Electronic Thesis and Dissertation Repository

Degree

Master of Science

Program

Computer Science

Supervisor

Professor Michael A. Bauer

Abstract

All drivers have their own driving style while performing different driving maneuvers. They vary in using vehicle’s control devices such as the steering wheel, pedals, gears etc. In this thesis, we analyze driving behavior in different timeframes prior to turns. We employ data obtained from actual driving behavior in an urban environment collected from the CAN-Bus of an instrumented vehicle. Five CAN-Bus signals, vehicle speed, gas pedal pressure, brake pedal pressure, steering wheel angle, and acceleration, is collected for 5, 10, and 15 seconds of driving prior to each turn. We consider all turns for each driver as well as look specifically and right and left turns. We use cluster analysis to see if we can categorize drivers into possible groups of driving styles. In our first approach, we use hierarchical clustering on statistical features extracted from the signals. The results show that using this approach we can effectively cluster drivers into two groups, moderate and aggressive drivers. This pattern is also reflected in the analysis of right and left turns. Another approach makes use of the Dynamic Time Warping (DTW) technique to identify the distance between signals of each pair of drivers, and based on these distances, a cluster analysis using hierarchical clustering is performed as well. The results show high consistency in the membership within a cluster throughout different timeframes.


Share

COinS