Human driving data-based design of a vehicle adaptive cruise control algorithm

This paper presents a vehicle adaptive cruise control algorithm design with human factors considerations. Adaptive cruise control (ACC) systems should be acceptable to drivers. In order to be acceptable to drivers, the ACC systems need to be designed based on the analysis of human driver driving behaviour. Manual driving characteristics are investigated using real-world driving test data. The goal of the control algorithm is to achieve naturalistic behaviour of the controlled vehicle that would feel natural to the human driver in normal driving situations and to achieve safe vehicle behaviour in severe braking situations in which large decelerations are necessary. A non-dimensional warning index and inverse time-to-collision are used to evaluate driving situations. A confusion matrix method based on natural driving data sets was used to tune control parameters in the proposed ACC system. Using a simulation and a validated vehicle simulator, vehicle following characteristics of the controlled vehicle are compared with real-world manual driving radar sensor data. It is shown that the proposed control strategy can provide with natural following performance similar to human manual driving in both high speed driving and low speed stop-and-go situations and can prevent the vehicle-to-vehicle distance from dropping to an unsafe level in a variety of driving conditions.     

基于SOM神经网络的多工况驾驶风格识别

不同的驾驶员对车辆的各项性能可能有个性化地要求,因此有必要对驾驶风格的分类与识别问题进行研究。首先在驾驶模拟器上采集不同驾驶员在多工况下的数据,利用主成分分析法选取驾驶员在各个工况下的特征参数,SOM神经网络分别对起步、加速及制动工况下的驾驶数据进行了聚类分析,然后以驾驶风格聚类分析结果为基础,建立了基于SOM神经网络的驾驶风格识别系统,该系统可根据驾驶员驾驶历史数据来判断其驾驶风格,最后以某一温和型驾驶风格识别结果为例验证了系统的合理性。     

个体差异对转向指标疲劳辨识能力的影响分析

驾驶人个体差异是影响疲劳驾驶辨识准确性的重要因素。为了探究个体差异与基于转向行为的疲劳辨识效果之间的关系，量化个体差异对转向特征指标疲劳辨识能力的影响程度，通过自然驾驶试验，采集被试在清醒状态和疲劳状态下的真实驾驶行为数据，结合观察员问询打分和被试面部视频得到疲劳水平信息。设置双层滑动时间窗对每位驾驶人的自然驾驶行为数据进行处理，挖掘出9个疲劳驾驶转向特征指标。对每位驾驶人清醒和疲劳状态下的指标样本进行Wilcoxon检验，用Wilcoxon检验的|Z|值表示指标对驾驶疲劳的分类性能。以清醒和疲劳状态下指标有显著差异的被试数目最多为优化目标，得到指标最优的双层时间窗设定值。将|Z|值最大的被试逐个与其他被试两两组合，对清醒和疲劳状态下混合两被试指标样本数据进行Wilcoxon检验，得到被试组合指标的|Z|值。计算两被试的综合个体差异值，基于线性模型拟合两被试组合Wilcoxon检验的|Z|值和个体差异值，以拟合直线的斜率绝对值|k|量化个体差异对指标疲劳辨识能力的影响。研究得到基于自然驾驶行为数据的9个疲劳驾驶转向特征指标的最优时间窗，发现指标对疲劳驾驶的分类性能存在个体差异，并且指标的疲劳辨识能力会随个体差异增加而降低，进而影响基于转向行为指标疲劳辨识的准确性，其中方向盘转角下四分位标准差（X_q1std）的斜率绝对值最大（1.17），方向盘转角标准差（X_jstd）的斜率绝对值最小（0.44），疲劳辨识能力受个体差异影响最大和最小的指标分别是X_q1std和X_jstd。研究结果可为利用自然驾驶行为数据的疲劳驾驶特征提取及考虑个体差异的疲劳驾驶建模提供参考。     

