CN103760569B - Travelable area detecting method based on laser radar - Google Patents

Travelable area detecting method based on laser radar Download PDF

Info

Publication number
CN103760569B
CN103760569B CN201310751817.8A CN201310751817A CN103760569B CN 103760569 B CN103760569 B CN 103760569B CN 201310751817 A CN201310751817 A CN 201310751817A CN 103760569 B CN103760569 B CN 103760569B
Authority
CN
China
Prior art keywords
data
laser radar
segments
travelable area
travelable
Prior art date
Application number
CN201310751817.8A
Other languages
Chinese (zh)
Other versions
CN103760569A (en
Inventor
薛建儒
张春家
杜少毅
戚晓林
王迪
程皓洁
Original Assignee
西安交通大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 西安交通大学 filed Critical 西安交通大学
Priority to CN201310751817.8A priority Critical patent/CN103760569B/en
Publication of CN103760569A publication Critical patent/CN103760569A/en
Application granted granted Critical
Publication of CN103760569B publication Critical patent/CN103760569B/en

Links

Abstract

本发明公开一种基于激光雷达的可行驶区域检测方法,利用激光雷达扫描实际交通场景,获取当前场景的平坦与起伏状况;然后通过高度变化情况将激光雷达返回的扫描数据分为若干个片段,随后使用迭代的直线拟合算法从所有的片段中提取出合适的直线特征,接着从满足约束条件的片段中找到平坦区域,再接着按照车宽要求从平坦区域确定可行驶区域;如果是多线激光,则可以由近及远,以前一线数据的可行驶区域为基础,从该线数据反映的平坦区域中确定的对应的可行驶区域,本发明可以有效的完成实际交通场景中可行驶区域的检测,准确的描述出汽车的安全通行区域,且本发明具有很好的实时性、易用性和鲁棒性。 Travelable region detection method based on the present invention discloses a laser radar, a laser radar scanning using actual traffic scenes, and obtaining a flat condition undulating current scene; followed by changes in the height of the scanning laser radar return data is divided into a plurality of segments, subsequently using an iterative fitting algorithm to extract a straight line from all the segments of a suitable linear features, and then find the segment from the flat region satisfy the constraints, and then subsequently determining in accordance with the requirements in the vehicle width region from the flat region; if it is a multi-line laser, it is possible and far from near, the previous line of data travelable area determined on the basis of the flat region of the line data reflect the corresponding travelable area, the present invention can effectively complete the actual traffic scene travelable area detection, accurate description of safe passage out of the area of ​​the car, and the present invention has a good real-time performance, ease of use and robustness.

Description

一种基于激光雷达的可行驶区域检测方法 Travelable area detecting method based on laser radar

技术领域 FIELD

[0001] 本发明属于无人驾驶汽车技术中的交通场景感知领域,具体涉及一种使用激光雷达作为数据来源完成实际交通场景中可行驶区域感知的方法。 [0001] The present invention belongs to the unmanned vehicle perception scene automotive technology field, particularly relates to a laser radar as the data source performs the actual traffic scenes perceived travelable area method.

背景技术 Background technique

[0002] 近年来随着汽车产业的高速发展,交通事故已经成为全球性的问题,全世界每年交通事故的死伤人数估计超过50多万人,因此集自动控制、人工智能、模式识别等技术于一体的无人驾驶应用而生。 [0002] In recent years, with the rapid development of automobile industry, traffic accidents has become a global problem, the number of casualties traffic accidents worldwide each year is estimated more than 50 million people, and therefore set the automatic control, artificial intelligence and pattern recognition technology to one of the unmanned applications for us. 为了实时可靠的感知环境信息,无人驾驶汽车上装备了各种主动传感器与被动传感器,包括相机、激光雷达、毫米波雷达和GPS,可行驶区域检测是无人驾驶技术中的关键部分之一。 For real-time reliable perception of environmental information, unmanned equipment on cars of various active sensors and passive sensors, including cameras, laser radar, millimeter-wave radar and GPS, can travel area detection is a key part of the unmanned technology . 一直以来,视觉技术通常是研究人员首选的研究方向,这是因为它信息含量大、成本低、运行功率低、扫描时间短。 All along, vision researchers usually the first choice of research direction, because it is information content, low cost, low power, short scan time. 但是,视觉技术一个最主要的不足就是对于外界环境的要求比较严格。 However, a major lack of vision technology requirements for the external environment is more stringent. 对于环境中出现阴影、驾驶环境较为复杂、路边标记(或标志) 丢失、光照不佳、能见度低或天气恶劣等情况,视觉技术(摄像机)获取的图像信息往往信噪比很低,使得特征提取的方法很难处理数据。 For shadowing environment, the driving environment is more complex, roadside markers (or marker) is lost, poor lighting, low visibility or bad weather, the vision (camera) information is often acquired image signal to noise ratio is very low, so that the feature the method of processing data extraction difficult. 从而难以将图像中的各个区域进行准确的分害J,也就无法准确的检测到可行驶区域。 Thereby making it difficult to respective regions in the image for accurate damage points J, it can not accurately detect the drivable region.

发明内容 SUMMARY

[0003] 本发明的目的在于提供一种高效、实时、鲁棒的基于激光雷达的可行驶区域检测方法。 [0003] The object of the present invention to provide an efficient, real-time, robust travelable region detection method based on laser radar.

[0004] 为达到上述目的,本发明采用了以下技术方案。 [0004] To achieve the above object, the present invention employs the following technical solutions.

[0005] 该可行驶区域检测方法包括以下步骤: [0005] The travelable area detecting method comprising the steps of:

[0006] 首先标定激光雷达,然后将采集到的激光雷达数据进行坐标转换;然后根据激光雷达的工作原理分割激光雷达数据为若干片段;对片段内的数据做杂点消除以及滤波处理以估计数据的真值;然后进行激光雷达数据的特征提取;然后根据可行驶区域约束以及车身宽度约束条件从提取的特征中确定可行驶区域。 [0006] First, the calibration of the laser radar, and the collected laser radar data coordinate conversion; then divided laser radar data into a plurality of segments according to the principle of the laser radar; data within the segment do Noise Reduction and filtered to estimate the data truth value; then the lidar data feature extraction; travelable area is then determined from the extracted feature travelable area according to vehicle width constraints and constraints.

