Data-driven analysis of ACC car-following time gaps for different driving contexts

Abstract

Adaptive Cruise Control (ACC) is a key component of the Advanced Driver Assistance System which automatically adjusts the vehicle speed to maintain a safe distance from vehicles ahead. While current ACC strategies mainly take vehicle speed as the dominant factor, human drivers adapt their time headway (THW) according to a wider range of contextual factors, including infrastructure, traffic density, lead vehicle type, and environmental conditions. This thesis adopts a data-driven approach to analyze how such factors influence car-following time gaps. Using public trajectory datasets and the Volvo Cars sample dataset, we first extract steady carfollowing states and apply regression models with permutation importance to rank influential factors. Detailed analyses are then conducted on lead vehicle type, lane selection, and lighting conditions, supported by statistical comparisons across traffic groups. In addition, scenario-based studies examine interactions of multiple factors, such as cut-in intentions and on-ramp infrastructure, using clustering methods (Hidden Markov Models and K-means) and sensor signal filtering. The results show that contextual conditions affect human time gap preferences in predictable ways. These insights provide recommendations for more adaptive and human-like ACC time gap strategies, with potential benefits for driving comfort and user experience.