CN112489123B - Three-dimensional positioning method for surface target of truck in steel mill reservoir area - Google Patents

Three-dimensional positioning method for surface target of truck in steel mill reservoir area Download PDF

Info

Publication number
CN112489123B
CN112489123B CN202011186575.9A CN202011186575A CN112489123B CN 112489123 B CN112489123 B CN 112489123B CN 202011186575 A CN202011186575 A CN 202011186575A CN 112489123 B CN112489123 B CN 112489123B
Authority
CN
China
Prior art keywords
target
irregular
regular
point cloud
positioning
Prior art date
Legal status (The legal status 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 status listed.)
Active
Application number
CN202011186575.9A
Other languages
Chinese (zh)
Other versions
CN112489123A (en
Inventor
牛丹
魏双
陈夕松
王思敏
许翠红
刘子璇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Yunniu Intelligent Technology Co ltd
Jiangyin Zhixing Industrial Control Technology Co ltd
Southeast University
Original Assignee
Nanjing Yunniu Intelligent Technology Co ltd
Jiangyin Zhixing Industrial Control Technology Co ltd
Southeast University
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 Nanjing Yunniu Intelligent Technology Co ltd, Jiangyin Zhixing Industrial Control Technology Co ltd, Southeast University filed Critical Nanjing Yunniu Intelligent Technology Co ltd
Priority to CN202011186575.9A priority Critical patent/CN112489123B/en
Publication of CN112489123A publication Critical patent/CN112489123A/en
Application granted granted Critical
Publication of CN112489123B publication Critical patent/CN112489123B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • 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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • 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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • 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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Image Analysis (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

The invention discloses a three-dimensional positioning method for a surface target of a truck in a steel mill reservoir area, belonging to the field of mechanical automation and comprising the following specific steps of: obtaining point clouds of an irregular target and a regular target; obtaining a main direction vector and a central coordinate of each irregular target; obtaining the central coordinates of a standard irregular target; obtaining the center coordinates of the sorted irregular targets; obtaining the position coordinates of the irregular target; obtaining the central coordinates of the regular target point cloud; and obtaining the coordinates of the target position under the ground coordinate system. The invention is applied to the loading and unloading operation of the steel storage area, and can improve the accuracy of segmentation and positioning of irregular targets and regular targets on the truck. Identifying and segmenting target point cloud blocks on the surface of the truck by adopting a PointCNN segmentation network, and finally calculating the positions of an irregular target and a regular target by a specific positioning algorithm; the invention solves the problems of irregular target segmentation and positioning in the prior positioning technology, and improves the applicability and the identification efficiency of the positioning system.

