GENERATING REAL-TIME DRIVING VOLATILITY INFORMATION

摘要

Drivers make their short-term steering and speed decisions based on incoming information from several sources. In order to navigate through the transportation network they adjust their speeds and change lanes, exhibiting substantial variation in driving tasks during trips undertaken within urban areas. Large variations in accelerations and decelerations can be associated with fuel wastage, greater emissions, safety problems, and avoidable wear and tear on brakes and the engine. To reduce costs, this study explores how a profile of driving variations, especially hard accelerations and braking obtained using smart devices, can be used to generate actionable warnings and alerts. Hard accelerations or braking occurs when a driver applies greater than "normal" pressure on their accelerator or brake. As a result, vehicles experience higher levels of accelerations or decelerations. To provide drivers with real-time driving warnings, their instantaneous decisions can be monitored and analyzed. This study develops a fundamental understanding of their instantaneous driving decisions. It quantifies their driving style and captures their level of volatility during driving. Empirical analysis is based on a large-scale travel behavior survey, containing 51,370 trips and their associated second-by-second (total 36 million seconds) Global Positioning System (GPS) data from Atlanta, GA collected during 2011. It shows how acceleration and braking monitoring can generate warnings and alerts. Outlier driving patterns are the key to generating actionable volatility information. Results from rigorous statistical modeling reveal how driving volatility varies significantly between driver groups and trip characteristics. The implications of the findings and potential applications to fleet vehicles and driving population are discussed.