Risk Assessment Based on Microscopic Analysis of Vehicle Running Behavior at Lane Reduction Sections Due to Roadwork Regulations on Expressways

Abstract

In this study, we proposed a conflict index n-PIC (normalized-Polygon Index for Conflict) that is calculated by dividing the sum of the side lengths of polygons with each vehicle and the restriction start point as vertices by the number of vertices, for merging and parallelly running vehicles at the upstream end of road work regulation. Using vehicle running trajectory estimation data based on video observation, we showed that the conflict index can be applied to risk assessment of situations where merging is forced at the upstream end of restriction, and continuously before and after merging, in comparison with conventional conflict indexes. Based on the discrimination score of the lane-changing model re-identified using observed data in unregulated sections and the n-PIC value, the high/low risk threshold was set at 30.0m. In addition, a certain degree of suitability was confirmed for classification based on discriminant scores and n-PIC values from 33 cases of unregulated conditions (excluding high traffic flow rates) and 6 cases of high risk under regulated conditions. Furthermore, the difference in the average extreme acceleration values between the 34 low-risk and 28 high-risk cases, classified by n-PIC value, was highly significant for each sign.