[0007] 所述可行驶区域检测方法具体包括以下步骤: [0007] The travelable area detecting method includes the following steps:

[0008] 1)为了完成对于可行驶区域的描述,首先确定激光雷达在车辆上安装的高度,同时计算出激光雷达向下倾斜的角度,然后将采集到的激光雷达数据从2维极坐标系转化到局部3维直角坐标系; [0008] 1) In order to complete the description of the travelable area, first determine the height of a laser radar installed on a vehicle, and calculates the angle of downward inclination of the laser radar and the laser radar data collected from the two-dimensional polar coordinate system conversion to the local 3-dimensional rectangular coordinate system;

[0009] 2)激光雷达数据分割:如果相邻激光数据扫描到同一物体上,两个数据点间的高度差较小;相反相邻激光数据扫描到不同物体上,两个数据点间的高度差则较大,所以当用激光雷达倾斜向下扫描路面得到激光雷达数据后,采用相邻两点间的高度差值作为数据分割的标准,将激光雷达数据分割成若干个片段; [0009] 2) dividing the laser radar data: if the neighboring laser scan data to the same object, the smaller height difference between two data points; contrary adjacent laser scan data into a different object, the height between the two data points the difference is large, the road surface obtained when the scanning laser radar downwardly inclined lidar data, using the difference in height between two adjacent divided data as the standard, data is divided into a plurality of laser radar fragments;

[0010] 3)杂点消除及数据滤波:激光雷达数据中会存在一些与相邻两侧高度差都较大的数据点,本发明中将这类数据点称为杂点,其不反映真实空间位置的坐标,所以经过步骤2) 后,将仅包含单个激光雷达数据点的片段去除,然后对剩余的激光雷达数据片段进行滤波处理去除噪声得到用于特征提取的数据片段,由于受到系统噪声和随机噪声的影响,激光雷达数据反映的空间位置与实际对应的空间位置存在偏差,因此需要对激光雷达数据进行滤波处理,以便获得对实际空间位置更加准确的描述; [0010] 3) Noise Reduction and Data Filter: LIDAR be some difference in the height of the sides of adjacent data points are large, in the present invention, this data point is referred to Noise, which does not reflect the true position coordinate space, so after step 2), will contain only a single fragment of lidar data points removed, and the remaining laser radar data segments obtained is filtered to remove noise in the data segment used for feature extraction, due to the system noise and the influence of random noise, there is a deviation of the laser radar data reflects the actual spatial position corresponding to the spatial position of the laser radar data is required filtering process, in order to obtain a more accurate description of the actual spatial position;

[0011] 4)特征提取:经过步骤3)后,采用迭代的直线拟合算法从用于特征提取的数据片段中提取出对应的直线特征,使激光雷达数据变成对应实际场景中的平面; [0011] 4) feature extraction: After Step 3), using linear iterative fitting algorithm to extract data segments from a straight line feature extracting features corresponding to the laser radar data into a plane corresponding to the real scene;

[0012] 5)确定可行驶区域:从步骤4)提取到的直线特征中选择满足可行驶区域平坦约束、可行驶区域连续性约束以及车身宽度约束的直线特征作为可行驶区域。 [0012] 5) determined travelable area: line features extracted from step 4) to the selected area with a flat satisfy constraints can be characterized with continuity constraint and a straight region as a vehicle width constraints travelable area.

[0013] 所述步骤1)包括以下具体步骤: [0013] step 1) comprises the following steps:

[0014] 1. 1)以激光雷达安装位置为参考点定义局部3维直角坐标系,然后测量激光雷达距离水平面的安装高度H,并计算激光雷达倾斜向下扫描面相对于垂直方向的夹角α ; [0014] 1.1) laser radar mounted position as a reference point to define local 3-dimensional orthogonal coordinate system, and the distance measuring laser radar installation height level H, and calculating a scanning laser radar downwardly inclined surface to the vertical angle α ;

[0015] 1. 2)建立激光雷达扫描极坐标到局部3维直角坐标系中坐标的转换关系,根据转换关系获得激光雷达数据点Pi对应的3维直角坐标值,每一个激光雷达扫描极坐标数据点包括发射角度I与扫描距离1两个参数,则对应局部3维直角坐标系的坐标值为Pi=(Xi,yi, zj,转化方程如下: [0015] 1.2) to establish a laser radar scanning polar coordinate conversion relation to the local 3-dimensional coordinates in a Cartesian coordinate system, to obtain 3-dimensional Cartesian coordinates of the laser radar data based on the conversion point Pi corresponding relationship, each laser radar scanning polar transmitting data points includes two scanning angle I 1 from the parameters, corresponding to the local 3-dimensional rectangular coordinate system coordinate values ​​Pi = (Xi, yi, zj, the conversion equation is as follows:

[0016] xfsin α · cos β ; ·山 [0016] xfsin α · cos β; · Mountain

[0017] y^sin^ x · d, [0017] y ^ sin ^ x · d,

[0018] zfH-cos a · cos β ; ·山 [0018] zfH-cos a · cos β; · Mountain

[0019] 所述步骤2)包括以下具体步骤: [0019] step 2) comprises the following steps:

[0020] 根据激光雷达顺序扫描的原理,扫描到同一物体的相邻两个数据点的高度值不会存在太大的偏差,高度差较大的地方表示两个数据点扫描到不同的物体上,所以根据相邻数据点高度值之差将全部的激光雷达数据分为若干片段sk: [0020] The principle of laser radar sequential scanning, the scanning height values ​​of adjacent two data points of the same object does not present much of a deviation, where a large difference in the height data points represent two different objects to scan Therefore according to the difference between the height values ​​of adjacent data points all liDAR data into a plurality of segments sk:

[0021] S^tPj, (i=Nsk, ···, Nek) [0021] S ^ tPj, (i = Nsk, ···, Nek)

[0022] st I zs_zs ! I 彡Tz, I ze_ze+11 彡Tz [0022] st I zs_zs! I San Tz, I ze_ze + 11 San Tz

[0023] 其中,Pi表示激光雷达数据点,zs表示片段起始激光雷达数据点的高度,ze表示片段末尾激光雷达数据点的高度,Tz表示高度差阈值。 [0023] wherein, Pi represents a lidar data points, zs fragment represents the height of the starting point of the laser radar data, ze fragment represents the height of the end points of the laser radar data, Tz indicates the height difference threshold.

[0024] 所述滤波处理包括以下具体步骤: [0024] The filtering process comprises the following steps:

[0025] 对于数据片段,根据当前数据点Pi选择其设定窗口内的数据点进行均值滤波来估计数据点真值。 [0025] For data segments, mean filter to select the current data point Pi which data points within the window is set according to the data to estimate the true value point.

[0026] 所述步骤4)包括以下具体步骤: [0026] step 4) comprises the following steps:

[0027] 4. 1)将用于特征提取的数据片段分成靠近X轴与靠近Y轴两大类; Data segments [0027] 4.1) for feature extraction into close X-axis and Y-axis close to two categories;

[0028] 4. 2)根据用于特征提取的数据片段靠近不同的坐标轴选用不同的特征直线方程, 对于靠近X轴的数据片段采用y=kx+b的拟合方式确定其对应的特征参数(k,b),靠近Y轴的数据片段则采用x=ky+b的拟合方式确定其对应的特征参数(k,b); [0028] 4.2) close to the axis according to the different data segments for different selected feature extraction equation of the linear characteristic, determining the characteristic parameters corresponding to the X-axis of the data segments near y = kx + b using fitting manner (k, b), the data segment is used near the Y-axis x = ky + b fitting way to determine characteristic parameters (k, b) corresponding;

[0029] 4. 3)计算每个用于特征提取的数据片段与其特征直线方程的均方误差,如果均方误差值大于等于均方误差阈值,则说明该数据片段不能由一个直线方程描述,需要两个或更多的直线方程来描述该数据片段,找到该数据片段中距离特征直线方程最远的点,以此点将该数据片段一分为二,然后对两个新的数据片段分别采用步骤4. 2)的方法进行直线拟合直至均方误差值小于均方误差阈值。 [0029] 4.3) is calculated for each segment for its characteristic feature extraction equation of the linear mean square error, mean square error if the value is greater than a threshold value equal to the mean square error, then the data segment can not be described by a linear equation, two or more are required to describe a linear equation of the data segment, the data segment to find the point farthest from the characteristic equation of a straight line, thus the data points into two segments, then two new data segments are employing step 4.2) the method of fitting a straight line until the mean square error is smaller than the mean square error threshold.