Development of an errorable car-following driver model

An errorable car-following driver model is presented in this paper. An errorable driver model is one that emulates human driver’s functions and can generate both nominal (error-free), as well as devious (with error) behaviours. This model was developed for evaluation and design of active safety systems. The car-following data used for developing and validating the model were obtained from a large-scale naturalistic driving database. The stochastic car-following behaviour was first analysed and modelled as a random process. Three error-inducing behaviours were then introduced. First, human perceptual limitation was studied and implemented. Distraction due to non-driving tasks was then identified based on the statistical analysis of the driving data. Finally, time delay of human drivers was estimated through a recursive least-square identification process. By including these three error-inducing behaviours, rear-end collisions with the lead vehicle could occur. The simulated crash rate was found to be similar but somewhat higher than that reported in traffic statistics.     

Trip characteristics impact on the frequency of harsh events recorded via smartphone sensors

The Internet of Things (IoT) constantly offers new opportunities and features to monitor and analyze driver behavior through wide use of smartphones, effective data collection and Big Data analysis, resulting in assessment and improvement of driver behavior and safety. The objective of the present study is to investigate the impact of detailed trip characteristics on the frequency of harsh acceleration and harsh braking events through an innovative smartphone application developed within the framework of BeSmart project. A 200-driver naturalistic experiment spanning 12 months is carried out since July 2019. During the first two months, participants were asked to drive in the way they usually did, without receiving any feedback on their driving behavior from the application. Over the subsequent two months, participants were provided with personalized feedback, a trip list and a scorecard regarding their driving behavior, allowing them to identify their critical deficits or unsafe behaviors. Some of the most important risk factors, such as speed and driving above the speed limit, usage of mobile phone while driving and harsh events (acceleration and braking) are recorded through the application and subsequently analyzed. Generalized Linear Mixed-Effects Models were fitted to the trips of car drivers who made frequent trips for both experiment phases in order to model the frequencies of harsh events. Results indicate that maximum speed, the percentage of speeding duration and total trip duration are positively correlated with both harsh acceleration and harsh braking frequencies. On the other hand, the exposure metric of total trip distance was found to be negatively correlated with both harsh event types. A small positive correlation of the percentage of mobile use duration with harsh accelerations was also detected.     

Hardware and software for collecting microscopic trajectory data on naturalistic driving behavior

This article presents a method to collect naturalistic microscopic longitudinal vehicle trajectory data with a modest budget. The drivers studied are not aware that they are participating in an experiment; hence one can collect naturalistic driving behavior. This article presents the hardware and software developed, and we include a detailed example of a particular case study that was conducted with data collected from the system. The case study examines drivers' willingness to accept very short headways, and casts that behavior in light of their subsequent lane-changing decisions. The data show a statistically defensible connection between these behaviors. These phenomena are not new, but highlight the importance of the data quality and of observing naturalistic driving behavior, and this article demonstrates a method to calibrate specific parameters related to the behavior.     

驾驶人驾驶决策机制遵循最小作用量原理

为提升智能汽车的自主决策能力，使其能够学习人的决策智慧以适应复杂多变的道路交通环境，需要揭示驾驶人决策机制。首先通过对自然驾驶数据的分析，发现在车辆行驶过程中能够反映驾驶人决策行为的主要运动特征参数存在极值现象，而产生极值现象的内在动因是驾驶人遵循“趋利避害”的基本决策机制，即驾驶过程中驾驶人力图实现机动性和安全性综合性能最优。受自然界包括物理和生物行为上的众多极值现象遵循最小作用量原理的启发，提出驾驶人决策机制遵循最小作用量原理的假设。随后建立抽象描述驾驶过程的物理模型，并提出最小作用量决策模型（Least Action Decision-making Model，LADM），通过与传统驾驶决策模型（经典跟车模型和换道模型）对比，分析结果显示LADM模型更具通用性。最后开展了实车试验，采集20名驾驶人在自由行驶、跟车行驶和邻车切入3种工况下的试验数据，分析计算并检验了不同驾驶人行车过程的理论最小作用量和实际作用量。试验结果表明：驾驶人在驾驶过程中的实际作用量与最小作用量之间无显著性差异，体现出驾驶人在行车过程中对安全和高效具有共性追求，验证了驾驶人决策机制遵循最小作用量原理。     

