Analysis of Right Turning Vehicle Operation Characteristics at Intersections Using Trajectory Data

Abstract

The implementation of human-like control in autonomous vehicles greatly benefits from a thorough investigation of the operation characteristics of right-turning vehicles at intersections, including driving behavior, speed, and path characteristics. But most existing studies consider the right-turning speed, path, and behavior characteristics separately, so the analysis of turning speeds considering different driving styles is rarely combined with the path model. Therefore, based on trajectory data, this paper made a comprehensive analysis. Firstly, conflicts at an intersection were extracted using extended Time-To-Collision (ETTC) as the indicator and the driving behaviors of right-turning vehicles were divided into aggressive and normal styles. Secondly, a binary Logit model was built to study the significant influencing variables on driving behavior choice after relevant variables were screened by Random Forest. Finally, the relationship between path and speed was considered, and the influencing factors of speed curves under different driving styles were analyzed. The results show that the minimum speed, the mean speed, the mean speed in the lateral direction, and the number of general conflict points around the right-turning vehicles have the greatest influence on driving behavior choice. The acceleration and deceleration stages of aggressive and normal speed curves are affected by different factors, such as vehicle types, the speed into turning stage, and the minimum ETTC value. In this paper, the operation characteristics of right-turning vehicles at an intersection were analyzed as a whole, and the conclusions can lay a foundation for the development of human-like autonomous driving and the construction of path planning models.