Big data analytics to identify deceleration characteristics of an older driver

Abstract

This paper presents the analysis of all driving by a single (female) older diver over a one year period from the Candrive project. Data analytics techniques have been applied to this unique big data set that includes 1 Hz sampled Global Positioning System (GPS) and Geographic Information System (GIS) data and includes the analysis of 1562 trips covering 13,425 km. The driver is known to have stable general, cognitive and physical health through clinical testing at the start and end of the 1 year period. The paper specifically explores the deceleration habits of the driver by locating all deceleration events over the period with a net velocity drop of 4km/hr or more resulting in 24,794 events being identified. The paper finds that the mean and minimum deceleration values for the events, both have two phases where the deceleration values increase with the size of the velocity drop (-0.252 and -0.0593 hr·m/km·s2 respectively) until the drop exceeds 27.5km/hr and then the second phase has a much lower slope (-0.027 and -0.0053 hr·m/km·s2 respectively). Subsets of the deceleration events such as posted speed limit on road and decelerations ending with a stopped vehicle exhibit the same two phase relationship. The two phases and their transition are attributes of the deceleration habits for the driver that may potentially be used to distinguish between drivers of a vehicle.