[0030] 所述步骤5)包括以下具体步骤: [0030] step 5) comprises the following steps:

[0031] 5. 1)选择直线特征满足可行驶区域平坦约束的数据片段,删除不满足可行驶区域平坦约束的数据片段,所述可行驶区域平坦约束指直线与局部3维直角坐标系中Y轴夹角大于等于设定的阈值Τ γ; [0031] 5.1) Select line characteristic data segments satisfies constraints flat travelable area, delete the travelable area does not satisfy the constraint data segments flat, planar region with said constraint means 3-dimensional straight line and the local Cartesian coordinate system Y greater than the threshold shaft angle equal to the set Τ γ;

[0032] 5. 2)经过步骤5. 1)后,合并满足可行驶区域连续性约束的数据片段,将其视为一块水平区域,作为描述可行驶区域的候选对象;所述可行驶区域连续性约束是指相邻数据片段的高度差值以及水平距离差值分别小于设定的对应阈值T z、Td; [0032] 5.2) through step 5.1), the fragments may be combined with the data area satisfy the continuity constraints, be considered as a horizontal area, it can be described with a candidate region; the region can be continuous with constraint means that the difference in height and a horizontal distance difference between adjacent data segments are smaller than a threshold value corresponding to the set T z, Td;

[0033] 5. 3)经过步骤5. 2)后,选择候选对象中满足车身宽度约束、且让车辆尽量少做横向运动的候选对象用于描述可行驶区域。 [0033] 5.3) after step 5.2), selecting a candidate object satisfies the vehicle width constraints, let the vehicle as little as possible candidates for describing the transverse movement travelable area.

[0034] 本发明的有益效果体现在: [0034] Advantageous effects of the present invention are embodied in:

[0035] 本发明利用激光雷达扫描实际交通场景,获取当前场景的平坦与起伏状况;然后通过高度变化情况将激光雷达返回的扫描数据分为若干个片段,随后使用迭代的直线拟合算法从片段中提取出合适的直线特征,接着从满足约束条件的片段中找到平坦区域,再接着按照车宽要求从平坦区域确定可行驶区域;如果是多线激光,则可以由近及远,以前一线数据的可行驶区域为基础,从该线数据反映的平坦区域中确定对应的可行驶区域,本发明可以有效的完成实际交通场景中可行驶区域的检测,准确的描述出汽车的安全通行区域, 且本发明具有很好的实时性和易用性。 [0035] The present invention utilizes a laser radar scanning actual traffic scenes, and obtaining a flat condition undulating current scene; followed by changes in the height of the scanning laser radar return data is divided into several segments, and then using an iterative fitting algorithm from a straight line segment extracted suitable linear characteristic, and then find the flat region from satisfy the constraints fragments, followed by determining from the flat region may travel region in the vehicle width requirement; if it is a multi-line laser, may be far from the close previous line data the travelable area basis, to determine a corresponding travelable area from the flat region of the line data is reflected, the present invention can effectively complete the actual traffic scene can travel detection area, an accurate description of the safe passage area of ​​the vehicle, and the present invention has good real-time performance and ease of use.

[0036] 本发明利用激光雷达作为数据来源,通过坐标系转换、数据分割、数据滤波、特征提取和确定可行驶区域等步骤,完成可行驶区域检测的功能,由于激光雷达的工作原理与相机等不同,能够很好的克服天气、光照等环境因素的影响,所以本发明可以实现可行驶区域鲁棒检测的功能。 [0036] The present invention utilizes a laser radar as the data source, through coordinate transformation, data splitting, data filtering, feature extraction and determination drivable step regions, and complete travelable area detecting function, the laser radar works with cameras different, it is possible to overcome the environmental factors good weather, light, etc., the present invention can be realized with the function of robust detection region.

附图说明 BRIEF DESCRIPTION

[0037] 图1为抽象的道路模型; [0037] FIG. 1 is the abstract model of the road;

[0038] 图2为激光雷达标定及局部坐标系; [0038] FIG. 2 is a laser radar calibration and a local coordinate system;

[0039] 图3为激光雷达可行驶区域检测总体框图; [0039] FIG. 3 is a general block diagram of a laser radar detection area travelable;

[0040] 图4为激光雷达数据分割流程图; [0040] FIG 4 is a flowchart of lidar data division;

[0041] 图5为激光雷达单线数据分割结果; [0041] FIG. 5 is the result of dividing a single line laser radar data;

[0042] 图6为激光雷达数据真值估计流程图; [0042] FIG. 6 is a lidar flowchart estimate the true value;

[0043] 图7为激光雷达数据片段特征提取流程图; [0043] FIG. 7 is a lidar flowchart feature extraction data segments;

[0044] 图8为激光雷达数据片段直线拟合方式划分区域; [0044] FIG. 8 is a lidar data segments fitted to the linear divided areas;

[0045] 图9为迭代直线拟合示意图; [0045] FIG. 9 is a schematic diagram of an iterative fitting a straight line;

[0046] 图10为迭代直线拟合结果; [0046] FIG. 10 is a linear iterative fitting result;

[0047] 图11为实际交通场景4线激光雷达可行驶区域检测的结果; [0047] FIG. 11 is actual traffic scenes travelable 4 lidar detection result of the area;

[0048] 图12为实际交通场景64线激光雷达可行驶区域检测结果。 [0048] FIG. 12 is actual traffic scene 64 lidar travelable area detection result.

具体实施方式 Detailed ways

[0049] 下面结合附图对本发明作详细说明。 [0049] DRAWINGS The present invention will be described in detail.

[0050] 为了能够实时、稳定的感知周围交通场景,本发明给出了一种基于激光雷达数据的实际交通场景可行驶区域检测方法,具体包括激光雷达标定、数据分割、杂点消除与数据滤波、特征提取以及确定可行驶区域五个部分,如图3所示。 [0050] For real-time and stable perception of surrounding traffic scenario, the present invention presents a practical laser radar based traffic scene data area detecting driving method, including laser radar calibration, data partitioning, data filtering and Noise Reduction , and feature extraction region determining section with five, as shown in FIG. 该方法具体按以下步骤进行: The method specifically perform the following steps:

[0051] 步骤1. 1 :为了实现激光雷达可行驶区域检测的功能,需要将激光雷达安装于车头正前方,斜向下扫描,保证其与路面相交。 [0051] Step 1.1: To achieve the lidar travelable area detecting function needs to be attached to the front of the laser radar front, obliquely scan, which intersects with the road surface to ensure that.