Description

Three-dimensional positioning method for surface target of truck in steel mill reservoir area
Technical Field
The invention belongs to the field of mechanical automation, and relates to a three-dimensional positioning method for a surface target of a truck in a steel mill reservoir area; in particular to a positioning method of an unmanned travelling system of a steel plant based on a deep learning algorithm.
Background
With the development of automation technology, the full-automatic loading and unloading of the steel plant reservoir area has become an important trend. In order to realize loading and unloading automation, the core task is to accurately position the material on the truck and the position of a stack position.
The traditional automatic positioning method based on three-dimensional laser is a research hotspot of current loading and unloading positioning, however, the method can only identify regular steel coils and stack positions, or can only position truck carriages with specific shapes and specifications, and does not have the identification capability of irregular type targets, so that the positioning efficiency is limited by target types, and the efficiency is low.
The invention discloses a three-dimensional positioning method for a surface target of a truck in a steel mill reservoir area on the basis of a traditional three-dimensional identification and positioning system. Taking irregular wooden strip stacking positions common in steel mills as an example, the wooden strip stacking positions are often seriously worn due to long-time use, no regular geometric model corresponds to the irregular wooden strip stacking positions, and the regular characteristic identification algorithm is difficult to accurately divide the wooden strip stacking positions. The method trains the PointCNN segmentation network by manufacturing a data set of site regular steel coils, regular stack positions and irregular batten stack positions, so that the network can identify and segment three kinds of target point clouds with large shape differences, and then designs a corresponding target positioning algorithm, thereby effectively solving the problem of irregular target identification and positioning in the prior art, and improving the applicability and the identification efficiency of the system in different scenes.
Disclosure of Invention
Aiming at the problems, the invention provides a three-dimensional positioning method for a surface target of a truck in a steel mill reservoir area.
The technical scheme of the invention is as follows: a three-dimensional positioning method for a surface target of a truck in a steel mill reservoir area comprises the following specific steps:
step (1.1), adopting PointCNN network branchCutting the point cloud on the surface of the truck to obtain a point cloud H of an irregular target1,H2...HmAnd point cloud P of regular objects1,P2...Ps；
Step (1.2), point cloud H of the segmented irregular target1,H2...HmObtaining a main direction vector w of each irregular target by adopting a directional bounding box algorithmh1,wh2...whmAnd center coordinate ch1,ch2...chm；
Step (1.3), carrying out vector w of main direction of irregular targeth1,wh2...whmChecking, and eliminating the placed irregular target which inclines and deviates from the center of the truck carriage to obtain the standard center coordinate c of the irregular targeth1,ch2...chlWherein l is less than or equal to m;
step (1.4), calculating the irregular target center coordinate c of the directional bounding box according to the parking direction of the truckh1,ch2...chlSorting is carried out to obtain a sorted central coordinate ch1',ch2'...chl'；
Step (1.5), calculating the middle points of adjacent irregular targets in pairs to obtain the position coordinates h of the irregular targets1,h2...hl/2；
Step (1.6), obtaining a regular target point cloud P by adopting a regular target positioning algorithm1,P2...PsCentral coordinate p of1,p2...pfWherein f is less than or equal to s;
and (1.7) converting the obtained position coordinates of all the irregular targets and the regular targets to obtain the target position coordinates in the ground coordinate system.
Furthermore, the three-dimensional positioning method is that the three-dimensional laser point cloud on the surface of the truck in the steel mill reservoir area is segmented and positioned, and a target positioning algorithm is designed to calculate the positions of the irregular target and the regular target.
Further, in the step (1.1), the irregular target is a batten stacking position, and the regular target is a regular steel coil and a regular stacking position.
Further, in step (1.1), the PointCNN network partition is obtained by training a library area field data set, and the field data set is specifically produced by the following steps:
(1.1.1) collecting three-dimensional laser data of a field positioning target, and dividing the three-dimensional laser data into point cloud blocks only containing a single target;
(1.1.2) increasing or reducing the number of each target point cloud to a set point cloud number through up-sampling and down-sampling, and simultaneously carrying out averaging and normalization on the point clouds;
(1.1.3) repeating (1.1.1) - (1.1.2), making a category data set, and labeling a category label;
(1.1.4) randomly dividing the manufactured category data and the labels into a training set and a testing set, and synthesizing an HDF5 file required by PointCNN training by utilizing an HDF5 tool;
(1.1.5) inputting the training set and the testing set into a PointCNN network for training and testing to obtain a PointCNN segmentation network capable of segmenting different types of target point clouds.
Further, in the step (1.6), the regular target positioning algorithm includes a positioning algorithm of a regular steel coil and a positioning algorithm of a regular stack position, wherein the positioning algorithm of the regular steel coil specifically includes the following steps:
(1.6.1) for the regular steel coil point cloud segmented by the PointCNN, obtaining model parameters of the steel coil point cloud by using a RANSAC cylinder fitting algorithm, wherein a cylinder geometric model is as follows:
Figure BDA0002751605680000021
in the formula (x)0,y0,z0) Is a point on the cylindrical shaft, (m, n, l) is a unit direction vector on the axis of the steel coil, and r is the radius of the cylinder;
(1.6.2) projecting the steel coil point cloud onto the obtained model central axis vector (m, n, l), wherein the projection formula is as follows:
Figure BDA0002751605680000022
where k (m, n, l) is an axial unit vector, P0(x0,y0,z0) Is a point on the cylinder axis, Pi(xi,yi,zi) (i 1, 2.. n.) is the point cloud on the circular arc, n is the number of the point clouds, Qi(xi,yi,zi) Is PiA projected point on the cylindrical axis;
(1.6.3) calculating the mean value of all the projection points to obtain the central coordinate p of the steel coil:
Figure BDA0002751605680000031
(1.6.4) calculating the directional bounding box of the steel coil point cloud to obtain the steel coil width information;
(1.6.5), verifying the steel coil according to the z coordinate, wherein if the central coordinate z is smaller than the center z' of the bounding box, the central coordinate is the central coordinate of the effective steel coil, otherwise, the central coordinate is discarded.
The invention has the beneficial effects that: the invention discloses a method for positioning a surface target of a truck in a steel mill reservoir area. By training the PointCNN segmentation network, the invention effectively solves the problem of irregular target positioning in the prior positioning technology and improves the applicability and the recognition efficiency of the positioning system.
Drawings
FIG. 1 is a flow chart of the architecture of the present invention;
FIG. 2 is a schematic view of the warehousing operation of the present invention;
FIG. 3 is a schematic diagram of a PointCNN split network according to the present invention;
FIG. 4 is a schematic view of a steel coil in the present invention;
FIG. 5 is a schematic view of a standard berth in the present invention;
FIG. 6 is a schematic illustration of a stuff pile in the present invention;
FIG. 7 is a schematic view illustrating the recognition effect of the steel coil according to the present invention;
FIG. 8 is a schematic diagram of the identification effect of the stacking position of the battens in the invention.
Detailed Description
In order to more clearly illustrate the technical solution of the present invention, the following detailed description is made with reference to the accompanying drawings:
the invention provides a three-dimensional positioning method for a truck surface target in a steel mill reservoir area, which comprises the steps of using a PointCNN partition network to partition and identify point clouds of regular and irregular targets, and then calculating the central position of the target by adopting designed various category positioning algorithms; the method comprises the following steps of obtaining a compartment point cloud containing a target by reducing and filtering three-dimensional laser original data.
The method takes a cold-rolled raw material warehouse of a certain steel mill as a research object, a three-dimensional laser acquisition device is arranged above a parking space of a warehouse area, and hardware equipment comprises a ground operation station, a PLC electrical control cabinet and a hand-held remote controller; the specific implementation flow of the method is shown in figure 1; wherein, the software development tools are Visual Studio2015, Qt5.8.0, Point Cloud Library and MySQL database, and the Point CNN operation platform is Pycharm, Tensorflow and Anaconda 2; the warehousing operation process of the method in a steel mill is shown in figure 2, a PointCNN segmentation network is shown in figure 3, and the installation height of a field three-dimensional scanning device is 13 meters; after the point cloud data on the surface of the truck is acquired, the specific processing mode is as follows:
step (1.1): adopting a PointCNN network to segment the point cloud on the surface of the truck to obtain a point cloud H of an irregular target1,H2...HmAnd point cloud P of regular objects1,P2...Ps(ii) a Wherein the irregular target is a batten stacking position, and the regular target is a regular steel coil and a regular stacking position;
the PointCNN network is a three-dimensional segmentation network taking point cloud as input, and is mainly characterized by the production of a field data set and the training of the PointCNN segmentation network, wherein the production steps of the data set are as follows:
(1.1.1) acquiring three-dimensional laser data of a plurality of targets on site, and dividing the three-dimensional laser data into point cloud blocks only containing a single target;
(1.1.2) increasing or reducing the number of each target point cloud to a set point cloud number through up-sampling and down-sampling, and simultaneously carrying out averaging and normalization on the point clouds;
(1.1.3) repeating (1.1.1) - (1.1.2), making a certain number of category data sets, and labeling category labels;
