Automatic Lane-Level Intersection Map Generation using Low-Channel Roadside LiDAR

A lane-level intersection map is a cornerstone in high-definition (HD) traffic network maps for autonomous driving and high-precision intelligent transportation systems applications such as traffic management and control, and traffic accident evaluation and prevention. Mapping an HD intersection is time-consuming, labor-intensive, and expensive with conventional methods. In this paper, we used a low-channel roadside light detection and range sensor (LiDAR) to automatically and dynamically generate a lane-level intersection, including the signal phases, geometry, layout, and lane directions. First, a mathematical model was proposed to describe the topology and detail of a lane-level intersection. Second, continuous and discontinuous traffic object trajectories were extracted to identify the signal phases and times. Third, the layout, geometry, and lane direction were identified using the convex hull detection algorithm for trajectories. Fourth, a sliding window algorithm was presented to detect the lane marking and extract the lane, and the virtual lane connecting the inbound and outbound of the intersection were generated using the vehicle trajectories within the intersection and considering the traffic rules. In the field experiment, the mean absolute estimation error is 2 s for signal phase and time identification. The lane marking identification Precision and Recall are 96% and 94.12%, respectively. Compared with the satellite-based, MMS-based, and crowdsourcing-based lane mapping methods, the average lane location deviation is 0.2 m and the update period is less than one hour by the proposed method with low-channel roadside LiDAR.      

P分位改进算法及其在路面裂缝检测中应用

针对传统的P分位分割算法容易受到道路标线的干扰,设计了一种改进型P分位算法分别对含有或没有含有道路标线的裂缝图像进行二值化,其中对不含道路标线的裂缝图像采用传统的P分位法,而对含道路标线的裂缝图像分三种情况进行细化处理。实验结果表明,与传统P分位算法相比,改进后的算法对路面裂缝图像分割及后续的图像处理产生较好的效果。     

复杂场景下视觉先验信息的地图恢复SLAM

目的 目前已有的单目视觉SLAM（simultaneous localization and mapping）系统每次开始运行时都将初始帧而不是绝对位置设置为参考帧,不能在一个固定的坐标系中获得位姿,导致无法重用已有的建图信息,而且在复杂场景中相机容易跟踪失败,需要当前帧与已有的关键帧非常相似时才能重定位并继续建图。针对这个问题,提出一种具有重新初始化、地图重用与地图恢复能力的视觉SLAM系统。方法 首先,加载先验地图,通过ORB（oriented brief）特征匹配SLAM系统当前帧与先验地图关键帧,并结合重定位方法完成SLAM系统的初始化。接着,为了避免丢失地图,建立一种应对SLAM系统跟踪失败的地图保存机制,保存跟踪成功地图,并提出一种自适应快速重新初始化算法,引入灭点检测,自动选择最佳重新初始化策略,保证SLAM系统继续跟踪与建图,建立的地图称为恢复地图。最后,对于跟踪成功地图与恢复地图,采用改进的回环方法获得它们之间的转换关系,并提出一种地图恢复法,减少跟踪成功地图与恢复地图尺度不一带来的误差,确保得到的全局一致地图更加准确。结果 在经过加噪处理的KITTI数据集上进行地图恢复融合的测试,实验结果表明,在KITTI00、KITTI02、KITTI05数据集下,本文提出的SLAM系统比ORB-SLAM2系统分别可以多获得39.25%、47.75%、32.46%的地图信息。在EuRoC数据集上的运行结果表明,本文提出的单目视觉SLAM系统不仅在建图精度方面与ORB-SLAM2效果相当,还在跟踪稳定性方面有显著提升。结论 本文提出的SLAM系统可以在跟踪失败的情况下有效恢复地图;此外,还可以高效重用SLAM系统已有的建图结果,固定SLAM地图坐标系,提升SLAM系统运行稳定性。     

