Methodology to Events Identification in Vehicles Using Statistical Process Control on Steering Wheel Data

Driving a vehicle has become a challenging task, especially because up to 90% of accidents involving vehicles are caused by drivers' errors. In this paper, we propose a method for events identification with the use of an SPC (Statistical Process Control) technique, known as EWMA (Exponentially Weighted Moving Average) control chart, which correlates the SWMs (Steering Wheel Movements) with deviations on the road. It has low cost, provides quick detection and, more importantly, can be used in online applications. A total of 43 sets of driving data were collected from various driving simulations performed by subject drivers. Results show that the application of this technique is feasible, allowing the identification of dangerous lane departure events with a 91.42% precision and low false positive rate.