Mathematical Modeling of the Effects of Speech Warning Characteristics on Human Performance and Its Application in Transportation Cyberphysical Systems

Transportation cyberphysical systems (CPS) aim to improve driving safety by informing drivers of hazards with warnings in advance. The understanding of human responses to speech warnings is essential in the design of transportation CPS to eliminate hazards and accidents. To date, many works have addressed diverse warning characteristics with experimental approaches. However, the computational model to quantify the effects of warning characteristics on human performance in responses to speech warnings is still missing. Mathematical equations were built to model the effects of lead time, loudness, and signal word choices on human perceptual, cognitive, and motor activities involved in speech warning responses. Different levels of lead time, levels of loudness, and signal word choices served as inputs in the model to predict human error rate and reaction time of speech warning responses. The model was validated with drivers' crash rates and reaction times to speech warnings of upcoming hazards in driving assistant systems in two empirical studies. Results showed a good prediction of human performance in responding to the speech warnings compared with the empirical data. The application of the model to identify optimal parameter settings in the design of speech warnings in order to achieve greater safety benefits is later discussed.