S. Siuhi, M. Kaseko

Pages: 81-96

Abstract
This study developed a nonlinear family of car-following models that emulate driving behavior in congested freeway traffic conditions. The study developed separate sub-models for acceleration and deceleration responses. The study calibrated these models using individual vehicle trajectory data for “automobile following automobile” collected on a segment of Interstate 101 in Los Angeles, California. The study used nonlinear regression with robust standard errors to estimate the model parameters and to obtain their distributions across drivers. The stimulus response thresholds that delimited the acceleration and deceleration responses were determined based on Signal Detection Theory. The results indicated that the average driver’s response time lag was lower for the deceleration response than for the acceleration response. This result was expected, since deceleration response is related to safety, therefore, drivers tend to respond faster than for acceleration response. The acceleration response is related to drivers’ desire to attain maximum speed, which is a less critical need than deceleration response. Due to similar reasons, the results also showed that the average stimulus response threshold was lower for deceleration response than acceleration response. Furthermore, the deceleration response had higher magnitude of parameters than the acceleration response, which further indicated that, on the average, drivers were more aggressive when required to decelerate than when they wanted to accelerate. Additionally, drivers’ response to negative stimuli is sometimes further aided by the activation of brake lights for a leading vehicle that is braking.

Keywords: Stimulus-response; car-following models; relative speed