Enhancement in identification of unsafe driving behaviour by blending machine learning and sensors

The majority of traffic accidents are caused by human drowsiness, weariness and absentmindedness. ML “Machine learning” technology has recently been used to accurately identify styles of driving and recognise risky DB “Driving Behaviour” using signals of Inside Car sensors “In-Car sensors”. For example speed of engine and car, position of throttles, load on engine during driving. Through detailed study and analysis we brainstormed the prominence of fusion and blend various Outside Car sensors “Out-Car sensors”, like GPS, magnetometer or a gyrometer with In-Car sensors in order to yield enhanced accuracy in identification of DB style. The proposed integration of In-Car and Out-Car sensors will enhance ML identification of unsafe DB. From Out-Car sensors a set of capable features can be computed which may precisely portray the DB. The factual data may be utilized to assess classification performances through a specific target strategy dependent on the connection among car speed, longitudinal acceleration and lateral longitudinal acceleration of car. Based on this technical fusion of ML and Out-Car sensors the potential capability of system can definitely take an accuracy leap in identification of unsafe DB. In this paper, we first discuss the idea of risky driving behaviour and the significance of technology in helping safe driving, followed by a review of related work in the field of DB identification. Following that, we propose our sensor fusion technology for improving DB identification or detection results. We go over numerous In-Car/Out-of-Car sensors and discuss the performance matrix. Finally, this study highlights the paper's conclusion and future scope of work in this area, which may contribute in further enhancing the precision of DB identification results.