Modeling drivers' speed choice on roads with complex shapes based on the curvature of the preview trajectory. (October 2018)

• Record Type: Journal Article
• Authors: Xu, Jin; Zhao, Jun; Shao, Yiming
• Is Part Of: Simulation. Volume 94:Number 10(2018)
• Page Start: 911
• Page End: 930
• Publication Date: 2018-10
• Subjects: Driver–vehicle–road system -- driving behavior -- desired speed -- target speed -- decisions -- driver model -- intelligent driving; Computer simulation -- Periodicals; 003.3
• DOI: 10.1177/0037549717751633
• Languages: English
• ISSNs: 0037-5497

Abstract

The target speed needs to be determined based on the geometry of the road ahead and current operational state in order to realize automated driving along complex mountain roads. However, existing speed prediction and V 85 models are unable to do so. This paper proposes a speed decision algorithm that is based on the curvature of the preview trajectory. Firstly, the trajectory point in each preview cross-section within the driver's sight window is determined; the curvature of the preview trajectory is calculated and used as input data. Secondly, an objective function is selected to set the speed control pattern: minimum travel time; maximum comfort; minimum deviation from the reference speed; or multiple weighted objectives. Thirdly, the speed value for the preview cross-section is solved by optimizing the selected objective function. As the sight window rolls ahead, the target speed can be determined along the traveling distance. Because the trajectory generated within the boundary of the pavement is used by the driver, parameters such as the change in curve radius, deflection angle, lane width, and spiral length can change the driving boundary, which can change the trajectory characteristics and thus affect the target speed. Therefore, the proposed algorithm is able to determine the speed for roads with complex shapes. The field validation and simulation cases demonstrated that the target speed profile of different driving patterns can be obtained for different combinations of trajectory and speed patterns. Thus, the model can simulate diverse real-world driving behaviors and explain the diversity in driving speeds of different drivers.