【作者】 赵炜华；

【导师】 刘浩学；

【作者基本信息】 长安大学 ， 交通环境与安全技术， 2010， 博士

【摘要】 因人的因素在车辆运行安全中具有重要作用,且驾驶人对环境运动信息感知是操作行为的基础,要保障驾驶人-车辆-道路系统协调安全,必须提高人的可靠度,改善动态环境中的信息认知。在驾驶人所获信息的多个通道中,视觉提供的信息占90%以上。在车辆运行状态下,驾驶人对外界环境信息正确认知,尤其是空间距离辨识,对人的行为决策、道路设施设计等具有重要意义。空间距离判识除受驾驶人生理和心理影响外,环境照度、物体色调等均存在较大影响。本文从驾驶人对地面动态环境空间距离信息加工的生理和心理角度出发,通过实际道路实验,研究分析车辆运行状态下,驾驶人在昼间、黄昏、夜间环境中,对空间距离判识的变化规律及差异。主要研究内容及成果如下：1.系统分析国内外对驾驶人动态视觉信息加工的研究现状,尤其是空间距离判识等相关研究及存在的问题；以认知工程和实验心理学为基础,结合道路交通系统和车辆运行特点,设计了不同环境下的空间距离判识研究实验方案、数据处理和建模方法。2.统计分析了昼间动态实验数据分布特征,获得驾驶人空间距离判识结果。针对不同空间距离和色调障碍物,建立距离判识结果随速度变化的一元线性回归模型及BP神经网络模型；检验并比较不同色调障碍物距离判识差异,利用神经网络建立了距离判识差异变化模型；考虑速度和空间距离的影响,综合红绿色障碍物距离判识结果,利用二元曲面回归和神经网络方法,分别建立了昼间动态空间距离判识变化模型。3.根据黄昏静态距离判识实验结果,建立了不同颜色障碍物判识距离随环境照度变化的一元对数回归模型；考虑黄昏不同颜色的视认特性,检验并比较红绿色障碍物距离判识差异,利用神经网络建立了驾驶人判识结果及判识差异随环境照度和空间距离变化的模型。4.依据夜间驾驶人空间距离判识实验数据,建立了空间距离判识结果随速度变化的负指数函数模型；检验低照度下红绿色障碍物距离判识差异显著度,比较红绿色障碍物间的距离判识差异,并建立了差异随照度和距离变化的BP网络模型；综合红绿色障碍物判识结果,利用泰勒公式建立了判识距离随速度和距离变化的二元二次广义回归模型及神经网络模型；利用相同条件下昼夜距离判识差异,建立了昼夜距离判识差异变化的BP神经网络模型。5.利用眼动仪检测视觉特征变化,记录并研究距离判识过程中定任务的特征参数,分析注视参数变化规律,及距离判识过程中速度与空间距离对动态视觉特征的影响。更多还原

【Abstract】 Because the human factor has an important role in road traffic safety and the driver correctly percept external environment and vehicle movement information is the basis of the corresponding driving behavior, a good way to assure system safety and coordination is to improve people’s reliability and information cognition capability. In the multiple channels to percept information of driver information provided by visual sensation is more than 90%.The driver correctly aware external environment information especially the spatial distance is of great significance to decision-making on people’s behavior and road infrastructure design. Apart from the driver’s physical and psychological factors illumination and color and other environmental conditions may influence distance cognition results. This paper on the basis of the information processing and the physical and psychological theory research laws of distance cognition and variance in daytime, dusk and night through dynamic experiments. The main research contents are as follows:1.Systematic analyze dynamic visual information processing research of the driver at home and abroad and particularly concern sub-consciousness of space distance and other unsolved problems. On the basis of cognitive engineering and experimental psychology this paper design experiment programs of the different environments and research method of data disposition and model foundation combined with the road traffic system and vehicle operating characteristics.2.Statistically analyze the distribution of experimental data in daytime and obtain characteristics of the results from the sub-consciousness.Study laws of distance discriminating results to red and green obstacles change with velocity at different depths distance and establish a linear regression model and BP neural network model.Test and compare the differences between two colors obstacles and establish models of cognition difference variances in use of neural networks. To consider the impact of speed and spatial distance use the dual surface regression and neural network method to establish dynamic distance perception model for sub-space distance.3.In use of the results of static experiment cognition establish logarithmic regression model to study the laws of cognition distance change with the environmental illumination. Test difference between red and green obstacles and utilize neural networks discriminating differences variance models in consideration of characteristics of visual recognition at dusk. Establish neural network models about cognition distance change with velocity and space distance through two input layer neurons network.4. According to drivers’recognition results at night establish negative exponential function model to describe cognition distance change with velocity. Test significance between red and green under low illumination and establish BP neural network model illuminate differences change with illumination and distance. Integrate recognition results from red and green obstacle and use Taylor’s formula to establish binary quadratic generalized regression model and neural network model of cognition result change with the speed and distance. Utilize distance recognition results under the same conditions from day and night establish difference between day and night change with velocity and distance in BP network model.5.Record and study eye movement parameters in course of distance cognition by eye tracking system. Analyze laws of fixation parameters variance and study the influence of velocity and space distance on dynamic visual feature in course of experiment.更多还原