Institute of Transportation, MOTC

       According to statistic and previous researches, at least 90 percent of crashes involve human factors, 80 percent of which can be attributed to driving errors. Most of these past findings were based on self-reported errors recorded in in-depth crash investigations or driving behavior questionnaire studies, and hence they are often challenged on the grounds that these self-reported errors cannot be verified. In any case, it is still unclear how these driving errors occur and how exactly they contribute to crash occurrence. Consequently, this study aims to identify the effects of driving situations and the driving environment on driving errors under real-world driving conditions. Specifically, this study seeks to examine the associations between driving errors, driving situations, roadway attributes, and safety-related events involving lane changes. For these lane change events, a driver’s behaviors 10 seconds prior to lane changing was coded in order to shed some light on the outcome. For instance, when exactly did the drivers start checking their back mirrors, and what were the driving situations that existed when drivers executed a lane change, etc. from the driver’s face view, forward view, and side view cameras. SCEs were also identified when a lane change event involved an imminent rear-end crash risk. The model results shows that the driving dynamics and roadway geometry are both influence drivers make errors during lane changes and the combination of driving dynamics and driving errors cause safety correlative events(SCEs) more easily. Besides, drivers usualy have more than one error when they change lanes. Thus, this study also sort out the common error combinations from nine types of driving errors during information gathering, decision-making, and action phases, respectively, throughout the entire lane changes were defined and identified.