A human factors-based modeling framework to mimic bus driver behavior

Over the past 50–60 years, numerous driver behavior models have been proposed in the literature. However, the literature still lacks models describing bus drivers’ behavior in traffic streams, even though buses comprise a non-negligible component of the traffic mix in many cities. Further, bus driver behavior might differ from other vehicles due to the differences in size, kinematic characteristics, maneuvering capabilities, and the number of occupants. Moreover, human factors such as multi-vehicle anticipation and stimuli perception contribute to this difference in driver behavior. Motivated by these reasons, this study presents a new modeling framework for mimicking bus driver behavior. The framework incorporates two important aspects of bus driver behavior: multi-vehicle anticipation and stimuli perception. Based on the proposed modeling framework, the study modifies the widely used Intelligent Driver Model (IDM). A variance-based sensitivity analysis is carried out to recognize the influence of model parameters (specifically, the new parameters) on the output of the IDM model. The modified IDM model is calibrated and validated using an empirical trajectory dataset of about 90 buses from a traffic stream in Chennai, India. In doing so, the study also contributes to modelling driver behavior in heterogeneous and disorderly traffic streams found in Indian cities and elsewhere. The parameter calibration results show that the average calibrated parameters of the modified IDM offer realistic interpretations, and the calibration and validation errors are small. Furthermore, it is evident from the results that the perceived space gaps by bus drivers can be longer or shorter than the actual space gaps. Overall, the modified IDM model outperformed the original IDM, highlighting the efficacy of the proposed multi-vehicle anticipation and stimuli perception features in the model. Finally, the study also evaluates the model performance by analysing its stability and macroscopic properties.