Toward detection of unsafe driving with inertial head-mounted sensors

Abstract

This paper explores the potential for the inertial sensors on head-mounted devices (HMDs) such as Google Glass, and mobile devices to identify driving activities and unsafe driving. Particularly, we study whether the inertial sensors on HMDs, can detect whether their user is operating the vehicle and infer where the driver's visual attention is focused at by tracking the driver's head without relying on cameras on the HMD or the vehicle. Detecting user's vehicle operation through head movements can help prevent the driver's interactive usage of HMD, which can cause distracted driving. Tracking driver's head movements and the driver's focus of visual attention can detect unsafe driving and additionally enable many safety applications. Our approach relies on head movements that are specific to a driver. Once the system detects the vehicle's operation, head movements can be further utilized to detect distracted driving, to predict driver behaviors and to detect inadequate surveillance cases where the driver failed to look in the appropriate place to complete a maneuver.