老年驾驶人的驾驶自我调节研究述评

随着社会人口老龄化的发展，老年驾驶人的占比逐年增加，提升老年人的驾驶安全性对于其安全自主出行和公共交通安全均具有重要意义。驾驶自我调节是老年驾驶人为适应身体、认知功能变化而对驾驶行为做出的主动调整，是其提升驾驶安全性、延长驾驶生命和维持自主行动能力的有效补偿策略。通过对已有关于老年驾驶人的驾驶自我调节研究进行系统回顾，介绍了老年驾驶人的驾驶自我调节行为的定义及其表现，归纳分析了其驾驶自我调节行为的影响因素及产生机制，在此基础上总结了现有研究的局限，并指出了未来进一步研究的主要思路和方向。对文献的梳理和分析表明：老年驾驶人的驾驶自我调节包括减少驾驶频率和回避具有挑战性的驾驶情境2种主要形式，并可分为策略性、战术性和生活目标性自我调节3种不同的层次水平；驾驶自我调节是一个复杂的过程，社会人口因素、生理健康和功能状况、心理因素等均可对其产生影响；驾驶自我调节的产生机制可以被概括为是个体从认知到态度改变，再到形成调节行为意向，直至最终发生驾驶行为改变的动态决策过程。未来对老年驾驶人的驾驶自我调节行为研究应更进一步将客观驾驶行为数据、医疗机构数据与驾驶人主观自我报告相结合，适当开展追踪研究，考察驾驶自我调节行为随年龄的发展变化趋势，深入探究驾驶自我调节的产生机制及其在降低事故发生和提升驾驶安全性方面的作用。     

基于主成分分析和学习向量量化神经网络的制动工况路面识别与验证

开展车辆制动时路面类型识别的研究，提出一种基于主成分分析-学习向量量化神经网络 （Principal Component Analysis - Learning Vector Quantization，PCA-LVQ） 的制动工况路面识别方法。利用主成分分析对多维度驾驶数据降维处理，提取能表征路面特征的主要成分，采用学习向量量化神经网络对降维处理后的驾驶数据进行训练，并用于路面特征分类，使用制动工况下实车试验数据和硬件在环仿真数据进行验证。结果表明，所提出的 PCA-LVQ算法能准确识别路面类型特征，路面识别的精度达到 97%，与传统 BP神经网络的路面类型特征识别精度提升 7%；同时，在不同车速下，基于PCA-LVQ算法也能较准确地识别路面类型特征。     

Normal and risky driving patterns identification in clear and rainy weather on freeway segments using vehicle kinematics trajectories and time series cluster analysis

Enhancing traffic safety on freeways is the main goal for all transportation agencies. However, to achieve this goal, many analysis protocols of network screening models need to be improved through considering human factors while analyzing traffic data. This paper introduces one on the new analysis protocol of identifying and discriminating between normal and risky driving in clear and rainy weather. The introduced analysis protocol will consider the effect of human factors on updating the networking screening process of identifying hotspots of crash risk. This paper employs the Second Strategic Highway Research Program (SHRP2) Naturalistic Driving Study (NDS) data to investigate the behavior of normal and risky driving under both rainy and clear weather conditions. Near-crash events on freeways, which were used as Surrogate Measure of Safety (SMoS) for crash risk, were identified based on the changes in vehicle kinematics, including speed, longitudinal and lateral acceleration and deceleration rates, and yaw rates. Through a trajectory-level data analysis, there were significant differences in driving patterns between rainy and clear weather conditions; factors that affected crash risk mainly included driver reaction and response time, their evasive maneuvers such as changes in acceleration rates and yaw rates, and lane-changing maneuvers. A cluster analysis method was employed to classify driving patterns into two clusters: normal and risky driving condition patterns, respectively. Statistical results showed that risky driving patterns started on average one second earlier in rainy weather conditions than in clear weather conditions. Furthermore, risky driving patterns extended in average three seconds in rainy weather conditions, while it was two seconds in clear weather conditions. The identification of these patterns is considered as a primary step towards an automated development that would distinguish between different driving patterns in a Connected Vehicle CV environment using Basic Safety Messages (BSM) and to enhance the network screening analysis for increased crash risk hotspots.     

驾驶行为与驾驶风险国际研究进展

驾驶人在驾驶车辆的过程中总会面临由自身或外界条件所带来的或高或低的风险，即驾驶风险，通过对驾驶风险进行识别、分析及评估是对风险进行管理的有效对策，明确由人为因素（即驾驶人个体特征及驾驶行为）所带来的驾驶风险并对驾驶人进行安全管理尤为重要。为了全面了解各类危险驾驶行为和各种驾驶人群体的驾驶风险行为研究进展，对驾驶风险领域重点问题进行了总体概述。从驾驶人个体特征及驾驶行为的角度出发，探究了驾驶风险领域目前的研究现状，并利用科学知识图谱展示驾驶风险领域研究的发展进程与结构关系。通过Web of Science核心合集数据库获取了3 406篇在1986~2020年（截至2020年2月29日）间出版的驾驶风险研究相关英文文献，共涵盖8 684位作者及6 018个关键词，基于科学知识图谱对该领域文献进行梳理与分析。结果表明：驾驶风险领域的国外研究在驾驶人选择方面主要从年轻驾驶人、老年驾驶人、新手驾驶人及职业驾驶人的角度进行切入，重点围绕酒驾、药驾、分心驾驶及疲劳驾驶等主题开展研究。与国外研究相比，中国在分心驾驶、疲劳驾驶领域的研究相对丰富，而针对酒驾、药驾的研究试验手段较为单一，研究不够全面；在研究对象的选取上，有必要进一步增加老年驾驶人及新手驾驶人的深入研究，包括老年驾驶人适驾性评估与教育培训，以及新手驾驶人驾照分级制度的可行性探索。在研究方法方面，国外常见研究方法包括问卷调查、驾驶模拟器试验、实车试验以及自然驾驶研究等，而中国在自然驾驶研究领域尚未充分开发利用；未来应考虑多种方法相结合并从不同角度促进对驾驶行为及驾驶风险的全面理解。     

线形空间几何突变对曲线路段车道偏移的影响

为了探讨道路线形变化对侧碰、刮擦等侧向安全事故的影响，以三维空间线形的曲率和挠率作为公路线形几何特征描述参数，以车道偏移量作为侧向行车安全的表征指标，剖析了线形在空间层面发生的几何突变对车道偏离的影响。在山区高速公路开展实车试验，采用侧向行车视频记录连续的车道偏移，进行图像距离与实际距离的标定，并通过图像识别技术自动读取连续的车道偏移曲线，从中获取最大车道偏移作为分析变量。采用单因素方差分析方法，对新手驾驶人和熟练驾驶人在线形空间几何特征不同的曲线路段所表现的最大车道偏移结果展开统计分析和检验。分析结果表明：当空间曲率突变超过一定的临界值时，空间曲率突变与最大车道偏移显著正相关；挠率突变对车道偏移产生的影响主要取决于线形扭转的方向，当线形扭转方向与路拱横坡反向时，会明显加剧最大车道偏移；而线形扭转方向与路拱横坡同向时，会降低最大车道偏移但降低效果不明显；熟练驾驶人的最大车道偏移小于新手驾驶人，这种现象在空间曲率突变较大和挠率突变不利的路段尤为明显。研究结论可为公路线形安全性评价、线形设计优化和路侧安全改善提供参考。     

基于一维卷积神经网络的驾驶人身份识别方法

近年来，智能网联汽车（ICV）已成为智能工业时代最有前景的发展方向。作为现代移动的重要模式，ICV的设计和开发越来越强调个性化需求。提出一种仅使用车载CAN总线行车状态数据，基于深度学习的驾驶人身份识别通用框架。首先采集20名驾驶人在固定试验路线下，包括不同道路类型、不同交通条件下的自然驾驶行车状态数据集；其次对9种类型的CAN信号行车数据进行数据清洗与重采样，构建数据样本集。搭建了由卷积层、池化层、全连接层、SoftMax层构成的一维卷积神经网络（1-D CNN）驾驶人身份识别模型，并且使用Adam算法、L2正则化、Dropout、小批量梯度下降等方法对模型性能进行优化。算法验证过程中，探讨了模型卷积核占比、卷积核数量、卷积层层数、全连接层节点规模对模型识别准确率的影响，进而对模型结构参数进行优选。进一步地，将该算法与K近邻（KNN）、支持向量机（SVM）、多层感知器（MLP）等传统机器学习方法及深度学习算法长短时记忆网络（LSTM）进行对比分析，同时探究样本时间窗口大小、样本数据重叠度、驾驶人数量对模型识别结果的影响。在数据时间窗口为1 s、数据重合度80%的条件下，对20名驾驶人进行识别，评价指标宏观F₁分数可达99.1%，表明该模型表现明显优于其他对比模型算法，其对驾驶人身份识别表现稳定，鲁棒性强。     

基于多分类监督学习的驾驶风格特征指标筛选

交通事故与驾驶风格具有强烈的相关性,而驾驶风格的直观体现是驾驶行为.为深入分析驾驶行为与驾驶风格的关联性,探索不同驾驶风格群体之间的差异,筛选驾驶风格分类与识别影响因素,建立驾驶风格识别模型并验证有效性.依托车联网实验数据,利用K-means++算法对驾驶员样本数据集进行驾驶风格聚类,设计支持向量机-递归特征消除(SV...     

Characteristic analysis for cognition of dangerous driving using automobile black boxes

Automobile black boxes are devices that collect information regarding vehicle operation and the driver’s operating situation in the case of a traffic accident. The information collected from the automobile black box, which can also be used during normal driving, can provide information about dangerous driving cognition. This study was designed to analyze characteristics of dangerous driving data and build a dangerous driving cognition system as follows. First, dangerous driving is divided into four types by considering the vehicle’s movement, such as acceleration, deceleration, turning and statistical data of traffic accidents. Second, dangerous driving data were collected by vehicle tests using the automobile black box, and characteristics of the driving data were analyzed to classify dangerous driving. Third, a standard threshold was chosen to recognize dangerous driving, and an algorithm of dangerous driving cognition was created. Finally, verification was conducted by vehicle tests with automobile black boxes embedded with the developed algorithm. The presented recognition methods of dangerous driving can be used for on/off-line management of drivers and vehicles. Scientific traffic accident databases can be built with this driving and accident information, and can be used in various industrial areas.     

跟驰过程中驾驶员认知结构模型的建立

在道路交通4要素中(人、车、路、环境),人以其主动性和智慧性起着支配作用,是其中的主体要素。基于认知心理学的有关知识,论文采用因子分析法对五轮仪实验系统观测到的车辆跟驰数据进行分析,确定了对车辆跟驰信息提取过程有独立作用的4个因素,相应地将驾驶员认知过程划分为4个阶段,建立了车辆跟驰过程的驾驶员认知结构模型。为驾驶行为研究和车辆跟驰模型的建立提供了理论基础。     

基于小波和粒子群算法的HEV行驶状况辨识方法研究

针对混合动力汽车(HEV)行驶状况(道路坡度和整车载荷)变化难以有效识别,导致驱动系统控制策略不能有效满足驾驶员意图问题,以混联式HEV为研究对象,提出了基于小波滤波和粒子群算法的HEV行驶状况辨识方法。首先建立了汽车行驶状况辨识模型,采用最小二乘法确立了优化目标函数,其次研究了基于小波滤波和粒子群算法的HEV行驶状况辨识原理,最后进行了行驶状况粒子群智能算法辨识试验。在采集实车数据的基础上,对实车数据进行小波滤波,并运用行驶状况辨识方法对道路坡度和整车载荷进行了辨识,并对辨识结果进行小波滤波,结果表明,试验工况下整车载荷辨识的相对误差绝对平均值为2.71%,道路坡度辨识的相对误差绝对平均值为3.85%,验证了所提出方法的有效性。     

Which factors lead to driving errors? A structural equation model analysis through a driving simulator experiment

As driving error is a main contributory factor of road accidents, its causes and consequences are of great interest in the road safety decision making process. This paper investigates several factors (including driver distraction, driver characteristics and road environment) that affect overall driving error behaviour and estimates a new unobserved variable which underlines driving errors. This estimation is performed with data obtained from a driving simulation experiment in which 95 participants covering all ages were asked to drive under different types of distraction (no distraction, conversation with passenger, cell phone use) in rural and urban road environment, as well as in both low and high traffic conditions. Driving error was then modeled as a latent variable based on several individual driving simulator parameters. Subsequently, the impact of several risk factors such as distraction, driver characteristics as well as road environment on driving error were estimated directly. The results of this complex model reveal that the impact of driver characteristics and area type are the only statistically significant factors affecting the probability of driving errors. Interestingly, neither conversing with a passenger nor talking on the cell phone have a statistically significant impact on driving error behaviour which highlights the importance of the present analysis and more specifically the development of a measure that represents overall driving error behaviour instead of individual driving errors variables.     

螺旋匝道和螺旋桥的小客车行驶速度特征

为明确螺旋匝道和螺旋桥处的驾驶行为模式和汽车运行特征，在涪陵长江一桥、乌江二桥、重庆融侨大道和涪陵金凯环形高架4处地点开展螺旋匝道实车试验，用车载仪器采集自然驾驶状态下的汽车连续行驶轨迹、速度以及周围行驶环境等信息。基于自然驾驶数据，研究螺旋匝道范围内的速度变化模式、幅值特性以及影响因素。研究结果表明：单车道螺旋匝道的速度变化模式多样化，双车道螺旋匝道的行驶速度在整体上维持稳定，匝道范围内的连续升坡和降坡并未导致速度出现趋势性衰减和趋势性升高；螺旋匝道并入主线时，驾驶人在合流鼻之前有明显的、共性的减速行为，这与现行设计标准中的设计假定相反；除涪陵长江一桥之外，其余3处都是下行速度低于上行速度；螺旋匝道设计速度越低，实测速度与设计速度之间的偏离越严重，并且速度幅值离散化，因此不建议使用20 km·h^-1的匝道设计速度；螺旋匝道运行速度与匝道半径成正相关。     

基于轨迹数据的危险驾驶行为识别方法

连续的跟驰行为和换道行为是驾驶行为的主要构成部分，对交通拥挤和交通事故有着重要影响。通过无人机视频拍摄和图像处理方式，提取了曹安公路沿线的2个交叉路口间正常交通流状态下共600条多车高精度轨迹数据。首先，考虑车辆类型对驾驶行为产生直接的影响，分析了大车和小车的车辆轨迹特征变量分布的差异性，包括速度、加速度、碰撞时间倒数、车头时距等，在标记危险驾驶行为的过程中考虑车辆类型的影响。其次，针对不同的车辆类型，利用修正碰撞裕度对跟驰行为和换道行为进行风险性评估，将其划分为安全型和风险型。根据风险型行为发生的顺序以及持续时间，评估驾驶人的整体驾驶状态是否危险，作为危险驾驶行为识别的样本标记。分别利用离散小波变换和统计方法提取车辆轨迹的关键特征参数，为了提高模型识别效率，将关键特征参数进行排序，从而确定最优判别指标；最后，利用轻量梯度提升机（LGBM）算法对危险驾驶行为进行识别，并与随机森林、多层感知器、支持向量机等算法在精度上进行比较。研究结果表明：在上述研究条件下，LGBM算法对危险驾驶行为的理论识别率最高可达93.62%，可以实现基于机器学习算法的危险驾驶行为的高精度自动识别，该结果对于智能驾驶辅助系统的设计、道路交通安全决策的制定具有显著的意义。