WO2023005566A1 - 基于点云尾气过滤技术的水雾尾气噪声处理方法 - Google Patents
基于点云尾气过滤技术的水雾尾气噪声处理方法 Download PDFInfo
- Publication number
- WO2023005566A1 WO2023005566A1 PCT/CN2022/101873 CN2022101873W WO2023005566A1 WO 2023005566 A1 WO2023005566 A1 WO 2023005566A1 CN 2022101873 W CN2022101873 W CN 2022101873W WO 2023005566 A1 WO2023005566 A1 WO 2023005566A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- exhaust gas
- point cloud
- water mist
- mist
- obstacles
- Prior art date
Links
- 239000003595 mist Substances 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 16
- 238000001914 filtration Methods 0.000 title claims abstract description 10
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 41
- 238000005516 engineering process Methods 0.000 claims description 8
- 238000003672 processing method Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000001514 detection method Methods 0.000 abstract description 5
- 238000013145 classification model Methods 0.000 abstract description 3
- 238000005065 mining Methods 0.000 description 4
- 230000008447 perception Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
Definitions
- the invention relates to a water mist tail gas noise processing method based on point cloud tail gas filtering technology.
- lidar As a sensor to sense surrounding obstacles, it is easily affected by the exhaust gas of the own vehicle, especially when the truck has high power and the working environment is relatively cold, the exhaust gas or water vapor generated by the own vehicle will be more obvious.
- the laser radar produces interference, which may easily cause false detection of obstacles, causing the main vehicle to brake frequently or even stop;
- the existing water mist exhaust noise processing methods generally train a special classification model or detection model to detect the water mist exhaust noise that may be encountered, or combine other sensor information, such as camera information, to pre-process the point cloud data , before the perception system performs perception, remove the noise of water mist exhaust.
- a common noise solution is to increase training data and use more data containing water mist exhaust to train detection models or classification models, but this method requires a large amount of new data , and the generalization ability is poor, when the automatic driving system encounters new models or different scenes, it will be easily disturbed by water mist and exhaust gas again, and this solution consumes computing resources and consumes a lot of computing time; another method is to use the camera
- the combination of information and lidar projects the image segmentation results onto the point cloud, but the shortcomings of this method are also obvious, such as poor generalization, requiring more computing resources, and is easily affected by dark lighting.
- the purpose of the present invention is to provide a water mist exhaust noise processing method based on point cloud exhaust filtration technology, which overcomes the traditional technology that is easy to cause false detection of obstacles, makes the main vehicle brake frequently or even stops, and requires more computing resources.
- the disadvantages of mining trucks effectively ensure the driving safety of automatic driving vehicles for mining trucks.
- a water mist exhaust noise processing method based on point cloud exhaust filtering technology includes the following steps: first, after using the CNN model to detect obstacles in the point cloud, a number of obstacles in the point cloud image will be obtained, and only Identify the water mist obstacles and delete them to achieve the purpose of removing water mist noise, and for each obstacle, make the following judgments to know whether it is water mist;
- the first level indicates that the range is [min, min+step), ..., and the last level is [min+(k-1)step ,max];
- H the probability that the larger H is, the greater the probability is water mist, and the smaller H is, the smaller the probability is water mist
- the present invention mainly provides a water mist exhaust noise processing method based on point cloud exhaust filtering technology, which can effectively solve the problem that the prior art requires a large amount of new data, and has poor generalization ability and is easily affected by light and darkness. Defects, using the point cloud topology point cloud exhaust filtering technology, through two distinct point cloud topologies to distinguish the two, the more evenly distributed obstacles between points, several obstacles in the point cloud map , just identify the water mist obstacles and delete them to achieve the purpose of removing the water mist noise, effectively ensuring the safe and accurate driving of unmanned mining vehicles.
- a water mist exhaust noise processing method based on point cloud exhaust filtering technology includes the following steps: first, after using the CNN model to detect obstacles in the point cloud, a number of obstacles in the point cloud image will be obtained, and only Identify the water mist obstacles and delete them to achieve the purpose of removing water mist noise, and for each obstacle, make the following judgments to know whether it is water mist;
- the first level indicates that the range is [min, min+step), ..., and the last level is [min+(k-1)step ,max];
- H the probability
- H the probability
- Water mist exhaust noise (hereinafter referred to as water mist) is relatively evenly distributed and can penetrate.
