Investigation of Driver Adaptations in a Mixed Traffic Environment

Existing mathematical models for car-following are mostly descriptive and do not inherently estimate behavioral responses due to different traffic conditions, such as changes in roadway, environment, or vehicle conditions. These models include behavioral parameters (e.g., reaction time, degree of aggressiveness, etc.), which are calibrated with aggregate data collected under various traffic conditions. However, these models fail to capture changes in driver behavior caused by changes in the driving environment and thus fail to address vehicle interactions and the mechanisms that lead to breakdown phenomena.

Car-following has been studied extensively during the past few decades. Car-following models are grouped into three categories (van Wageningen-Kessels, 2015): (i) safe-distance, (ii) stimulus-response, and (iii) psycho-physical models. A limited number of car-following models consider human factors. For example, the Gipps model considers human factors by introducing behavioral parameters such as desired speed, reaction time, and estimation of braking rate of the leading vehicle; however, it exhibits some problems, as it is difficult to define what constitutes “safe headway,” (i.e., safe time distance between two vehicles) or to determine how a driver estimates the deceleration of its leader. Psycho-physical models were introduced by Wiedemann (1974), and Brackstone and McDonald (1999) to address the inability of all previous models to “provide a psychologically plausible characterization of how humans think about and address the driving problem” (2014).

"Task 1: Behavioral Data Collection We will collect car-following data using the KU Driving Simulator and invite 60 participants equally split between males and females. Participants will be asked to drive a scenario with three phases: free driving, following driving without automation, and following driving with automation. The free-driving phase captures the participant’s desired speed and maximum acceleration components on an empty highway (for manual mode). The study scenarios will be designed to allow the drivers to experience significant variation in task complexity with respect to their driver capabilities. Task 2: Calibration of Automated Driving b-IDM During this task the research team will analyze the driving simulator data and will calibrate the automated driving b-IDM to reflect how changes in the driving environment, perceived difficulty, as well as engagement with automation may affect the ability of drivers to maintain a target gap with the leading vehicle. We will also group drivers to different clusters and focus on the specific clusters to identify potential differences between the different groups, and incorporate these variations into the IDM. Task 3: Final Report The research team will compose the Final Report that describes the efforts undertaken during the final stage of the project. "

This project will develop a car-following model that realistically replicates driver behavior and behavioral changes due to vehicle automation.

Download the Final Report