[0052] 此时定义如图2所示局部空间直角坐标系,测量激光雷达距离局部空间直角坐标系水平面(XY,其中Y向为车头正前方所指的方向)的安装高度Η及斜向下扫描面相对于垂直方向的夹角α,确定激光雷达安装位置在局部空间直角坐标系的位置。 [0052] At this time, as shown in local space defined Cartesian coordinate system shown in FIG. 2, the laser radar measuring spatial rectangular coordinate system from the local horizontal plane (the XY, where Y is the direction referred to in front of the front) is mounted obliquely and the height Η scan plane angle α in the vertical direction, the laser radar mounted position is determined in the local spatial rectangular coordinate system.

[0053] 步骤1. 2 :由于激光雷达是扫描式的主动传感器,其数据是以激光雷达为极点,正前方或是其他特定方向为极轴的极坐标系的距离和角度,而后面的计算全部是基于直角坐标系来实现的,所以需要建立激光雷达扫描数据极坐标系到局部空间直角坐标系坐标的转换关系,获得激光点对应的直角坐标值。 [0053] Step 1.2: Since the scanning laser radar is an active sensor, a laser radar which data is to the pole, or other specific front direction and an angle from the polar axis of a polar coordinate system, while the later calculations are all implemented based on a rectangular coordinate system, it is necessary to establish a laser radar scan data into the polar coordinate system conversion relationship local space coordinate Cartesian coordinate system, is obtained Cartesian coordinates corresponding to the laser spot.

[0054] 每一个激光数据点包括发射角度I与扫描距离七两个参数,即Pfd I),而局部空间直角坐标系则需要Χι,Υι,Zi3个坐标值,即Ρι=(Χι,Υι,Ζι)。 [0054] Each data point comprises a laser emitting angle I and the scanning distance seventy-two parameters, i.e. Pfd I), the local spatial rectangular coordinate system is required Χι, Υι, Zi3 coordinate values, i.e. Ρι = (Χι, Υι, Ζι). 根据图2所示的激光雷达安装位置在局部空间直角坐标系中的位置,可以通过三角函数完成极坐标到直角坐标的转换,转换等式如下所示: The spatial position of the local rectangular coordinates, the polar coordinates can be accomplished by the installation position of the laser radar according to a trigonometric function shown in FIG. 2 to transform the Cartesian coordinates, the following conversion equation:

[0055] xfsin α · cos β ; ·山 [0055] xfsin α · cos β; · Mountain

[0056] y^sin β ; · d; [0056] y ^ sin β; · d;

[0057] zfH-cos a · cos β ; ·山 [0057] zfH-cos a · cos β; · Mountain

[0058] 上式将只有两个数值的极坐标值转换到包含3个数值的直角坐标系,其中包括数据点在局部空间直角坐标系的高度值^,这对于后面的处理相当重要。 [0058] The above formula is only two values ​​of the polar coordinate values ​​converted to a Cartesian coordinate system comprising three values, including the value of the height data points in a local rectangular coordinate system ^ space, which is important for later processing.

[0059] 步骤2. 1 :根据激光雷达的安装位置,如图2所示,其扫描平面为局部空间直角坐标系中一个斜向下的平面,所以当相邻两个数据点扫描到同一物体时其高度值基本相同, 不会存在太大的偏差,反之扫描到不同的物体上高度值则会有较大的偏差,所以根据相邻数据点高度值之差将全都激光雷达数据分为若干片段S k。 [0059] Step 2.1: The mounting position of the laser radar shown in Figure 2, which local spatial scanning plane is a plane in a Cartesian coordinate system obliquely downward, so when the two adjacent scan data points to the same object value substantially the same as the height, will be no big variation, whereas the value of the scan height to different objects will have a large deviation, the height difference according to the value of the adjacent data points will be divided into all lIDAR segment S k. 激光雷达数据点分割的流程如图4所示,按照下面的步骤进行: Lidar data point divided flow shown in Figure 4, in accordance with the following steps:

[0060] 1)将所有数据点按照X值升序的顺序排列,保证X值从左到右依次变大; [0060] 1) all data points in ascending order of X values ​​are arranged to ensure that the value of X increases from left to right;

[0061] 2)创建一个新的片段Si,其中只包含第一个数据点P1; [0061] 2) Create a new segment Si, which contains only a first data point Pl;

[0062] 3)依次读取存储序列中的数据点P1; [0062] 3) sequentially reads the data stored in the sequence of points Pl;

[0063] 4)计算卩1与前一个数据点P i丨的高度差,即Δ z=| z ^ 1 [0063] 4) Calculate the height of the previous Jie 1 data point P i Shu difference, i.e., Δ z = | z ^ 1

[0064] 5)如果高度差小于阈值Tz,即Δ Z〈TZ,将次序存入到当前片段S#,然后返回到3); [0064] 5) If the height difference is smaller than the threshold value Tz, i.e. Δ Z <TZ, the order of deposit into the current segment S #, and then returns to 3);

[0065] 6)如果高度差不小于阈值TZ,即ΔΖ >TZ,则创建一个新的片段Sk+1,将Pi按次序存入到Sk+1中,然后返回到3); [0065] 6) If the height difference is not smaller than the threshold value the TZ, i.e. ΔΖ> TZ, create a new segment Sk + 1, Pi is stored in sequence to Sk + 1, and then returns to 3);

[0066] 通过上面的循环计算,可以获得一个片段序列,参见图5,图5中的空心圆圈表示数据片段的端点,其中每个片段有如下的表达形式: [0066] By the above calculation cycle, a fragment of a sequence can be obtained, see FIG. 5, FIG. 5, open circles indicate the endpoints for the data segments, wherein each segment has the following expressions:

[0067] Sk= {Pj,(i=Nsk,…,NJ [0067] Sk = {Pj, (i = Nsk, ..., NJ

[0068] st | zs_zs ! | 彡Tz,| ze_ze+11 彡Tz [0068] st | zs_zs |! San Tz, | ze_ze + 11 San Tz

[0069] 其中,zs表示片段起始激光雷达数据点的高度,%表示片段末尾激光雷达数据点的高度。 [0069] wherein, ZS indicates the starting segment height lidar data points, expressed in% height lidar data segment end points.

[0070] 步骤3. 1 :激光雷达的工作特性导致会有一些噪声数据的存在,为了能够更加真实的反映实际场景,需要将这些噪声数据清除掉。 [0070] Step 3.1: the operating characteristics of the laser radar results in the presence of noise there will be some data, in order to more truly reflect the actual scene, these noise data needs to be removed. 本发明中通过判断上面获得的激光雷达数据片段包含的数据点数来实现这个功能,只包含一个激光雷达数据点的片段,被认为是不能反映场景实际情况的杂点,予以消除。 Laser radar data points in the data segments obtained by the present invention comprising the above judgment to achieve this function, a fragment contains only a lidar data point is considered not to reflect the actual situation of the scene point heteroaryl, be eliminated.

