基于多传感器数据融合的无人车行驶策略研究

发布时间：2018-01-30 03:51

本文关键词： 传感器数据融合驾驶行为 D-S证据理论　出处：《西安工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：为了提高道路交通安全,现代的无人车系统中广泛采用各类传感器来获得车辆行驶所需要的环境信息。面对逐渐复杂的实际环境,传统的单一传感器由于不能满足实时、快速地提供具有高精度和高可靠性的监测与定位信息,同时又不具备自主性好、对环境变化的适应能力强、抗干扰性强以及高性能价格比等要求,从而影响无人车行驶策略的制定。我们知道良好行驶策略的制定需要通过大量的环境信息对车辆关键部位行驶状态进行监测,进而能够更好地了解整个车辆的性能,并及时地制定出相应的调整策略来提高车辆的可靠性和安全性。因此,本文从如何更加合理利用传感器信息的角度出发对车辆运行状态进行检测,并在多传感器信息融合技术下对无人车行驶策略进行了相关的研究。首先,对无人车上基于环境信息采集系统中各个传感器的位置进行了设置；再次,由于环境信息的复杂多变,使得传感器获得的信息带有一定程度的不确定性,本文在综合考虑各类融合算法的基础上,采取将处理结果利用D-S证据理论进行融合很好的处理了不确定性的问题。最后,针对D-S证据理论自身不确定性与不能处理高冲突的问题,本文提出矩阵分析的理论修正了该问题；通过上述所得的融合结果结合相应的准则,我们就可以分析决策并制定无人车行驶策略。通过在实际道路中进行的无人车行驶实验,得出本文提出的算法能准确表示驾驶员的不确定性先验知识,能够保证在一个或多个传感器失效的情况下仍然具有良好的容错性。
[Abstract]:In order to improve road traffic safety, is widely used in various types of unmanned vehicle system of modern vehicle sensors to obtain the necessary environmental information. In the face of the actual environment is gradually complicated, the traditional single sensor because it can not meet the real-time, quickly provide the advantages of high precision and high reliability of monitoring and positioning information, but do not have good autonomy the change of environment, strong adaptability, strong anti-interference and high ratio of performance and price requirements, thus affecting the unmanned vehicle driving strategy. We know that to develop good driving strategy needs to be monitored by a large number of environmental information on the key parts of the vehicle, and to better understand the performance of the whole vehicle, and timely make the corresponding adjustment strategy to improve the reliability and safety of the vehicle. Therefore, this article from how to reasonably use of sensor information. Point of view to detect the running state of the vehicle, and the multi-sensor information fusion technology under the driving strategy of driverless vehicle is studied. Firstly, the unmanned vehicle on each sensor environment information collection system based on the location of the set; thirdly, because the environmental information is complicated and changeable, so that the sensor information obtained with a certain degree of uncertainty, based on the comprehensive consideration of all kinds of fusion algorithms, the results will be taken by the D-S evidence theory in data fusion to deal with uncertain problems. Finally, the needle can not handle the high uncertainty and conflict on D-S evidence theory, this paper puts forward the correction matrix analysis theory through the integration of these problems; the results obtained with the corresponding criterion, we can analyze the decision and to develop unmanned vehicle driving strategy. Through the actual road The experiment of driverless vehicle running in road shows that the algorithm proposed in this paper can accurately express the priori knowledge of uncertainty of drivers, and it can ensure that one or more sensors fail to have good fault tolerance.

【学位授予单位】：西安工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;U463.6

【引证文献】