基于SLIC和改进区域生长的非结构化道路识别

非结构化道路一般没有车道标识线且道路边界模糊,区分道路区域与背景区域难度较大。针对现有非结构化道路识别方法存在全像素域计算分类处理实时性差、易受噪声数据干扰等问题,提出一种基于SLIC（simple linear iterative clustering）超像素分割和改进区域生长算法的非结构化道路识别方法。利用均匀化初始聚类中心的SLIC算法生成低分辨率超像素特征图。在此基础上,利用聚类算法与邻域搜索算法自适应选择种子点,并引入CIEDE2000色差理论作为区域生长法生长准则,初步确定道路区域。根据道路连续一致特点,优化超像素级生长图并映射轮廓区域至原图,获得道路最终区域。基于数据集及真实场景的实验结果表明,该方法具有较高的识别率和抗干扰能力。     

基于激光测距的环境地图动态创建技术研究

本文主要研究完全未知结构化环境下的移动机器人二维地图构建与标图技术。本文以激光测距仪为环境探测传感器,采用几何特征法创建地图。对局部地图创建中的区域分割方法进行了改进,提出了基于线性阈值法的区域分割方法;给出了基于相关线段和线段缓冲区的全局地图创建方法。实验结果表明：本方法实现了基于实时的激光测距数据的局部地图动态创建和全局地图的实时更新,算法有效且可行。     

彩色公路交通地图图像道路提取

根据彩色公路交通地图的图像特征,提出一种新颖的道路识别与提取方法。这种方法包括三个关键步骤。首先,根据区域的特征,提取出区域的灰度值;其次,根据道路的颜色和形状特征以及数字图像处理的一些方法（如对象的连通成分等）,识别并提取出道路的颜色;最后,为了获得完整的道路网络,一些道路连接方法被提出。这种算法已经被应用于许多彩色公路地图图像中去提取道路网络。大量成功的实例表明这个算法是非常有效的。     

车辆视觉导航系统中的实时道路检测

提出了一种新的道路检测算法。该算法中首先采用基于线段的区域增长法将采集到的实际道路边缘图像分割成道路区域和非道路区域，使下一步搜索道路标志的区域限定在道路区域；然后恢复道路标志并根据其特征定位道路标志线；最后采用数据拟合的方法找出道路轨迹线。在复杂路况下可以准确、快速估算出车道的延伸方向，实现车辆的防偏预报。、     

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.     

Perceived density of road maps

Maps should be designed so that users can comprehend and use the information. Display decisions, such as choosing the scale at which an area is shown, depend on properties of the displayed information such as the perceived density (PD) of the information. Taking a psychophysical approach we suggest that the PD of information in a road map is related to the scale and properties of the mapped area. 54 participants rated the PD of 60 maps from different regions. We provide a simple model that predicts the PD of electronic road map displays, using the logarithm of the number of roads, the logarithm of the number of junctions and the length of the shown roads. The PD model was cross-validated using a different set of 60 maps (n = 44). The model can be used for automatically adjusting display scales and for evaluating map designs, considering the required PD to perform a map-related task.     

Automatic and Accurate Extraction of Road Intersections from Raster Maps

Since maps are widely available for many areas around the globe, they provide a valuable resource to help understand other geospatial sources such as to identify roads or to annotate buildings in imagery. To utilize the maps for understanding other geospatial sources, one of the most valuable types of information we need from the map is the road network, because the roads are common features used across different geospatial data sets. Specifically, the set of road intersections of the map provides key information about the road network, which includes the location of the road junctions, the number of roads that meet at the intersections (i.e., connectivity), and the orientations of these roads. The set of road intersections helps to identify roads on imagery by serving as initial seed templates to locate road pixels. Moreover, a conflation system can use the road intersections as reference features (i.e., control point set) to align the map with other geospatial sources, such as aerial imagery or vector data. In this paper, we present a framework for automatically and accurately extracting road intersections from raster maps. Identifying the road intersections is difficult because raster maps typically contain much information such as roads, symbols, characters, or even contour lines. We combine a variety of image processing and graphics recognition methods to automatically separate roads from the raster map and then extract the road intersections. The extracted information includes a set of road intersection positions, the road connectivity, and road orientations. For the problem of road intersection extraction, our approach achieves over 95% precision (correctness) with over 75% recall (completeness) on average on a set of 70 raster maps from a variety of sources.