[0071] 步骤3. 2:测量噪声和系统噪声是数据测量中普遍存在的两大噪声数据,为了能够得到数据的真值需要对获得的测量数据进行滤波处理,消除噪声的干扰。 [0071] Step 3.2: measurement noise and system noise are prevalent in the two data measurement noise data, in order to obtain the true value of the data requires that the measurement data obtained by performing filter processing to eliminate the noise interference. 本发明选择经典的均值滤波方法完成每一个数据片段内测量数据的真值估计,如图6所示,对于每一个激光雷达数据片段,根据当前数据Pi选择其规定窗口长度N内的数据点c (i)来估计该点的真值,当数据侧或是右侧数据个数小于规定窗口长度时,调整滤波器的规定窗口长度N。 Classical mean filter selection method of the invention is completed for each data segment within the true value of the estimated measurement data, shown in Figure 6, a laser radar for every data segment, which is selected according to a predetermined current data Pi data points within the window length N c (i) to estimate the true value of the point, when the data or the number of right-side data is less than a predetermined window length, adjusting a predetermined length of the filter window N. 此处选择滤波器的窗口如下所示: Here selection filter window as follows:

Figure CN103760569BD00081

[0075] 其中,Nsk是参与滤波计算的第一个数据点的索引,N &是参与滤波计算最后一个数据点的索引,通过上式,可以使得数据点的坐标值更加接近真值。 [0075] where, is the participation of Nsk first filter calculation data point index, N & involved filtering the last data point index calculated by the equation, coordinate value data can be made closer to the true value point.

[0076] 步骤4. 1 :由于本发明中特征提取的主要手段是直线拟合,而不同位置的直线拟合方程是不同的,通常靠近X轴的数据会以X轴作为自变量,靠近Y轴的会以Y轴作为自变量,这样拟合出来的直线能够更加准确的描述数据点的特征。 [0076] Step 4.1: Since the main feature extraction means of the present invention is a straight line fit, the equation of a straight line fit different locations are different, the data typically will be close to the X-axis as the independent variable X axis, near the Y axis as the Y-axis will be independent variables, wherein the fitting out of this straight line can more accurately describe the data points. 所以如图8所示,将所有数据片段分成靠近X轴与靠近Y轴两大类。 Therefore, as shown in Figure 8, close to all the data segments into X-axis and Y-axis close to the two categories.

[0077] 步骤4. 2 :根据上一步对数据片段的划分,采用最小均方误差的直线拟合方式完成数据片段的特征提取。 [0077] Step 4.2: The last step of dividing the data segments, characterized in using a linear minimum mean square error fit is accomplished extracted data segment. 实现流程如图7所示,根据数据片段靠近不同的坐标轴选用不同的拟合函数,对于靠近X轴的数据片段采用y=kx+b的拟合方式确定其对应的特征参数(k,b),靠近Y轴的数据片段则采用x=ky+b的拟合方式确定其对应的特征参数(k,b)。 Implementation process shown in Figure 7, close to the different axes use different fit function according to the data segment, close to the X-axis for the data segment using y = kx + b fitting way to determine characteristic parameter (k corresponding, b ), data segments near the Y-axis is x = ky + b using fitting way to determine characteristic parameters (k, b) the corresponding.

[0078] 步骤4. 3 :虽然采用了最小均方误差的拟合方式,但是最小均方误差的方法只能保证对于当前的直线方程有最小的误差,不能够说明一条直线就可以准确的描述数据片段的特征。 [0078] Step 4.3: Although the use of minimum mean square error fitting manner, but the minimum mean square error method is only guaranteed for the current linear equation with a minimum error can not be a straight line can be described accurately describe characteristic data segment. 因此当均方误差较大时,说明一条直线已经不能充分的描述该数据片段的特征, 需要再引入一条或多条直线来更加细致的刻画数据片段的特征。 Thus when the mean square error is larger, a straight line can not be described fully described features of the data segments, we need to be introduced into one or more lines more detailed characterizations of data segments. 该功能的实现过程如下所示: This implementation functions as follows:

[0079] 1)判断直线拟合的均方误差是否小于阈值,如果小于阈值则进行下一段数据片段特征的提取; [0079] 1) to determine the mean square line fitting error is less than the threshold value, the threshold value is less than the period of the extracted feature data segments is performed;

[0080] 2)如果大于等于阈值,则如图9所示,寻找该数据片段中距离拟合得到的直线距离最大的点P n; [0080] 2) if not less than the threshold value, as shown in FIG. 9, the linear distance to find the distance data pieces obtained by fitting maximum point P n;

[0081] 3)从P"处将数据片段一分为二得到片段邑= %···.ξ,丨和片段么+i = K,··· A!; [0081] 3) from P "of the data segments is divided into two fragments obtained Yi =% ··· .ξ, Shu, and fragments of it + i = K, ··· A !;

[0082] 4)在片段&后插入新的片段:,用片段基代替片段Sk; [0082] 4) insert a new segment after segment &: replace Sk fragment by fragment group;

[0083] 5)从片段&开始进行特征提取,返回1); [0083] 5) starts from the segment feature extraction & Returns 1);

[0084] 通过上面的处理,可以保证所有的数据片段都找到其对应的直线方程来描述其特征,如图10所示,这些特征就比较客观的反映实际交通场景的情况,为下一步的可行驶区域检测提供了稳定的数据支持。 [0084] By the above process, will ensure that all data segments have found a linear equation which corresponds to be characterized, shown in Figure 10, these features are more objectively reflect actual traffic scenes, and may be for the next travel area detection data to provide a stable support. 对应每个数据片段都有如下的描述格式: Each segment has a corresponding data format is described as follows:

[0085] Sf {{f, ki, bj,{Pm,…,Pn}} [0085] Sf {{f, ki, bj, {Pm, ..., Pn}}

[0086] 其中f用来标记拟合方式,U h为数据特征,P ",···,Pn为该片段包括的数据点。 [0087] 步骤5. 1 :客观环境中的可行驶区域一定是一块平坦的区域,不会存在高低不平的情况,在本发明中这个特点被称为平坦性原则。实际交通场景模型如图1所示,所以选择特征满足平坦性原则要求的数据片段作为可行驶区域的候选对象。选择方法如下所示: [0086] wherein f is used to mark fitting manner, U h feature data, P ", ···, Pn segment that includes data points [0087] Step 5.1: objective driving environment can be constant region is a flat area, there will be no case where rugged, this feature is referred to in the present invention, the flatness of the principle of the actual traffic scene model shown in Figure 1, so the selected feature data segments meet flatness requirements as a principle candidate target travel region selection method is as follows:

[0088] 1)顺序读取数据片段S1; [0088] 1) Sl sequentially read data segments;

[0089] 2)计算特征直线与Y轴的夹角θι; [0089] 2) Calculation of characteristic lines and the Y-axis angle θι;

[0090] 3)如果该直线与Υ轴夹角较小,即| Θ ^90 | <ΤΥ,则认为该段数据点扫描到非水平面上,不是可行驶区域,将其删除; [0090] 3) If the angle between the straight line and the axis Υ small, i.e., | Θ ^ 90 | <ΤΥ, the segment is considered point scan data to a non-horizontal surface, not the travelable area, delete;

[0091] 4)如果该直线与Υ轴夹角较大,即I Θ i-901彡Τγ,则认为该段数据点扫描到水平面上,是可行驶区域,将其保留; [0091] 4) If the angle between the straight line and Υ axis larger, i.e. I Θ i-901 San Τγ, the segment is considered point scan data to the horizontal plane, a travelable area, which is retained;

[0092] 通过上面的处理,可以将不符合交通场景模型中路面的区域删除掉,只保留符合平坦性原则的平面。 Area [0092] Through the above process, can not meet the traffic scene model pavement removed, leaving only the plane in line with the principle of flatness.