- the lidar scans the water mist, it can generate a uniformly distributed obstacle in space, while the general sense in road driving Obstacles, due to impenetrability and self-occlusion, the shape of the generated point cloud is often a surface of an obstacle that can be scanned by a lidar, that is, the points are concentrated on one side of the obstacle, such as most vehicles
- the shape of the point cloud is often an "L" shape.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Evolutionary Biology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Processing (AREA)
Abstract
一种基于点云尾气过滤技术的水雾尾气噪声处理方法。现有技术水雾尾气噪声处理方法,增加了训练数据,使用更多包含水雾尾气的数据训练检测模型或分类模型,需要大量的新数据,而且泛化能力差,遇到新的车型或者不同场景时,会再次容易被水雾尾气等干扰。本发明该方法包括如下步骤:首先是在使用CNN模型检测完点云中的障碍物后,会得该点云图中若干障碍物,只需识别出其中的水雾障碍物,并将其删除,即可达到去除水雾噪声的目的,并对于每一个障碍物,进行以下判断,即可知道其是否为水雾;(1)对于3d矩形box表示的一个障碍物,在xoy平面,将其栅格化;(2)用pi表示落在第i级的频率。本发明用于基于点云尾气过滤技术的水雾尾气噪声处理方法。
Description
本发明涉及一种基于点云尾气过滤技术的水雾尾气噪声处理方法。
在矿卡自动驾驶场景中,使用激光雷达作为传感器感知周围障碍物时,容易受到自车尾气影响,尤其是卡车功率大、工作环境较为寒冷时,自车产生的尾气或者水汽会比较明显,对激光雷达产生干扰,容易造成障碍物误检,使主车频频急刹甚至刹停;
现有技术的水雾尾气噪声处理方法,一般为训练一个专门的分类模型或者检测模型,检测可能遇到的水雾尾气噪声,或者结合其他传感器信息,如相机信息,对点云数据进行前处理,在感知***进行感知之前,将水雾尾气噪声去除,常见的噪声解决方案为增加训练数据,使用更多包含水雾尾气的数据训练检测模型或者分类模型,但是这种方式需要大量的新数据,而且泛化能力差,自动驾驶***遇到新的车型或者不同场景时,会再次容易被水雾尾气等干扰,且该方案耗费计算资源,计算时间消耗较大;还有一种方法是使用摄像头信息和激光雷达相结合的方式,将图像分割结果投影到点云之上,但是该方法的缺点同样明显,泛化性差,需要较多计算资源,而且极易受到光照明暗的影响。
发明内容
本发明的目的是提供一种基于点云尾气过滤技术的水雾尾气噪声处理方法,该方法克服了传统技术容易造成障碍物误检、使主车频频急刹甚至刹停、需要较多计算资源的弊端,有效保证了矿卡自动驾驶车的行驶安全。
上述的目的通过以下的技术方案实现:
一种基于点云尾气过滤技术的水雾尾气噪声处理方法,该方法包括如下步骤:首先是在使用CNN模型检测完点云中的障碍物后,会得该点云图中若干障碍物,只需识别出其中的水雾障碍物,并将其删除,即可达到去除水雾噪声的目的,并对于每一个障碍物,进行以下判断,即可知道其是否为水雾;
(1)对于3d矩形box表示的一个障碍物,在xoy平面,将其栅格化,举例来说,分为n*n的矩形格子,然后计算在每个格子里,障碍物点云的数量,这样就可以得到一张矩阵,所述的矩阵上每个数字表示这个位置的格子里点的数量,所述的格子里最大数字为max,最小值为min,设置一个超参数k,step=(max-min)/k,
这样就可以得到k个等级,然后统计落在每一级里点的数量,如,第一级表示范围为[min,min+step),…,最后一级为[min+(k-1)step,max];
即可表示点云的拓扑分布,其中H为概率H越大,是水雾的概率越大,H越小,是水雾的概率越小;
(3)点和点之间分布越均匀的障碍物,是水雾的概率越大,点分布集中在一个平面的障碍物,是水雾的概率越小。
1.本发明主要是提供一种基于点云尾气过滤技术的水雾尾气噪声处理方法,该方法能够有效解决现有技术需要大量的新数据,而且泛化能力差,易受到光照明暗的影响的缺陷,采用点云拓扑结构的点云尾气过滤技术,通过两种差异明显的点云拓扑结构,将二者区分开来,点和点之间分布越均匀的障碍物,点云图中若干障碍物,只需识别出其中的水雾障碍物,并将其删除,即可达到去除水雾噪声的目的,有效保证无人驾驶矿车能够安全精准行驶。
具体实施方式
实施例1:
一种基于点云尾气过滤技术的水雾尾气噪声处理方法,该方法包括如下步骤:首先是在使用CNN模型检测完点云中的障碍物后,会得该点云图中若干障碍物,只需识别出其中的水雾障碍物,并将其删除,即可达到去除水雾噪声的目的,并对于每一个障碍物,进行以下判断,即可知道其是否为水雾;
(1)对于3d矩形box表示的一个障碍物,在xoy平面,将其栅格化,举例来说,分为n*n的矩形格子,然后计算在每个格子里,障碍物点云的数量,这样就可以得到一张矩阵,所述的矩阵上每个数字表示这个位置的格子里点的数量,所述的格子里最大数字为max,最小值为min,设置一个超参数k,step=(max-min)/k,
这样就可以得到k个等级,然后统计落在每一级里点的数量,如,第一级表示范围为[min,min+step),…,最后一级为[min+(k-1)step,max];
即可表示点云的拓扑分布,其中H为概率,H越大,是水雾的概率越大,H越小,是水雾的概率越小;
(3)点和点之间分布越均匀的障碍物,是水雾的概率越大,点分布集中在一个平面的障碍物,是水雾的概率越小。
点云数据中,水雾尾气噪声点云分布和普通障碍物点云分布,有着非常明显的差别,利用这种差别,可以很好的将二者区分开来,达到去噪的目的;
