Understanding Take-Over in Automated Driving: A Human Error Analysis

Automation offers a new way of driving, but often the human error (HE) in the process of take-over results in adverse effects of unrecognized risks. Hence, the impact of HE in safety of automated driving remains a major problem. This paper proposed a Human error analysis method based on analysis of the root cause of HEs events to understand the process of take-over and identify root cause of take-over failure in automated driving. Simulated driving practice with videos and questionnaire were conducted to identify the main factors leading to HEs in take-over. Human factors events diagram was used to better understand take-over as a human factor event and to provide information for root cause of take-over failure recognition. The results reveal that the most common failure mode in take-over is cognition error caused by driver poor mental state such as driver fatigue and reaction ability, followed by control error caused by inappropriate take-over request (TOR). Determination of these failure modes provide evidence for increasing or repairing barriers in the process of take-over. The suggested cognition-corresponding model of take-over showed that take-over is a complex human-machine interaction process, thus the causes of HEs should be discussed from a multi-dimensional perspective, and explored through empirical research.