Liu Xinhua, Mei Huafeng, Lu Huachang, Kuang Hailan, Ma Xiaolin
School of Information Engineering, Wuhan University of Technology, Wuhan 430070, China.
Sensors (Basel). 2017 Mar 20;17(3):633. doi: 10.3390/s17030633.
Recognizing how a vehicle is steered and then alerting drivers in real time is of utmost importance to the vehicle and driver's safety, since fatal accidents are often caused by dangerous vehicle maneuvers, such as rapid turns, fast lane-changes, etc. Existing solutions using video or in-vehicle sensors have been employed to identify dangerous vehicle maneuvers, but these methods are subject to the effects of the environmental elements or the hardware is very costly. In the mobile computing era, smartphones have become key tools to develop innovative mobile context-aware systems. In this paper, we present a recognition system for dangerous vehicle steering based on the low-cost sensors found in a smartphone: i.e., the gyroscope and the accelerometer. To identify vehicle steering maneuvers, we focus on the vehicle's angular velocity, which is characterized by gyroscope data from a smartphone mounted in the vehicle. Three steering maneuvers including turns, lane-changes and U-turns are defined, and a vehicle angular velocity matching algorithm based on Fast Dynamic Time Warping (FastDTW) is adopted to recognize the vehicle steering. The results of extensive experiments show that the average accuracy rate of the presented recognition reaches 95%, which implies that the proposed smartphone-based method is suitable for recognizing dangerous vehicle steering maneuvers.
识别车辆的转向方式并实时向驾驶员发出警报对车辆和驾驶员的安全至关重要，因为致命事故往往是由危险的车辆操纵行为引起的，比如急转弯、快速变道等。现有的利用视频或车载传感器的解决方案已被用于识别危险的车辆操纵行为，但这些方法会受到环境因素的影响，或者硬件成本非常高。在移动计算时代，智能手机已成为开发创新型移动情境感知系统的关键工具。在本文中，我们提出了一种基于智能手机中低成本传感器（即陀螺仪和加速度计）的危险车辆转向识别系统。为了识别车辆转向操纵行为，我们关注车辆的角速度，它由安装在车辆中的智能手机的陀螺仪数据表征。定义了包括转弯、变道和掉头在内的三种转向操纵行为，并采用基于快速动态时间规整（FastDTW）的车辆角速度匹配算法来识别车辆转向。大量实验结果表明，所提出的识别方法的平均准确率达到了95%，这意味着所提出的基于智能手机的方法适用于识别危险的车辆转向操纵行为。
