Personalized Driver Assistance for Signalized Intersections Using V2I Communication
Personalized Driver Assistance for Signalized Intersections Using V2I Communication
复制标题
DOI：
10.1109/tits.2016.2515023
复制
发表时间：
2016-01
影响因子：
8.5
通讯作者：
V. A. Butakov;Petros A. Ioannou
中科院分区：
文献类型：
--
作者：
V. A. Butakov;Petros A. Ioannou
交叉路口穿越是城市驾驶中一种常见的驾驶行为。交通信号灯和停车标志迫使车辆停止和重新启动，造成沮丧的司机。为了帮助驾驶员接近和通过十字路口，车辆可以配备先进的驾驶员辅助系统（ADAS），该系统利用车辆到基础设施的通信。与集中式交通灯优化相反，车载系统使用交通灯位置和时间信息来找到个人最佳驾驶速度，然后将其传达给驾驶员。速度优化问题的传统方法是考虑到达时间、燃料消耗和排放等参数。然而，要有效地，该系统应考虑到另一个重要因素-驾驶员的偏好和驾驶特性，以提高系统的接受度。在本文中，我们提出了一个个性化的速度优化算法接近和通过信号交叉口，可用于自动驾驶辅助系统。它在驾驶过程中学习个人驾驶员的偏好和特征，并使用这些知识来计算驾驶速度，以提高燃油经济性，减少等待时间，并解决驾驶员的偏好。我们证明了所提出的方法，通过比较系统的建议，不同的偏好和驾驶风格的司机。驾驶员的特征提取从实验车辆上收集的数据。我们利用司机的个人资料，以找到一个最佳的驾驶速度为指定的路线，为每个特定的司机。我们通过模拟来验证该方法，以显示所生成的速度曲线的燃油经济性益处。
Intersection crossing is a frequent driving maneuver in urban driving. Traffic lights and stop signs force the vehicles to stop and restart causing frustration to drivers. To help the driver in approaching and passing intersections, the vehicle can be equipped with an advanced driver assistance system (ADAS) that utilizes vehicle-to-infrastructure communication. As opposed to centralized traffic light optimization, the in-vehicle system uses information on traffic light location and timing to find an individual optimal driving pace, which is then conveyed to the driver. The conventional approach to the pace optimization problem is to consider such parameters as time of arrival, fuel consumption, and emissions. However, to be effective, the system should take into account another important factor-driver's preferences and driving characteristics to improve the acceptance of the system. In this paper, we propose a personalized pace optimization algorithm for approaching and passing signalized intersections that can be used in ADAS. It learns the personal driver's preferences and characteristics during the course of driving and uses this knowledge to calculate driving pace that improves fuel economy, reduces waiting time, and addresses the driver's preferences. We demonstrate the proposed methodology by comparing the system's recommendations for drivers with different preferences and driving styles. Drivers' features were extracted from data collected on an experimental vehicle. We utilize the drivers' profiles to find an optimal driving pace for a specified route for each particular driver. We validate the methodology by simulation to show fuel economy benefits of the generated speed profiles.