基于UE4的在线驾驶仿真系统设计与实现

驾驶仿真系统在交通安全研究中发挥着重要作用,但目前仍存在三维道路建模受限、仿真场景还原度不高等问题;研究采用虚幻引擎开发仿真功能模块,根据道路几何线形与交通流原始资料,能高效搭建出驾驶仿真场景;采用像素流送技术将视频帧发送到前端,设置硬件设备的轴属性映射,支持模拟驾驶外部信号输入,并实时将驾驶行为数据存入数据库,设计了一种轻量化、自动化、智能化的在线驾驶仿真信息管理与可视化系统;系统主要实现了道路自动化建模、人-车-路-环境信息管理、车道级真实交通流场景还原、在线模拟驾驶与数据可视化等功能;经测试,系统部署后能在Web端稳定运行,弥补了现有驾驶仿真系统的不足,为实验人员提供轻量高效的驾驶仿真服务。     

Accurate lateral positioning from map data and road marking detection

We are witnessing the clash of two industries and the remaking of in-car market order, as the world of digital knowledge recently made a significant move toward the automotive industry. Mobile operating system providers are battling between each other to take over the in-vehicle entertainment and information systems, while car makers either line up behind their technology or try to keep control over the in-car experience. What is at stake is the map content and location-based services, two key enabling technologies of self-driving cars and future automotive safety systems. These content-based augmented geographic information systems (GIS) as well as Advanced Driver Assistance Systems (ADAS) require an accurate, robust, and reliable estimation of road scene attributes. Accurate localization of the vehicle is a challenging and critical task that natural GPS or classical filter (EKF) cannot reach. This paper proposes a new approach allowing us to give a first answer to the issue of accurate lateral positioning. The proposed approach is based on the fusion of 4 types of data: a GPS, a set of INS/odometer sensors, a road marking detection, and an accurate road marking map. The lateral road markings detection is done with the processing of two lateral cameras and provides an assessment of the lateral distance between the vehicle and the road borders. These information coupled with an accurate digital map of the road markings provide an efficient and reliable way to dramatically improve the localization obtained from only classical way (GPS/INS/Odometer). Moreover, the use of the road marking detection can be done only when the confidence is sufficiently high (punctual use). In fact, the vision processing and the map data can be used punctually only in order to update the classical localization algorithm. The temporary lack of vision data does not affect the quality of lateral positioning. In order to evaluate and validate this approach, a real test scenario was performed on Satory’s test track with real embedded sensors. It shows that the lateral estimation of the ego-vehicle positioning is performed with a sub-decimeter accuracy, high enough to be used in autonomous lane keeping, and land-based mobile mapping.     

结合先验知识和图像特征的道路提取方法

致力于在复杂环境下能对多种道路进行检测,提出了一种先验知识库与自适应区域增长相融合的道路检测方法,通过少量的道路样本采集,建立样本库,训练挖掘出道路知识模型,并结合区域增长方法对分割的实时道路影像进行道路区域增长。实验结果表明,该方法适用于多种不同环境道路的提取,鲁棒性强。     

基于环境信息熵的点线视觉里程计自适应优化器设计

针对即时定位与建图技术中点线视觉里程计在环境纹理发生变化时运行效率低下的问题,设计了一种基于环境信息熵的特征提取自适应优化器,以提高原有点线视觉里程计算法的效率及鲁棒性。优化器以图像信息熵作为主要影响因子,确定里程计的最优提取特征,生成包含特征提取选择的策略信息地图;对未探索区域的纹理环境进行预判性计算,与策略地图快速匹配,得到该区域的最优特征提取策略。在TUM数据集环境下测试了具有优化器的点线视觉里程计(APL-VO)的平均处理时间及建图效果。实验结果显示,与原有算法相比,具有自适应优化器的点线视觉里程计在复合环境中具有更强的鲁棒性及建图效率。     

