基于PCA-SVM的DCT汽车驾驶员起步意图识别

针对目前对于双离合自动变速器(Dual Clutch Transmission,DCT)汽车驾驶员起步过程中意图变化的研究有限,且现有方法的识别准确率不高这一问题,提出了一种基于主成分分析(Principal Component Analysis,PCA)和支持向量机(Support Vector Machines,SVM)的变化起步意图识别模型。根据驾驶员针对车辆反馈的反应时间,将起步过程划分为6个阶段;基于K均值聚类确定各个阶段的缓慢起步、一般起步、紧急起步这3种起步意图的界限;基于主成分分析挖掘出各个阶段起步意图识别的新特征;在此基础上构建6个SVM模型,并利用这6个模型分别对各阶段的起步意图进行识别。经过验证,该模型的平均测试准确率为94.92%,比只利用线性SVM模型高16.89%,且单个模型的平均耗时为0.008 s,能够快速有效地识别出DCT汽车驾驶员的起步意图。     

汽车语控智能电器系统

提出一种语音控制汽车智能电器系统的总体结构框架,包括语音识别和驾驶员意图识别的运算平台及汽车智能电器系统。在车载环境下,利用规范模式和有限状态机理论对驾驶员的自然语言进行分析,达到意图识别和语音控制的目的。引入抗干扰设计方法后系统对驾驶员意图识别的正确率有明显提高。最后通过试验系统验证了利用驾驶员语言进行汽车电器控制的可行性。     

考虑从众心理的城市居民购车意愿分析

正确并充分了解城市居民购车意愿,对于引导合理的私人小汽车需求,继而缓解道路交通压力具有重要意义.基于城市居民购车意愿问卷调查数据,将购车意愿分为3 个等级.考虑从众心理因素,运用多元Logistic 回归分析方法,分析比较不同类型城市居民购车意愿特性.结果表明,总体上从众心理、收入因素对购车意愿产生显著影响,年龄、性别、家至单位是否有地铁等因素影响不显著;通过对比有无小汽车限购政策城市,发现限 购政策的实施会刺激24-45 周岁年龄段2 年内购车意愿,指出相比强制性措施,应倡导引 导型策略,即通过发展公共交通来抑制购车意愿.     

The impact of outcome framing and psychological distance of air pollution consequences on transportation mode choice

Recent years, air pollution phenomenon has become one of the crucial problems of Tehran, Iran. Due to main political and economic role of Tehran, population of this metropolis is high and increasing. Urban transportation of this highly populated city contributes more than 70% of air pollution problem in this city. Although a number of urban transport developments, policy measures and regulations have been employed, Tehran’s air pollution has remained crucial thus far. Finding ways to encourage individuals to behave more sustainable can be considered as a substantial approach of tackling environmental problems such as air pollution, since it can be highly cost-effective and fast. This research attempt to evaluate the impacts of two factors of outcome framing and psychological distance of air pollution on citizen’s willingness to behave environmental friendly, particularly to change the travel mode choice. Results illustrate that communicating the consequences of air pollution can provoke individuals’ to act more environment friendly or in particular to change their intention for using more sustainable mode of transportation. Framing the positive consequences of mitigating air pollution take precedence over framing the negative consequences. Moreover the gains of mitigating air pollution have an impact on the willingness to use of bicycle and bus. Results also show that decreasing the psychological distance of air pollution in order to make manipulated frame more personally relevant has no significant impact on respondents.     

Six years of CO2-based tax incentives for new passenger cars in The Netherlands: Impacts on purchasing behavior trends and CO2 effectiveness

There is growing evidence that consumers respond more effectively to upfront price signals, such as vehicle purchase taxes and feebate policies, and to tax incentives that are more salient than others, such as company car taxes graded by CO₂ emissions. This paper examines tax changes in The Netherlands, which are among the most stringent and most salient in Europe, and assesses the ex-post purchasing impacts and CO₂ effectiveness of six years of CO₂-based tax incentives for low-carbon cars in The Netherlands. Dutch tax incentives resulted in 13 g/km, or 11% lower average CO₂ emissions in 2013. The Netherlands has moved from the 12th position before the tax changes in 2007 to become Europe’s number one in terms of the lowest average new car CO₂ emissions and highest share of electric vehicles in 2013. Tax incentives for new cars sold between 2008 and 2013 have resulted in 4.6 million tons of potential lifetime CO₂ abatement at the cost of a drop in tax revenues of 30–50%. However, when corrected for the Dutch policy-induced increasing real-world fuel-economy shortfall and leakage of carbon reduction potential through vehicle export of low-carbon cars, only 3.5 million tons or 75% of the CO₂ reduction remains. CO₂-based tax incentives for company cars seem to have contributed the most to the observed turnaround in purchasing behavior towards lower CO₂-emitting passenger cars.     

Consumer purchase intentions for electric vehicles: Is green more important than price and range?

