Fleet analysis of headway distance for autonomous driving

Abstract Introduction Modern automobiles are going through a paradigm shift, where the driver may no longer be needed to drive the vehicle. As the self-driving vehicles are making their way to public roads the automakers have to ensure the naturalistic driving feel to gain drivers’ confidence and accelerate adoption rates. Method This paper filters and analyzes a subset of radar data collected from SHRP2 with focus on characterizing the naturalistic headway distance with respect to the vehicle speed. Results The paper identifies naturalistic headway distance and compares it with the previous findings from the literature. Conclusion A clear relation between time headway and speed was confirmed and quantified. A significant difference exists among individual drivers which supports a need to further refine the analysis. Practical applications By understanding the relationship between human driving and their surroundings, the naturalistic driving behavior can be quantified and used to increase the adoption rates of autonomous driving. Dangerous and safety-compromising driving can be identified as well in order to avoid its replication in the control algorithms.