水雾尾气噪声(以下以水雾指代),分布较为均匀且可以穿透,当激光雷达扫描到水雾时,能够在空间中产生一个点分布均匀的障碍物,而道路驾驶中一般意义的障碍物,由于不可穿透、自遮挡的原因,产生的点云形状往往为一个激光雷达可以扫描到的障碍物的一个面,即点集中在障碍物的一侧的面上,如大部分车辆的点云形状,往往为一个“L”型。通过两种差异明显的点云拓扑结构,我们可以将二者区分开来,点和点之间分布越均匀的障碍物,是水雾的概率越大,点分布集中在一个平面的障碍物,是水雾的概率越小。
Claims (1)
- 一种基于点云尾气过滤技术的水雾尾气噪声处理方法,其特征是:该方法包括如下步骤:首先是在使用CNN模型检测完点云中的障碍物后,会得该点云图中若干障碍物,只需识别出其中的水雾障碍物,并将其删除,即可达到去除水雾噪声的目的,并对于每一个障碍物,进行以下判断,即可知道其是否为水雾;(1)对于3d矩形box表示的一个障碍物,在xoy平面,将其栅格化,举例来说,分为n*n的矩形格子,然后计算在每个格子里,障碍物点云的数量,这样就可以得到一张矩阵,所述的矩阵上每个数字表示这个位置的格子里点的数量,所述的格子里最大数字为max,最小值为min,设置一个超参数k,step=(max-min)/k,这样就可以得到k个等级,然后统计落在每一级里点的数量,如,第一级表示范围为[min,min+step),…,最后一级为[min+(k-1)step,max];即可表示点云的拓扑分布,其中H为概率,H越大,是水雾的概率越大,H越小,是水雾的概率越小;(3)点和点之间分布越均匀的障碍物,是水雾的概率越大,点分布集中在一个平面的障碍物,是水雾的概率越小。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110853179.5 | 2021-07-27 | ||
CN202110853179.5A CN113763263A (zh) | 2021-07-27 | 2021-07-27 | 基于点云尾气过滤技术的水雾尾气噪声处理方法 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2023005566A1 true WO2023005566A1 (zh) | 2023-02-02 |
Family
ID=78788001
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2022/101873 WO2023005566A1 (zh) | 2021-07-27 | 2022-06-28 | 基于点云尾气过滤技术的水雾尾气噪声处理方法 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN113763263A (zh) |
WO (1) | WO2023005566A1 (zh) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113763263A (zh) * | 2021-07-27 | 2021-12-07 | 华能伊敏煤电有限责任公司 | 基于点云尾气过滤技术的水雾尾气噪声处理方法 |
CN115271096B (zh) * | 2022-07-27 | 2023-08-18 | 阿波罗智能技术(北京)有限公司 | 点云处理和机器学习模型训练方法、装置及自动驾驶车辆 |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160103208A1 (en) * | 2014-10-14 | 2016-04-14 | Hyundai Motor Company | System for filtering lidar data in vehicle and method thereof |
CN110031823A (zh) * | 2019-04-22 | 2019-07-19 | 上海禾赛光电科技有限公司 | 可用于激光雷达的噪点识别方法以及激光雷达*** |
CN111291697A (zh) * | 2020-02-19 | 2020-06-16 | 北京百度网讯科技有限公司 | 用于识别障碍物的方法和装置 |
CN111880196A (zh) * | 2020-06-29 | 2020-11-03 | 安徽海博智能科技有限责任公司 | 一种无人矿车抗干扰方法、***及计算机设备 |
CN112513679A (zh) * | 2020-04-15 | 2021-03-16 | 华为技术有限公司 | 一种目标识别的方法和装置 |
CN112733813A (zh) * | 2021-03-30 | 2021-04-30 | 北京三快在线科技有限公司 | 一种数据降噪的方法及装置 |
CN113763263A (zh) * | 2021-07-27 | 2021-12-07 | 华能伊敏煤电有限责任公司 | 基于点云尾气过滤技术的水雾尾气噪声处理方法 |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10482619B2 (en) * | 2017-07-27 | 2019-11-19 | AI Incorporated | Method and apparatus for combining data to construct a floor plan |
CN109597097B (zh) * | 2018-12-06 | 2023-04-18 | 北京主线科技有限公司 | 基于多线激光的扫描式障碍物检测方法 |
CN111537994B (zh) * | 2020-03-24 | 2022-05-06 | 江苏徐工工程机械研究院有限公司 | 一种无人矿卡障碍物检测方法 |
CN111427047B (zh) * | 2020-03-30 | 2023-05-05 | 哈尔滨工程大学 | 一种大场景下自主移动机器人slam方法 |
CN112464994B (zh) * | 2020-11-05 | 2024-03-26 | 航天时代(青岛)海洋装备科技发展有限公司 | 基于PointNet网络的船艇尾浪识别与去除方法 |
