A model of human brake initiation behavior with implications for ACC design

Many automobile manufacturers have recently included or will soon include adaptive cruise control (ACC) systems in some vehicles. The operational limit of ACC-generated (hard) braking is a critical factor determining the human driver's interaction with the automation. Using a satisficing-based multiple mental model perspective, we generate a characterization of the natural onset of human-generated braking. We hypothesize that effective human interaction with an ACC system, that is, an interaction resulting in safe and comfortable transition from ACC to human control via human intervention, is achieved when (a) the onset of automated braking matches that of a skilled human operator and (b) a human driver can easily detect and interpret the operational limits of ACC-generated braking. This hypothesis, which is supported by experimental evidence obtained from both driving simulator and test track studies, implies that effective ACC designs should perform braking by either (a) matching human behavior or (b) augmenting human ability to detect and interpret the operational limits of ACC-generated braking through a surrogate such as a warning.