A Spatial-Temporal Analysis of Brain Activation to Explain the Take-Over Failure in Conditionally Automated Driving

Abstract

ObjectiveThis study aims to investigate the causes of take-over failures in conditional automated driving with spatial-temporal analysis of brain zone activation.BackgroundTake-over requires a human driver to resume the control of the vehicle when its automation system disengages. Existing studies have found that take-over failures occur frequently on some drivers, but the causes have not been thoroughly studied.MethodIn a driving simulator experiment, 40 drivers took over in critical freeway cut-in situations. Functional near-infrared spectroscopy (fNIRS) data were collected before and during the take-over process to evaluate brain zone activation. Successful and failed take-overs were compared with changes in fNIRS data based on spatial-temporal comparisons and cluster analysis.ResultsThe results suggested a significant difference in temporal brain activation between take-over failure and success conditions. Take-over failure conditions are mostly related to earlier and longer brain activation in most brain zones and repeated activation of the cognition brain zones. Drivers’ attention switches, steering, and braking patterns are also related to different brain zone activation orders.ConclusionThe results indicate the need to reduce the mental workload caused by the sudden system disengagement to prevent take-over failure.ApplicationFuture research and implementation should focus on earlier warnings of upcoming hazards and driver education in dealing with sudden system disengagement.