-
2021
- 2021-07-27 CN CN202110853179.5A patent/CN113763263A/zh active Pending
-
2022
- 2022-06-28 WO PCT/CN2022/101873 patent/WO2023005566A1/zh unknown
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160103208A1 (en) * | 2014-10-14 | 2016-04-14 | Hyundai Motor Company | System for filtering lidar data in vehicle and method thereof |
CN110031823A (zh) * | 2019-04-22 | 2019-07-19 | 上海禾赛光电科技有限公司 | 可用于激光雷达的噪点识别方法以及激光雷达*** |
CN111291697A (zh) * | 2020-02-19 | 2020-06-16 | 北京百度网讯科技有限公司 | 用于识别障碍物的方法和装置 |
CN112513679A (zh) * | 2020-04-15 | 2021-03-16 | 华为技术有限公司 | 一种目标识别的方法和装置 |
CN111880196A (zh) * | 2020-06-29 | 2020-11-03 | 安徽海博智能科技有限责任公司 | 一种无人矿车抗干扰方法、***及计算机设备 |
CN112733813A (zh) * | 2021-03-30 | 2021-04-30 | 北京三快在线科技有限公司 | 一种数据降噪的方法及装置 |
CN113763263A (zh) * | 2021-07-27 | 2021-12-07 | 华能伊敏煤电有限责任公司 | 基于点云尾气过滤技术的水雾尾气噪声处理方法 |
Also Published As
Publication number | Publication date |
---|---|
CN113763263A (zh) | 2021-12-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2023005566A1 (zh) | 基于点云尾气过滤技术的水雾尾气噪声处理方法 | |
Kurup et al. | Dsor: A scalable statistical filter for removing falling snow from lidar point clouds in severe winter weather | |
US9836660B2 (en) | Vision-based wet road surface condition detection using tire tracks | |
CN103760569B (zh) | 一种基于激光雷达的可行驶区域检测方法 | |
US9594964B2 (en) | Vision-based wet road surface detection using texture analysis | |
WO2021012987A1 (zh) | 物体检测方法、装置、***及设备 | |
CN113917487A (zh) | 基于激光雷达的封闭道路路沿和可行驶区域检测方法 | |
CN104050637A (zh) | 基于两次引导滤波的快速图像去雾方法 | |
WO2023092870A1 (zh) | 一种适用于自动驾驶车辆的挡土墙检测方法及*** | |
CN106326850A (zh) | 快速车道线检测方法 | |
CN115115595A (zh) | 一种面向森林火灾监测的机载激光雷达和红外相机的实时标定方法 | |
CN107194342A (zh) | 基于逆透视变换的车道线检测方法 | |
CN113243354A (zh) | 一种基于计算机视觉算法和人工智能技术的激光驱鸟方法、装置 | |
WO2020248118A1 (zh) | 点云处理方法、***、设备及存储介质 | |
CN112562419A (zh) | 一种基于离线多目标跟踪的天气规避区划设方法 | |
CN113313654B (zh) | 激光点云滤波去噪方法、***、设备及存储介质 | |
CN108229327A (zh) | 一种基于背景重建的车道线检测方法、装置与*** | |
CN114035202A (zh) | 基于激光雷达的障碍物检测方法及装置 | |
Kamiyama et al. | Recognition of parking spaces on dry and wet road surfaces using received light intensity of laser for ultra small EVs | |
WO2018180104A1 (ja) | 判定装置、判定方法並びに判定用プログラム及びデータ構造 | |
Cao et al. | Filter Methods for Removing Falling Snow from Light Detection and Ranging Point Clouds in Snowy Weather. | |
ApinayaPrethi et al. | Fog detection and visibility measurement using SVM | |
WO2024065685A1 (zh) | 一种处理点云的方法和雷达 | |
CN117974485A (zh) | 一种激光点云去噪方法、装置及设备 | |
Song et al. | Traffic meteorological visibility estimation based on homogenous area extraction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22848173 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |