基于换道意图识别的高速公路换道驾驶行为预测研究

为了预测高速公路换道驾驶行为,该研究在阐述高速公路换道意图、换道场景等基本概念的基础上,提出一种换道意图识别模型。在该模型中,引入BiLSTM、注意力机制、观测序列,以提升模型预测精准性。基于此,该文收集了某公路管理局提供的数据集中101车道的7:55—8:40时段内车道轨迹数据,对模型的预测结果进行试验分析。结果表明：该文提出的模型能够有效预测高速公路车辆换道意图;在0.5s内,预测结果精准性为98%;0.5s～3s的预测结果精准性为93%。     

车联网环境下车辆博弈换道协作策略

交互协作驾驶为提高车辆换道安全和效率提供了一条有效途径,然而可靠的人机协作(HMC)机制是其能有效运行的关键.基于博弈均衡思想,提出一种车联网环境下的车辆博弈换道协作策略.在捕捉和分析驾驶员不同换道类型的行为特征及驾驶期望的基础上,建立车辆换道协作博弈模型,通过求解模型的纳什均衡获取最优换道协作决策方案.特别地,设计了...     

驾驶员-智能体交通流混合仿真系统的建立

将交通流微观仿真系统和虚拟驾驶系统结合,提出并实现了驾驶员-智能体交通流混合仿真系统.介绍了混合系统开发的思路和多机数据传输方法,从平缓换道、转向灯闪烁和系统时间校准三个方面对混合系统进行了改善.最后利用混合系统对微观仿真模型的准确性进行了检验,结果表明在允许误差范围为5％时,跟驰模型可以准确地预测跟驰行为,换道模型的准确性仍然有待于提高.     

短时睡眠剥夺对驾驶人驾驶操作绩效的影响

目的本研究探索24 h内睡眠剥夺(SD)对驾驶人警觉性、驾驶肢体操作的影响。方法本研究在驾驶模拟器中设计了5个不同水平睡眠剥夺时间的实验,并对18名年轻驾驶人(平均24.7岁)展开仿真模拟实验,并通过驾驶模拟器采集他们的驾驶操作数据。结果在24 h内,随着睡眠剥夺时间增加,驾驶人的警觉性先降低后有所提升,但仍低于起床后3 h状态。驾驶人的驾驶肢体操作随着睡眠剥夺时间先变好后变坏,并且在睡眠剥夺16 h后驾驶人更加倾向于简单的动作。此外,驾驶人车道保持能力和换道次数呈现从差变好再变差的趋势,在睡眠剥夺16 h时达到最优。结论驾驶员的警觉性整体下降,而肢体操作能力在16 h后逐渐下降,但是随着清醒时间增长驾驶人动作的灵活性、驾驶人车道保持方面和驾驶人警觉性在24 h睡眠剥夺后是否还能增加,仍需进一步研究。     

经验因素与车道类型对汽车驾驶员变道决策的影响

目的考察驾驶经验与车道类型对驾驶员变道决策的影响以及不同经验驾驶员的眼动注意搜索模式。方法使用Tobii眼动仪和驾驶场景视频收集了35名驾驶员的变道决策反应时、变道频率及眼动数据。结果驾驶经验与车道类型对驾驶员变道决策反应时交互作用显著;驾驶经验、车道类型与兴趣区对驾驶员总注视时间交互作用显著,同时驾驶员前方视野总注视时间对其边侧车道变道决策反应时有负向预测作用。结论中间车道条件下,实习驾驶员变道决策反应时更慢;边侧车道条件下结论相反;中间车道条件下,实习驾驶员对于侧视镜的总注视时间更长;边侧车道条件下,实习驾驶员对于前方视野的总注视时间更长;边侧车道条件下,实习驾驶员对于前方视野注视时间越长,变道决策反应时越快,这可以被视为实习驾驶员变道决策反应时更快的主要视觉注意机制。本研究对驾驶员改善变道行为和安全驾驶具有重要意义。     

驾驶模拟器培训的有效性实验研究

为了探究驾驶模拟培训的有效性,针对小型客车(C1车型)驾驶培训,进行了"模拟与实车"混合培训实验。结合驾驶培训新大纲的要求,以国内常见的驾驶模拟器为辅助培训工具,通过实验组与对照组在驾校进行的培训实验和相关考核,确定了模拟培训的有效性及相关参数。研究表明,在培训效果基本等同的前提下,采用驾驶模拟器可以减少场内实车培训时间24%,道路培训时间46%;在此基础上,提出了小型车模拟培训在驾驶培训中的时间安排和所占比例,为我国相关部门提供依据和参考。     

飞机着陆数值仿真及机场道面动载特性研究

为研究飞机着陆滑跑过程民用机场道面动荷载特性,以Boeing737-800机型为例,基于动力学仿真软件VI-Aircraft,建立了机身、起落架及轮胎三维数值仿真模型,根据某机场道面实测平整度数据创建道面仿真模型,形成了一套考虑气动力变化特性的飞机着陆冲击仿真方法,并通过相关起落架系统落震试验以及飞机-地面运动学理论解析两方面验证了仿真方法的可靠性。此外,系统讨论了各类着陆状态参数对道面动载特性的影响,明确了不同着陆状态参数影响下道面动载系数量化取值范围,揭示了各着陆状态参数对道面动载响应的影响规律及影响机理。研究结果表明：随着陆质量、接地速度及滚转角增大,道面动载响应显著增强;随着陆航向速度增大,道面动载响应明显减小;而随俯仰角增大,道面动载响应整体呈现波动减小的趋势。飞机着陆过程中道面动载系数敏感性因素从大到小依次为：航向速度、着陆质量、接地速度、滚转角与俯仰角,充分考虑各着陆状态参数影响,一般情况下道面动载系数D_IM分布区间为1.18～1.80。研究成果可进一步拓展用于飞机着陆跑道桥的分析研究。     

车辆换道过程中目标车辆与周围车辆的安全距离

首先对车辆换道的动态与安全进行分析,然后通过建立换道模型的方式,对换到车辆与周围车辆之间的最小安全距离展开计算。     

复合机铝箔换卷中的速度损失研究

目的减少铝箔复合机换卷时的速度降幅及变化,改善降速换卷中容易断幅的问题,从而提高设备的综合效率(OEE)。方法通过对实验数据的5W1H和张力分析,建立复合机铝箔换卷时铝箔张力的动力模型,探讨了张紧辊张力、压辊压力、铝箔偏心度、运动离心力方向角和速度差等影响铝箔张力的因素,并对实际操进行了实验改进。结果铝箔在换卷时速度变化得到改善,改善了断幅的问题。结论调整影响铝箔张力等一系列因素,综合效率(OEE)达到92.1%,提高了生产效率。     

基于仿真的中韩青年驾驶员驾驶绩效对比研究

以驾驶仿真模拟器为平台,从速度(平均速度、最大速度和速度变化)、车道保持(车道位置和车道变化)、险情感知能力(险情点速度、反应时间和碰撞时间)和事故频率四个方面,应用SPSS的独立样本T检验和MannWhitney U检验,对30名中国青年驾驶员和30名韩国青年驾驶员的驾驶绩效进行了对比分析.结果显示,两国青年驾驶员在平均速度、车道保持和险情感知能力上有显著性差异,在最大速度、速度变化和事故频率三方面差异不大;中国青年驾驶员比韩国青年驾驶员平均速度低,车道变化小,车道位置更偏向道路中心线,且处理险情更为谨慎.Pearson相关系数显示驾驶里程与平均速度、最大速度和车道变化有显著的正相关性.     

基于车道线斜率的车道偏离检测

针对结构化道路上运动车辆的车道偏离检测问题,采用单目视觉作为感知手段,分别从针孔模型下摄像机成像的空间几何关系和图像中车道线消失点的位置特点两个方面出发,推导出车辆直行情况下车道偏离率与两侧车道线斜率比之间的简单函数关系,该函数与摄像机内外参数无关。同时完成了在摄像机不同方向角下的车道偏离率测量试验,结果表明由于行车方向瞬时变化引起的摄像机方向角微小改变对车道偏离检测的影响可以忽略。道路现场试验结果表明,上述视觉测量方法得到的车道偏离率与手工实测结果相比,其相对误差小于5%。基于车道线斜率的单目视觉检测方法避开了繁琐的摄像机参数标定过程,达到了较高的车道偏离检测精度。     

Effects of work zone configurations and traffic density on performance variables and subjective workload

This paper investigates the effect of changing work zone configurations and traffic density on performance variables and subjective workload. Data regarding travel time, average speed, maximum percent braking force and location of lane changes were collected by using a full size driving simulator. The NASA-TLX was used to measure self-reported workload ratings during the driving task. Conventional lane merge (CLM) and joint lane merge (JLM) were modeled in a driving simulator, and thirty participants (seven female and 23 male), navigated through the two configurations with two levels of traffic density. The mean maximum braking forces was 34% lower in the JLM configuration, and drivers going through the JLM configuration remained in the closed lane longer. However, no significant differences in speed were found between the two merge configurations. The analysis of self-reported workload ratings show that participants reported 15.3% lower total workload when driving through the JLM. In conclusion, the implemented changes in the JLM make it a more favorable merge configuration in both high and low traffic densities in terms of optimizing traffic flow by increasing the time and distance cars use both lanes, and in terms of improving safety due to lower braking forces and lower reported workload.     

智能车辆视觉系统的障碍物边缘检测

对采集的道路图像中的障碍物进行边缘检测,其过程为:道路图像预处理、二值化、边缘检测和数学形态学处理.同时,在MATLAB7.0中对图像进行仿真.     