[0093] 步骤5. 2 :可行驶区域是一块连续的平坦区域,中间不存在较大的间断和高度差, 在本发明中这个特点被称为连续性原则。 [0093] Step 5.2: travelable area is a continuous flat region, intermediate and large discontinuities height difference does not exist, in the present invention, this feature is called the principle of continuity. 上面得到的片段均表示一块平面,但是没有表明相邻片段是否满足连续性原则,实际上两个相邻且没有明显间隔和高度差的片段,描述的是同一个平面区域,所以将其合并以更好的描述该平面区域。 The above fragments were obtained showing a plane, but did not indicate whether the adjacent segments meet the principle of continuity, and in fact the two adjacent intervals and no significant difference in the height of fragments, is described with a planar region, which is incorporated in better description of the planar region. 实现步骤如下所示: Implement the following steps:

[0094] 1)从第二个片段开始,顺序读取片段S1; [0094] 1) from the second segments are sequentially read starting segment Sl;

[0095] 2)计算片段Si与前一片段S ii两相邻端点的高度差Δ z与水平距离差Δ d ; [0095] 2) Si height difference calculated fragment of the previous segment S ii endpoint two adjacent horizontal distance and difference Δ z Δ d;

[0096] 3)如果高度差与水平距离差有一个不小于给定阈值,即Δ d彡Td或是Δ z彡T z, 则认为这两个片段描述的不是同一平面,返回1); [0096] 3) If the height difference between the horizontal distance and the difference is not less than a given threshold value, i.e., Δ d Δ z San San Td or T z, is not the same plane that the two fragments described, returns 1);

[0097] 4)如果高度差与水平距离差都小于给定阈值,即Δ (1〈1^且Δ Z〈T z,则认为这两个片段描述的是同一平面,将片段31与S ii合并,然后返回1); [0097] 4) If the height difference between the horizontal distance difference is less than a given threshold value, i.e., Δ (1 <1 ^ and Δ Z <T z, it is considered that two fragments describe the same plane, the segments 31 and S ii were combined, and then return 1);

[0098] 通过上面的处理,将所有的数据片段按照其描述的平面进行合并,这样合并后的数据片段就是道路模型中的一整块平面。 [0098], all of the data segments are combined according to the plane thereof by the above described process, the data segment is a road model so the combined planar piece.

[0099] 步骤5. 3 :现在所有的数据片段都描述一块连续的平面,但是并不全是可行驶区域,因为有些区域小于车身的宽度、有些区域距离车辆太远,所以要从中选择出最合适的数据片段作为可行驶区域。 [0099] Step 5.3: Now all the data segments have described a continuous plane, but not all travelable area, because some areas less than the width of the vehicle body, some areas far from the vehicle, it is most appropriate to select from as the data segment travelable area. 本发明采用了一个让车辆尽量少做横向运动的假设,即选择车辆能够通过且位于车头前方的最大区域作为可行驶区域,实现方式如下所示: The present invention uses a vehicle as little as possible so that lateral movement of the assumption that the vehicle can pass through and select the area in front of the front at the maximum as a travel area, implementation as follows:

[0100] 1)依次读取数据片段,计算数据片段起始点与终止点的水平距离ds; [0100] 1) sequentially reading a data segment, the data segment calculating the horizontal distance between the start point and termination point DS;

[0101] 2)如果水平距离小于车宽,即ds〈Wv,认为该区域不能满足车辆通过要求,为不可行驶区域,返回1)。 [0101] 2) If the distance is less than the horizontal width of the vehicle, i.e., ds <Wv, that the vehicle passes through this area does not meet the requirements for non-driving area, to return 1).

[0102] 3)如果水平距离大于等于车宽,即ds> Wv,则认为满足车身宽度约束条件,接着计算该数据片段中心点与坐标原点构成的直线与Y轴的夹角的绝对值Θ s; [0102] 3) If the distance is greater than equal to the horizontal width of the vehicle, i.e., ds> Wv, is considered to meet the vehicle width constraints, then calculates the angle of the line and the Y-axis data segments constituting the center point of the coordinate origin of the absolute value of Θ s ;

[0103] 4)如果当前的0s小于当前可行驶区域的夹角Θ。 [0103] 4) If the current is less than the current travelable area 0s angle Θ. ,即θ'θ。 That θ'θ. ,则将当前数据片段作为新的可行驶区域,且令L=0 s,返回1)。 , The current data segment as a new travelable area, and let L = 0 s, 1 returns).

[0104] 经过上面的处理,从所有片段中选择满足宽度约束的位于车头正前方的平面区域作为可行驶区域,如果有多线激光数据,则以前一线的可行驶区域为基础,依次向外扩展, 确定其数据片段中的可行驶区域,参见图11以及图12,其中颜色较浅的灰度区域为检测到的可行驶区域。 [0104] After the above process, that satisfies the constraints of the width of the region located directly in front of the front plane from all the segments as the travelable area, if a plurality of data line laser, the previous line of the travelable area based on successively extended outwardly determining which data segments travelable area, see FIGS. 11 and 12, wherein the lighter gray area is detected drivable region.

Claims (4)

1. 一种基于激光雷达的可行驶区域检测方法,其特征在于:该可行驶区域检测方法包括以下步骤: 首先标定激光雷达,然后将采集到的激光雷达数据进行坐标转换;然后分割激光雷达数据为若干片段;对片段做杂点消除以及滤波处理后进行激光雷达数据的特征提取;然后根据可行驶区域约束以及车身宽度约束条件从提取的特征中确定可行驶区域; 所述可行驶区域检测方法具体包括以下步骤: 1) 首先确定激光雷达安装的高度,同时计算出激光雷达向下倾斜的角度,然后将采集到的激光雷达数据从2维极坐标系转化到局部3维直角坐标系; 2) 激光雷达数据分割:采用相邻两点间的高度差值作为数据分割的标准,将激光雷达数据分割成若干个片段; 3) 杂点消除及数据滤波:经过步骤2)后,将仅包含单个激光雷达数据点的片段去除, 然后对剩余的激光雷达 Travelable area detecting method for a laser-based radar, wherein: the travelable region detection method comprising the steps of: a laser radar calibrated first, and then the collected data, the laser radar coordinate transformation; and divided LIDAR several fragments; fragments do Noise Reduction feature lidar and extracting the filtered data; travelable area is then determined from the extracted feature travelable area according to the vehicle width constraints and constraints; the travelable region detection method It includes the following steps: 1) first, to determine the height of the laser radar installation, while calculating the angle of downward inclination of the laser radar and the laser radar data collected from the two-dimensional polar coordinate conversion to a local 3-dimensional rectangular coordinate system; 2 ) laser radar data segmentation: the difference in height between adjacent two divided data as the standard, data is divided into a number of lidar segments; 3) Noise Reduction and data filter: after step 2), will contain only fragment single laser radar data points removed, and the remaining lidar 数据片段进行滤波处理去除噪声得到用于特征提取的数据片段; 4) 特征提取:经过步骤3)后,采用迭代的直线拟合算法从用于特征提取的数据片段中提取出对应的直线特征,使激光雷达数据变成对应实际场景中的平面; 5) 确定可行驶区域:从步骤4)提取到的直线特征中选择满足可行驶区域平坦约束、可行驶区域连续性约束以及车身宽度约束的直线特征作为可行驶区域; 所述步骤4)包括以下具体步骤: 4. 1)将用于特征提取的数据片段分成靠近X轴与靠近Y轴两大类; 4. 2)根据用于特征提取的数据片段靠近不同的坐标轴选用不同的特征直线方程,对于靠近X轴的数据片段采用y=kx+b的拟合方式确定对应的特征参数(k,b),靠近Y轴的数据片段则采用X=ky+b的拟合方式确定对应的特征参数(k,b); 4. 3)计算每个用于特征提取的数据片段与其特征直线方程的均方误差 Data segments obtained is filtered to remove noise in the data segment used for feature extraction; 4) feature extraction: After Step 3), using linear iterative fitting algorithm to extract data segments from the straight line for extracting features corresponding to features, the laser radar data into a plane corresponding to the real scene; 5) determining travelable area: line features extracted from step 4) to the selected planar region satisfies travelable constraints, constraints and continuity travelable area of ​​the vehicle body width constraint linear wherein a travelable area; said step 4) comprises the following steps: 4.1) for feature extraction data segments into close close to the X-axis and Y-axis into two categories; 4.2) according to the extracted feature different data segments near the axes of different characteristics selected linear equation, for the X-axis data segments near y = kx + b using a fitting way to determine characteristic parameters (k, b) corresponding to the data segments near the Y-axis is used X = ky + b fitting way to determine characteristic parameters (k, b) corresponding to; 4.3) is calculated for each segment for its characteristic feature extraction equation of the linear mean square error 如果均方误差值大于等于均方误差阈值,则找到数据片段中距离特征直线方程最远的点,以此点将数据片段一分为二,然后对两个新的数据片段分别采用步骤4. 2)的方法进行直线拟合直至均方误差值小于均方误差阈值; 所述步骤5)包括以下具体步骤: 5. 1)选择直线特征满足可行驶区域平坦约束的数据片段,删除不满足可行驶区域平坦约束的数据片段,所述可行驶区域平坦约束指直线与局部3维直角坐标系中Y轴夹角大于等于设定的阈值Τγ; 5.2) 经过步骤5. 1)后,合并满足可行驶区域连续性约束的数据片段,作为描述可行驶区域的候选对象;所述可行驶区域连续性约束是指相邻数据片段的高度差值以及水平距离差值分别小于设定的对应阈值Tz、Td; 5.3) 经过步骤5. 2)后,选择候选对象中满足车身宽度约束、且让车辆尽量少做横向运动的候选对象用于描述可行 If the MSE is equal to the mean square error is greater than a threshold value, the data segment from the feature point furthest linear equation is found, this point is divided into two data segments, then two new data segments Step 4 respectively. 2) the method of fitting a straight line until the mean square error is smaller than the mean square error threshold; step 5) comprises the following steps: 5.1) select line characteristic data segments satisfies constraints flat travelable area, delete can not satisfy data segment with a flat area constraint, the constraint may be planar region refers to a straight line with the local 3-dimensional rectangular coordinate system in the Y-axis is greater than an angle equal to the set threshold value Τγ; 5.2) after step 5.1), the merger may be satisfied continuity constraint data segment traveling area, as may be described with candidate region; the travelable area continuity constraint is the difference in height and a horizontal distance difference between adjacent data segments are smaller than the threshold value corresponding to the set Tz, Td; 5.3) after step 5.2), the selected candidate object satisfies the vehicle width constraints, let the vehicle as little as possible candidates for describing the transverse movement of feasible 驶区域。 Driving area.
2. 根据权利要求1所述一种基于激光雷达的可行驶区域检测方法,其特征在于:所述步骤1)包括以下具体步骤: 1. 1)以激光雷达安装位置为参考点定义局部3维直角坐标系,然后测量激光雷达距离水平面的安装高度H,并计算激光雷达倾斜向下扫描面相对于垂直方向的夹角α; 1. 2)建立激光雷达扫描极坐标到局部3维直角坐标系中坐标的转换关系,根据转换关系获得激光雷达数据点Pi对应的3维直角坐标值,每一个激光雷达扫描极坐标数据点包括发射角度I与扫描距离1两个参数,则P1= (cU,对应局部3维直角坐标系的坐标值为Pi=(Xi,yi,I),转化方程如下: Xi= sin α · cos β i·山Yi= sinP i · di Zi= H-cos a · cos β i · di〇 The travelable area detecting method based on the laser radar as claimed in claim 1, wherein: said step a) comprises the following steps: 1) laser radar mounted position as a reference point defined 3-dimensional partial Cartesian coordinate system, and the distance measuring laser radar installation height level H, and calculating a scanning laser radar downwardly inclined surface angle α to the vertical; 1.2) to establish a laser radar scanning a three-dimensional polar coordinates to the local Cartesian coordinates conversion relation coordinates, conversion relation is obtained 3-dimensional rectangular coordinate values ​​of the laser radar according to the data corresponding to point Pi, a laser radar scanning each data point comprises a polar coordinate two parameters and emitting angle scanning distance I, the P1 = (cU, corresponding to local 3-dimensional coordinates of the rectangular coordinate system is Pi = (Xi, yi, I), conversion equations are as follows: Xi = sin α · cos β i · Hill Yi = sinP i · di Zi = H-cos a · cos β i · di〇
3. 根据权利要求1所述一种基于激光雷达的可行驶区域检测方法,其特征在于:所述步骤2)包括以下具体步骤: 根据相邻数据点高度值之差将全部的激光雷达数据分为若干片段Sk: Sk={PJ,i=Nsk, - ,Nek stIZs-Zs !I ^ Tz, IZe-Ze+11^ Tz 其中,? The travelable area detecting method based on the laser radar as claimed in claim 1, wherein: said step 2) comprises the following steps: The difference between the height values ​​of adjacent data points of all the data points lidar several segments Sk:! Sk = {PJ, i = Nsk, -, Nek stIZs-Zs I ^ Tz, IZe-Ze + 11 ^ Tz wherein? 1表示激光雷达数据点,Nsk是参与滤波计算的第一个数据点的索引,Nek是参与滤波计算最后一个数据点的索引,zs表示片段起始激光雷达数据点的高度,%表示片段末尾激光雷达数据点的高度,Tz表示高度差阈值。 1 denotes a laser radar data points, participation of Nsk is the first data point index filter calculation, filter calculation involved Nek index of the last data point ZS indicates the starting segment height lidar data points, expressed in% of the laser end segments height radar data points, Tz indicates the height difference threshold.
4. 根据权利要求1所述一种基于激光雷达的可行驶区域检测方法,其特征在于:所述滤波处理包括以下具体步骤: 对于数据片段,根据当前数据点? The travelable area detecting method based on the laser radar as claimed in claim 1, wherein: said filtering process comprises the following steps of: for data segments, based on the current data point? 1选择其设定窗口内的数据点进行均值滤波来估计数据点真值。 1 which is set to select the data points within the window mean filter to estimate the true value of the data point.
CN201310751817.8A 2013-12-31 2013-12-31 Travelable area detecting method based on laser radar CN103760569B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310751817.8A CN103760569B (en) 2013-12-31 2013-12-31 Travelable area detecting method based on laser radar

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310751817.8A CN103760569B (en) 2013-12-31 2013-12-31 Travelable area detecting method based on laser radar

Publications (2)

Publication Number Publication Date
CN103760569A CN103760569A (en) 2014-04-30
CN103760569B true CN103760569B (en) 2016-03-30

Family

ID=50527836

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310751817.8A CN103760569B (en) 2013-12-31 2013-12-31 Travelable area detecting method based on laser radar

Country Status (1)

Country Link
CN (1) CN103760569B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107851390A (en) * 2015-08-04 2018-03-27 日产自动车株式会社 Step detection means and step detection method

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20160054825A (en) * 2014-11-07 2016-05-17 현대모비스 주식회사 Apparatus and method for judging drivable space
CN104408442A (en) * 2014-12-15 2015-03-11 长春理工大学 Acceleration sensor correction and laser radar-based pavement type identification method and device
CN104597453B (en) * 2015-01-27 2017-06-20 长春理工大学 The vehicle safe driving method for detecting area and device of inertial measuring unit correction
CN104569998B (en) * 2015-01-27 2017-06-20 长春理工大学 The detection method and device in the vehicle safe driving region based on laser radar
CN104765025B (en) * 2015-03-31 2017-09-29 浙江大学 A kind of processing method and processing system of trailer-mounted radar data
CN105137412B (en) * 2015-08-19 2017-10-20 重庆大学 A kind of 2D laser radars range image middle conductor feature Accurate Curve-fitting method
CN106570446B (en) * 2015-10-12 2019-02-01 腾讯科技(深圳)有限公司 The method and apparatus of lane line drawing
CN105957182B (en) * 2016-04-21 2018-08-03 深圳市元征科技股份有限公司 A kind of method and device of rectilinear direction that correcting instruction vehicle traveling
CN107871399A (en) * 2016-12-27 2018-04-03 珠海市杰理科技股份有限公司 Vehicular automatic driving system and method
CN107167811B (en) * 2017-04-26 2019-05-03 西安交通大学 The road drivable region detection method merged based on monocular vision with laser radar
CN108868268A (en) * 2018-06-05 2018-11-23 西安交通大学 Based on point to identity distance from the unmanned vehicle position and orientation estimation method being registrated with cross-correlation entropy
CN110008941A (en) * 2019-06-05 2019-07-12 长沙智能驾驶研究院有限公司 Drivable region detection method, device, computer equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102622890A (en) * 2012-04-16 2012-08-01 公安部第三研究所 Laser ranging-based road vehicle information acquisition node device
CN102779280A (en) * 2012-06-19 2012-11-14 武汉大学 Traffic information extraction method based on laser sensor
CN103093088A (en) * 2013-01-09 2013-05-08 西安费斯达自动化工程有限公司 Safety evaluation method for steep slope and winding road

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3472815B2 (en) * 2000-10-31 2003-12-02 国土交通省国土技術政策総合研究所長 Crossing pedestrian detection method
JP5192868B2 (en) * 2008-03-25 2013-05-08 株式会社Ihi Traveling area discrimination device and traveling area discrimination method for mobile robot

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102622890A (en) * 2012-04-16 2012-08-01 公安部第三研究所 Laser ranging-based road vehicle information acquisition node device
CN102779280A (en) * 2012-06-19 2012-11-14 武汉大学 Traffic information extraction method based on laser sensor
CN103093088A (en) * 2013-01-09 2013-05-08 西安费斯达自动化工程有限公司 Safety evaluation method for steep slope and winding road

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107851390A (en) * 2015-08-04 2018-03-27 日产自动车株式会社 Step detection means and step detection method
EP3333829B1 (en) * 2015-08-04 2019-10-09 Nissan Motor Co., Ltd. Step detection device and step detection method

Also Published As

Publication number Publication date
CN103760569A (en) 2014-04-30

Similar Documents

Publication Publication Date Title
JP4797794B2 (en) Pedestrian detection device and pedestrian detection method
JP4650079B2 (en) Object detection apparatus and method
Yu et al. 3D reconstruction of road surfaces using an integrated multi-sensory approach
Yang et al. Semi-automated extraction and delineation of 3D roads of street scene from mobile laser scanning point clouds
EP2574958B1 (en) Road-terrain detection method and system for driver assistance systems
KR101411668B1 (en) A calibration apparatus, a distance measurement system, a calibration method, and a computer readable medium recording a calibration program
Hu et al. Automatic road extraction from dense urban area by integrated processing of high resolution imagery and lidar data
US20100100268A1 (en) Enhanced clear path detection in the presence of traffic infrastructure indicator
Hautière et al. Real-time disparity contrast combination for onboard estimation of the visibility distance
Zhou et al. Mapping curbstones in airborne and mobile laser scanning data
Moras et al. Credibilist occupancy grids for vehicle perception in dynamic environments
US20110242319A1 (en) Image processing system and position measurement system
JP2004037239A (en) Identical object judging method and system, and misregistration correcting method and system
Hata et al. Road marking detection using LIDAR reflective intensity data and its application to vehicle localization
US8238610B2 (en) Homography-based passive vehicle speed measuring
US20110243457A1 (en) Scene matching reference data generation system and position measurement system
Bertozzi et al. Obstacle detection and classification fusing radar and vision
JP6259928B2 (en) Lane data processing method, apparatus, storage medium and equipment
JP4363295B2 (en) Plane estimation method using stereo images
JP2015148601A (en) System and method for mapping, localization and pose correction
CN102779280B (en) Traffic information extraction method based on laser sensor
CN101750049A (en) Monocular vision vehicle distance measuring method based on road and vehicle information
Chen et al. Building reconstruction from LIDAR data and aerial imagery
Gern et al. Vision-based lane recognition under adverse weather conditions using optical flow
KR20130021018A (en) Method for separating object in three dimension point clouds

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model