| Summary: | This research performs comparative analyses on drivers’ behavior during mandatory and discretionary lane changes. We do this by examining the statistical properties of four lane changing decision variables that describe the gaps between the subject vehicle and the surrounding vehicles. Mandatory and discretionary lane changes in NGSIM’s I-80 Freeway and U.S. Highway 101 data collection sites were identified. First, for each variable at the same site, descriptive statistics for the two types of lane changes were compared, and hypothesis tests on the difference between two means were conducted. Then, for each decision variable at the same site, the observed cumulative distributions between the mandatory and discretionary lane changes were compared by means of the Kolmogorov–Smirnov test. This test was repeated for the fitted distributions of the same decision variable at the same site. The results show that, for the three decision variables associated with gaps in the target lane, the means and distributions between the two types of lane changes are not significantly different. The only variable found to have significant differences in means and distributions is the gap between the subject vehicle and the preceding vehicle in the original lane. This may be because this variable is not an important input in mandatory lane change decisions. This finding provides statistical justification for researchers to develop models with different inputs for mandatory and discretionary lane changes in driver assist systems, in autonomous vehicles, and in microscopic traffic simulation tools.
|
|---|
