for transferring the information to a particular person. The square pooling architecture is used in the convolution algorithm for extracting high-level features of the signals received from the sensors. The ELMA-DCR is evaluated by comparing the performance with other state-of-art algorithms. The results of the proposed method show that it is an accurate and efficient technique helpful for two-wheeler drivers. An experimental setup is constructed and tested under different test conditions.