(1.1.4) randomly dividing the manufactured category data and the labels into a training set and a testing set, and synthesizing an HDF5 file required by PointCNN training by utilizing an HDF5 tool;
(1.1.5) inputting the training set and the testing set into a PointCNN network for training and testing to obtain a PointCNN segmentation network capable of segmenting different types of target point clouds.
Through investigation, there are three main positioning targets in the reservoir area: the steel strip stacking device comprises a regular steel coil, a regular stacking position and a batten stacking position, wherein the regular steel coil is similar to a cylinder (see fig. 4), the regular stacking position is a pair of triangular columns (see fig. 5) with a certain inclination angle, the batten stacking position is a pair of long-strip-shaped wood blocks (see fig. 6), the batten stacking position is often seriously worn due to long-time use, and the batten stacking position is difficult to accurately divide by a conventional characteristic identification algorithm; in the field data set production, 486 pieces of three-dimensional data collected on the field are selected, the input point cloud number of the network is set to be 1024, and the table 1 shows produced labels and analog lists.
Table 1: label and category list
Figure BDA0002751605680000041
Step (1.2), point cloud H of the segmented irregular target1,H2...HmObtaining a main direction vector w of each irregular target (a batten stacking position) by adopting a directional bounding box algorithmh1,wh2...whmAnd center coordinate ch1,ch2...chm；
Step (1.3), carrying out main direction vector w on irregular target (wood bar stacking position)h1,wh2...whmChecking, and eliminating the irregular target (wood strip stacking position) which is inclined and deviated from the center of the truck carriage to obtain the standard central coordinate c of the irregular targeth1,ch2...chlWherein l is less than or equal to m;
step (1.4), calculating the center coordinate c of the irregular target (the stacking position of the battens) of the directional bounding box according to the parking direction of the truckh1,ch2...chlSorting is carried out to obtain a sorted central coordinate ch1',ch2'...chl'
Step (1.5), calculating the middle point of the stacking position of the adjacent battens in pairs to obtain the position coordinate h of the irregular target (the stacking position of the battens)1,h2...hl/2；
Step (1.6), obtaining a regular target point cloud P by adopting a regular target positioning algorithm1,P2...PsCentral coordinate p of1,p2...pfAnd f is less than or equal to s, wherein the regular steel coil positioning algorithm comprises a regular steel coil positioning algorithm and a regular stack position positioning algorithm, and the regular steel coil positioning algorithm comprises the following specific steps:
(1.6.1) for the regular steel coil point cloud segmented by the PointCNN, obtaining model parameters of the steel coil point cloud by using a RANSAC cylinder fitting algorithm, wherein a cylinder geometric model is as follows:
Figure BDA0002751605680000051
wherein (x)0,y0,z0) Is a point on the cylindrical shaft, (m, n, l) is a unit direction vector on the axis of the steel coil, and r is the radius of the cylinder;
(1.6.2) projecting the steel coil point cloud onto the obtained model central axis vector (m, n, l), wherein the projection formula is as follows:
Figure BDA0002751605680000052
where k (m, n, l) is an axial unit vector, P0(x0,y0,z0) Is a point on the cylinder axis, Pi(xi,yi,zi) (i 1, 2.. n.) is the point cloud on the circular arc, n is the number of the point clouds, Qi(xi,yi,zi) Is PiA projected point on the cylindrical axis;
(1.6.3) calculating the mean value of all the projection points to obtain the central coordinate p of the steel coil:
Figure BDA0002751605680000053
(1.6.4) calculating the directional bounding box of the steel coil point cloud to obtain the steel coil width information;
and (1.6.5) verifying the steel coil according to the z coordinate, wherein if the center coordinate z is less than the center z' of the bounding box, the steel coil is the center coordinate of the effective steel coil, and if not, the steel coil is discarded.
(1.6.6) converting the obtained position coordinates of all the irregular targets and the regular targets to obtain the target position coordinates in the ground coordinate system.
In order to verify the effectiveness of the method, the method trains the PointCNN network by using the manufactured field data set, and after testing, the trained neural network can distinguish three types with larger shape difference, so that the designed recognition positioning algorithm can achieve higher positioning precision; table 2 compares the segmentation effect of the non-neural network algorithm and the PointCNN-based segmentation network algorithm, and through experimental comparison, the present invention can effectively improve the positioning accuracy of the target, and the positioning effect is shown in fig. 7 and 8.
Table 2: accuracy of segmentation
Non-neural network algorithm PointCNN-based network
Regular steel coil 94.5％ 98.9％
Regular stack position 92.8％ 97.3％
Irregular batten pile position 85.6％ 96.2％
The foregoing is considered as illustrative of the preferred embodiments of the invention and is not to be construed as limiting the invention in any way. Therefore, any simple modification, equivalent changes and modifications of the above examples according to the technical essence of the present invention shall fall within the protection scope of the present invention, unless it departs from the technical solution.

