Detection of lateral hazardous driving events using in-vehicle gyro sensor data

Abstract

Hazardous driving maneuvers due to driver's inattentive behavior is highly associated with vehicle crash occurrence. Recent advances in sensors allow for valuable opportunities to monitor driving behavior and identify its characteristics. This study proposes an algorithm for detecting lateral hazardous driving events and classifying their severity using in-vehicle gyro sensor data. The detection of hazardous driving events focuses on two lateral hazardous driving events, i.e., lane changes and zigzag driving. The algorithm classifies lane change events into single-lane changes and double-lane changes using a well-known and robust pattern recognizer, Support Vector Machine (SVM). Similarly, the motion of zigzagging within a lane and zigzagging between lanes can be identified by the algorithm. The proposed algorithm uses maximum and minimum yaw rate, and duration of hazardous driving events obtained from a gyro sensor. Performance evaluations of the algorithm show promising results for actual implementation in practice. The proposed methodology is expected to be effectively used for a fundamental to devise various safety countermeasure. For example, in-vehicle warning information systems and differentiated insurance fees based on driver behavior can be taken into consideration as useful further applications.