In view of global warming and climate change, a transition from combustion to electric vehicles (EVs) can help to reduce greenhouse gas emissions and improve air quality. However, high acquisition costs and short driving ranges are considered to be main factors which impede the diffusion of EVs. Since electricity needs to be produced from renewable energy sources for EVs to be a true green alternative, the environmental performance of EVs is also presumed to be an important factor. This paper investigates the role of environmental performance compared to price value and range confidence regarding consumer purchase intentions for EVs. To develop our hypothesis, we interview 40 end-user subjects about their beliefs toward EVs. Then, we perform 167 test drives with a plug-in battery EV and conduct a survey with the participants to test the hypothesis. Results of a structural equation modeling support the hypothesis that the environmental performance of EVs is a stronger predictor of attitude and thus purchase intention than price value and range confidence.     

高速场景相邻前车驾驶行为识别及意图预测

相邻前车的驾驶行为会影响后车,因此先进的辅助驾驶系统需具备识别前车驾驶行为的能力. 对高速场景下相邻前车换道行为进行研究,分别提出双层连续隐马尔可夫模型-贝叶斯生成分类器(CHMM-BGC),以及基于双向长短时记忆网络(Bi-LSTM)的行为识别模型和意图预测模型. 采用自然驾驶数据集对模型的有效性进行测试验证. 实验分析表明：基于Bi-LSTM的行为识别模型相较于双层CHMM-BGC在平均识别率上提升了11.24%,两种行为识别模型均可在相邻前车换道过程的早期阶段识别换道行为;考虑相邻前车与周围环境车辆的交互作用,可使模型具有预测性,两种意图预测模型均可在车辆换道时刻前预测到驾驶人换道意图. 模型仿真计算时间可满足系统的实时性需求,为本车驾驶人预留出反应时间,为预测周围车辆行驶轨迹研究提供支持.     

高铁服务质量与乘客再搭乘意愿的实证分析

为提高高铁服务质量,从而获得更多的客流,将列车运营服务质量、列车响应性服务质量、车站服务质量作为外生潜在变量,再搭乘意愿作为内生潜在变量,构建结构方程模型,探讨 高铁服务质量和乘客再搭乘意愿之间的关系,进而通过问卷调查搜集数据,利用AMOS20.0 软件对初始模型和实证数据进行验证。研究结果表明：列车运营服务质量、车站服务质量、列车响应性服务质量对高铁乘客再搭乘意愿具有正向影响,列车运营服务质量对再搭乘意愿的影响程度最 大,其次是车站服务质量,最后是列车响应性服务质量,并且列车响应性服务质量对列车运营服务质量、车站服务质量具有不同程度的影响。最后,根据研究结果提出：第一,从乘客需求出发,坚持乘客至上的理念,加强各项配套设施的完善,提高列车运营服务质量;第二,改善车站候车环境,完善车站设施设备,提高车站服务质量;第三,加强与乘客的交流沟通,提高工作人员的服务水平,提高列车响应性服务质量,从而达到提高乘客再搭乘意愿的目的。     

智能船舶的反应型避碰规划研究

对于多船会遇场景而言,在反应型避碰规划中考虑操船者意图具有明显的优越性,可在紧迫局面下更合理地执行《1972年国际海上避碰规则》（即COLREGs）的背离规则。对此,提出一种基于隐马尔可夫模型的操船意图预报方法,将生成的目标船意图预报结果与经典速度障碍法相融合,产生更合理的碰撞危险评估结果,据此建立的反应型避碰规划算法比经典速度障碍法更安全。依托案例三体船完成的自主避碰试验,验证所提改进算法的有效性、可靠性和实时性,为智能船舶的自主避碰研究提供参考。     

基于多特征融合的行人意图以及行人轨迹预测方法研究

行人作为重要的交通参与者,其行走意图和轨迹预测对智能驾驶汽车的决策规划具有重要意义。基于注意力机制增强的长短时记忆(Long Short-term Memory, LSTM)网络,设计一种多特征融合的行人意图以及行人轨迹预测方法。该方法通过融合行人骨架和头部方向特征,以加强行人运动特征的表达,并将融合特征作为意图预测网络输入,继而得到行人意图;由于行人运动具有不确定性,将行人意图预测类别和历史轨迹坐标的联合向量作为行人轨迹预测网络的输入,以期生成更为精确的轨迹预测结果。此外,在轨迹预测网络中引入注意力机制,以加强LSTM对各个时刻编码向量的有效利用,从而提高网络的行人轨迹预测性能,并基于Daimler数据集进行训练及验证。研究结果表明：所提出的多特征意图预测方法准确率可达96.0%,优于基于骨架单特征的意图预测网络;在预测时域为1 s的情况下,预测轨迹的位置均方根误差为347 mm,相较于恒速度(Constant Velocity, CV)模型、交互多模型(Interacting Multiple Model, IMM)、常规LSTM等基线方法均有明显的提升;在实际场景分析中,提出的方...