基于道路结构特征的智能车单目视觉定位

高精度定位是实现自动驾驶的关键.在城市密集区域，全球定位系统（Global positioning system，GPS）等卫星定位系统受到遮挡、干扰、多路径反射等影响，无法保障自动驾驶所需的定位精度.视觉定位技术通过图像特征匹配进行位置估计，被广泛研究.然而传统基于特征点的方法容易受到移动目标的干扰，在高动态交通场景中的应用面临挑战.在结构化道路场景中，车道等线特征普遍存在，为人类驾驶员的视觉理解与决策提供重要线索.受该思路的启发，本文利用场景中的三垂线和点特征构建道路结构特征（Road structural feature，RSF），并在此基础上提出一个基于道路结构特征的单目视觉定位算法.本文利用在北京市区的典型路口、路段、街道等场所采集的车载视频数据进行实验验证，以同步采集的高精度GPS惯性导航组合定位系统数据为参照，与传统视觉定位算法进行比较.结果表明，本文算法在朝向估计上明显优于传统算法，对环境中的动态干扰有更高的鲁棒性.在卫星信号易受干扰的区域，可以有效地弥补GPS等定位系统的不足，为满足自动驾驶所需的车道级定位要求提供重要的技术手段.

Road Structural Feature Based Monocular Visual Localization for Intelligent Vehicle

Precise localization is an essential issue for autonomous driving applications, while global positioning system (GPS)-based systems are challenged to meet requirements such as lane-level accuracy, especially in crowded urban environment. This paper introduces a new visual-based localization approach in dynamic traffic environments, focusing on structured roads like straight roads or intersections. Such environments show several line segments on lane markings, curbs, poles, building edges, etc., which demonstrate the road's longitude, latitude and vertical directions. Based on this observation, we define a road structural feature (RSF) as sets of segments along three perpendicular axes together with feature points, and propose an RSF based monocular visual localization method. Extensive experiments are conducted on three typical scenarios, including highway, intersection and downtown streets. Results show better accuracy compared with a state-of-the-art visual localization method using feature points. We demonstrate that the proposed method can help improving localization accuracy in GPS restricted area, and discuss the remained challenges leading to future studies.