基于浮动网格的路段检索方法

地图匹配是将车辆原始的GPS轨迹数据映射到实际道路网络上的过程, 其中为GPS轨迹点检索候选路段是地图匹配的首要环节, 然而不同的候选路段检索方式会直接影响地图匹配的准确性和效率. 本文针对城市路网环境下的低频采样GPS轨迹数据, 提出了一种基于浮动网格的路段检索方法. 该方法利用GeoHash网格编码, 采用浮动GeoHash网格的方式, 为轨迹点检索候选路段. 其次为了验证方法的可行性, 本文通过隐马尔可夫模型, 结合道路网络的拓扑结构以及轨迹的时空约束条件, 采用增量的方式, 利用维特比算法计算得到局部最优解. 最后使用贪心策略, 从已经得到的局部最优解中依次延伸得到全局最佳匹配路径.     

Multiple Condensation filters for road detection and tracking

In this paper, we present a novel and robust road tracking system for vision-based personal navigation. Novelty of the work includes the use of multiple Condensation filters to track the road of arbitrary shape and automatic switching between trackers according to road conditions. The approach allows the road to be represented as a simple hyperbola. It also supports the representation of the road as a sequence of connected arcs/segments so that information from a digital map can be integrated into tracking. The parameters of the hyperbola road model are estimated using multiple vanishing points located in image strips. The road tracking method is robust in dealing with complex road shapes, background clutters, shadows, and road markings. Experiments using real videos demonstrate the robustness of our approach.     

栅格地图道路的快速矢量化方法研究与实现

以彩色数字栅格地图道路数据为处理对象,提出一种基于可变矩形跟踪框技术快速矢量化道路类线要素的方法。在确定道路矢量化规则的基础上,通过采用可变矩形跟踪框技术,结合端点(交点)判断及道路惯性延伸点判断方法实行栅格地图道路的快速矢量化,并克服了当前几种常用矢量化方法的不足。实验结果表明该方法的矢量化效果十分理想。     

基于时空分析的地图匹配算法研究

地图匹配技术被广泛用于GPS导航、城市道路交通状态分析等领域。针对目前城市浮动车数据量日益庞大,地图匹配算法实时性差、匹配率不高的缺点,提出了一种基于时空分析的地图匹配算法。算法在对城市路网建立网格索引的基础上,综合考虑了空间几何、路网拓扑信息及上下文因素对选取GPS投影点的影响,大大提高了匹配效率和匹配精度。实验结果表明,算法能够满足工程应用中浮动车地图匹配的实时性和准确性。     

基于Google影像的城市道路网提取及其应用

从遥感影像中提取道路信息制作专题图方法具有时效性强、周期短、操作快捷等特点。首先介绍了道路提取、处理以及专题图制作的方法与流程。其次以Google高分辨率影像为数据源,利用影像中道路的光谱和几何特征信息,结合计算机分类与形态学处理方法对实验区城市道路网络进行提取、处理和专题图的制作。实验结果表明:该方法能够清晰地提取出道路框架,满足一定的制图需求,而且对车辆、人群、阴影产生的道路内部空洞和边缘侵蚀具有很好的处理效果,具有较强的经济性和实用性。     

复杂路段的角度差和后续点地图匹配方法

针对仅依靠距离和轨迹与路径的相似性来判断正确道路的方法,在并行和交叉路段的复杂路网环境中,易匹配到相邻路段或部分不可达的路段,导致匹配错误的问题,提出采用线性回归模型的方法,其能更精确地描述道路形状在转弯处的变化,根据路段方向和移动对象移动的方向差判断并行或交叉路段,并通过参照多个后续点的匹配情况,实现复杂路段处的地图匹配,减少匹配错误。此外,还提出简化聚类的路网补全方法,可以解决部分GPS点周围缺失候选路段无法匹配的问题,并采用四叉树索引地图数据,提高效率。对比实验结果表明在轨迹点数较少时,与现有的基于隐马尔可夫模型的概率匹配方法相比,时间最多减少了62%,更适合时效性要求高的应用场景,与传统的几何匹配方法相比,匹配精度提高了3%,且更适合复杂路段匹配。