项目名称: 车联网环境下多意图跟踪的车辆碰撞预警方法研究
项目编号: No.61203236
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 自动化学科
项目作者: 马杰
作者单位: 武汉理工大学
项目金额: 24万元
中文摘要: 车联网环境下利用车车协同实现车辆碰撞预警,为提升汽车行驶安全水平创造了新途径。然而,由于没有考虑驾驶意图方面的因素,现有基于车车协同的汽车碰撞预警方法存在实时性低、误报率高的问题,难以适应有多车行驶的复杂道路环境。如何发现其他车辆运行状态背后的真实意图,识别其对本车构成的危险影响,是需要解决的关键问题。因此,本项目提出"相对意图"概念,通过"相对意图"与碰撞场景的关联,研究危险意图的动态演化形式;利用时空轨迹数据分析法,建立危险意图的时空特征数据模型;利用多维时间序列分析、动态贝叶斯网络、量子计算等,建立危险意图的识别模型和快速求解方法;运用D-S证据理论对危险意图、车辆运动状态和位置信息进行融合,建立车辆碰撞预测模型;基于这一模型实现对多个目标车辆的意图跟踪和碰撞预测。本项目试图在车联网环境下提出汽车碰撞预警的新方法,在理论和实践上为安全辅助驾驶研究提供新思路,促进车车协同的应用和推广。
中文关键词: 安全辅助驾驶;车联网;车辆碰撞预警;行车意图;
英文摘要: In vehicular networks, realizing inter-vehicle collision warning by using V2V cooperation creates a new way for improving vehicle driving safety. However, when discarding the driving intentions, the present V2V based inter-vehicle collision warning systems often behaves a low performance with high false reports, which have no capability to adapt to the multi vehicles driving road. In what way to find the true intentions hidden behind the neighboring vehicles and to perceive the danger situation caused by the neighboring vehicles are becoming a key scientific issue, which should be resolved at first. For this purpose, the research proposes a new concept: relative intention, and correlates it with the inter-vehicle collision scenarios to discover the danger intentions and its evolving forms. And then, model the forms with temp-spatial features by using temp-spatial data analyzing techniques. Taking advantages of multidimensional time series analysis, Dynamic Bayesian Network (DBN) and quantum computation techniques, the research constructs recognition patterns for the danger intentions and designs a real-time recognition algorithm. Base on the output of the recognitions, the recognized danger intentions,vehicle status and vehicle locations are fused using D-S theory. By this way, the inter-vehicle collision warnin
英文关键词: ADAS;Vehicular networks;Inter-vehicle collision warning;Vehicle motion intention;