车道变换碰撞预警分析及最小纵向安全距离模型的研究

分析了车道变换时车辆的各种运动状态,研究了避免发生碰撞的条件;根据车道变换,计算出不发生碰撞的起始最小纵向安全距离,在此基础之上,建立了最小纵向安全距离模型。     

双向拉伸聚酯薄膜拉展道的影响因素分析

通过检查双向拉伸聚酯薄膜（BOPET）的拉展道的表证，并采用厚度测试仪对双向拉伸聚酯薄膜拉展道进行研究，分析了拉展道的影响因素，并提出了相应的改善的方法。     

透明材料中冲击波速度直接测量技术

本文提出了超高压条件下应用于透明材料中冲击波速度直接测量的成像型速度干涉仪技术.介绍了成像型速度干涉仪技术的光路结构和基本原理,给出全系统的光路图.分析了预热效应产生和冲击波到达时刻的时间顺序,发现方波驱动脉冲平台的前沿到达时刻和X光离化效应出现的时刻相同,冲击波信号到达时刻晚于X光离化时刻,验证了冲击波到达窗口材料时形成反射面的反射率会突然增高,可以通过该现象确认冲击波到达的时刻.给出了实验结果,分析了条纹间断是由于X光预热效应引起的,并且发现预热效应在激光脉冲平台期前沿进入靶腔后开始,平台期后沿结束后约400 ps减弱到不影响冲击波信号探测,通过理论分析和实验数据比对的方法,确认了蓝宝石当中的减速曲线.给出了加蓝宝石窗口后的测速公式,经过实验对比,确认了测速公式的正确性.     

Powered two-wheeler drivers’ crash risk associated with the use of bus lanes

Context

Authorizing powered two-wheeler drivers to drive in lanes reserved to buses is a measure that is sometimes mentioned to improve mobility conditions for these users. But what effect would this measure have on the safety of these users and on the safety of the other users with whom they share the traffic space?

Objective

The objective of this study is to contribute elements to help answer this question. More precisely, the objective is to estimate the risk of having an accident per kilometer driven by powered two-wheeler drivers who drive in bus lanes and to compare this risk with that of powered two-wheeler drivers who drive in general traffic lanes.

Method

Using the bodily injury accidents recorded by the police over two years on 13 roads in the city of Marseille and a campaign of periodical observations of powered two-wheeler traffic, we estimated the risk per kilometer driven by powered two-wheeler drivers who drive in bus lanes and compared it with that of drivers who do not drive in them.

Results

The results show that the risk for powered two-wheeler drivers who drive in bus lanes of being involved in a bodily injury accident is significantly higher than the risk run by drivers who drive in general traffic lanes. For the 13 roads studied, it is on average 3.25 times higher (95% CI: 2.03; 5.21).

Conclusion

In the current situation, powered two-wheeler drivers who drive in bus lanes are less safe than those who drive in general traffic lanes. The analysis of police reports suggests that part of this increased risk comes from collisions between automobile drivers turning right and powered two-wheelers driving in the bus lane who continue straight ahead.     

Is the three-foot bicycle passing law working in Baltimore,Maryland?

Maryland (MD) recently became one of fourteen states in the United States to enact a traffic law requiring motor vehicles to pass bicyclists at a distance of greater than three feet. To our knowledge, motorist compliance with the law has never been assessed. This study measured the distance between overtaking motor vehicles and cyclists [e.g. vehicle passing distance (VPD)], to develop baseline metrics for tracking implementation of the three-foot passing law in Baltimore, MD and to assess risk factors for dangerous passes. During September and October 2011, cyclists (n = 5) measured VPD using a previously published video technique ( Parkin and Meyers, 2010). Cyclists logged a total of 10.8 h of video footage and 586 vehicle passes on 34 bicycle commuting trips. The average trip lasted 19.5 ± 4.9 min and cyclists were passed on average 17.2 ± 11.8 times per trip. VPDs of three feet or less were common when cycling in standard lanes (17%; 78 of 451 passes) and lanes with a shared lane marking (e.g. sharrows) (23%; 11 of 47 passes). No passes of three feet or less occurred in bicycle lanes (0 of 88 passes). A multiple linear regression model was created, which explained 26% of the variability in VPD. Significant model variables were lane width, bicycle infrastructure, cyclist identity, and street identity. Interventions, such as driver education, signage, enforcement, and bicycle infrastructure changes are needed to influence driving behavior in Baltimore to increase motorist compliance with the three-foot law.     

Designing a block stacked warehouse for dynamic and stochastic product flow: a scenario-based robust approach

The problem considered in this paper is motivated from a real life case. Conventional models in designing block stacked warehouse assume uniform and deterministic inflow and outflow of products in specific quantities and time intervals. These assumptions would lead to underestimation of space requirement for the case. Also, the possibility of leasing additional warehouse space during seasonal spikes was found to be unexplored in block stacking design literature. In this study, we develop a scenario-based model to address the decisions of warehouse size, lane depth and hiring of temporary additional warehouse capacity when flow of products is stochastic and dynamic in nature. A sequential decision model is presented with future operating scenarios assumed to branch out in the form of a tree. On comparison with conventional block stacking design methods, the model is found to provide robust solution in stochastic and dynamic environments in terms of risk reduction and infeasibility avoidance. Further, the model gives additional flexibility to the designer to trade-off model and solution robustness with predicted cost.