Claims (1)

1. A three-dimensional positioning method for a surface target of a truck in a steel mill reservoir area is characterized by comprising the following specific steps:
step (1.1), adopting a PointCNN network to segment the point cloud on the surface of the truck to obtain a point cloud H of an irregular target1,H2...HmAnd point cloud P of regular objects1,P2...Ps；
Step (1.2), point cloud H of the segmented irregular target1,H2...HmObtaining a main direction vector w of each irregular target by adopting a directional bounding box algorithmh1,wh2...whmAnd center coordinate ch1,ch2...chm；
Step (1.3), carrying out vector w of main direction of irregular targeth1,wh2...whmChecking, and eliminating the placed irregular target which inclines and deviates from the center of the truck carriage to obtain the standard center coordinate c of the irregular targeth1,ch2...chlWherein l is less than or equal to m;
step (1.4), calculating the irregular target center coordinate c of the directional bounding box according to the parking direction of the truckh1,ch2...chlSorting is carried out to obtain a sorted central coordinate ch1',ch2'...chl'；
Step (1.5), calculating the middle points of adjacent irregular targets in pairs to obtain the position coordinates h of the irregular targets1,h2...hl/2；
Step (1.6), obtaining a regular target point cloud P by adopting a regular target positioning algorithm1,P2...PsCentral coordinate p of1,p2...pfWherein f is less than or equal to s;
step (1.7), converting the obtained position coordinates of all irregular targets and regular targets to obtain target position coordinates under a ground coordinate system;
specifically, the three-dimensional positioning method comprises the steps of segmenting and positioning three-dimensional laser point clouds on the surfaces of trucks in a steel mill reservoir area, and designing a target positioning algorithm to calculate the positions of irregular targets and regular targets;
in the step (1.1), the irregular target is a batten stacking position, and the regular target is a regular steel coil and a regular stacking position;
in the step (1.1), the PointCNN network segmentation is obtained by training a library area field data set, and the manufacturing steps of the field data set are as follows:
(1.1.1) collecting three-dimensional laser data of a field positioning target, and dividing the three-dimensional laser data into point cloud blocks only containing a single target;
(1.1.2) increasing or reducing the number of each target point cloud to a set point cloud number through up-sampling and down-sampling, and simultaneously carrying out averaging and normalization on the point clouds;
(1.1.3) repeating (1.1.1) - (1.1.2), making a category data set, and labeling a category label;
(1.1.4) randomly dividing the manufactured category data and the labels into a training set and a testing set, and synthesizing an HDF5 file required by PointCNN training by utilizing an HDF5 tool;
(1.1.5) inputting the training set and the test set into a PointCNN network for training and testing to obtain a PointCNN segmentation network capable of segmenting different types of target point clouds;
in the step (1.6), the regular target positioning algorithm includes a positioning algorithm of a regular steel coil and a positioning algorithm of a regular stack position, wherein the positioning algorithm of the regular steel coil specifically includes the following steps:
(1.6.1) for the regular steel coil point cloud segmented by the PointCNN, obtaining model parameters of the steel coil point cloud by using a RANSAC cylinder fitting algorithm, wherein a cylinder geometric model is as follows:
Figure FDA0003193239150000021
in the formula (x)0,y0,z0) Is a point on the cylindrical shaft, (m, n, l) is a unit direction vector on the axis of the steel coil, and r is the radius of the cylinder;
(1.6.2) projecting the steel coil point cloud onto the obtained model central axis vector (m, n, l), wherein the projection formula is as follows:
Figure FDA0003193239150000022
where k (m, n, l) is an axial unit vector, P0(x0,y0,z0) Is a point on the cylinder axis, Pi(xi,yi,zi) (i 1, 2.. n.) is the point cloud on the circular arc, n is the number of the point clouds, Qi(xi,yi,zi) Is PiA projected point on the cylindrical axis;
(1.6.3) calculating the mean value of all the projection points to obtain the central coordinate p of the steel coil:
Figure FDA0003193239150000023
(1.6.4) calculating the directional bounding box of the steel coil point cloud to obtain the steel coil width information;
(1.6.5), verifying the steel coil according to the z coordinate, wherein if the central coordinate z is smaller than the center z' of the bounding box, the central coordinate is the central coordinate of the effective steel coil, otherwise, the central coordinate is discarded.
CN202011186575.9A 2020-10-30 2020-10-30 Three-dimensional positioning method for surface target of truck in steel mill reservoir area Active CN112489123B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011186575.9A CN112489123B (en) 2020-10-30 2020-10-30 Three-dimensional positioning method for surface target of truck in steel mill reservoir area

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011186575.9A CN112489123B (en) 2020-10-30 2020-10-30 Three-dimensional positioning method for surface target of truck in steel mill reservoir area

Publications (2)

Publication Number Publication Date
CN112489123A CN112489123A (en) 2021-03-12
CN112489123B true CN112489123B (en) 2021-09-10

Family

ID=74927515

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011186575.9A Active CN112489123B (en) 2020-10-30 2020-10-30 Three-dimensional positioning method for surface target of truck in steel mill reservoir area

Country Status (1)

Country Link
CN (1) CN112489123B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115201778B (en) * 2022-09-09 2023-02-03 广州小鹏自动驾驶科技有限公司 Irregular obstacle detection method, vehicle and computer-readable storage medium

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106503347A (en) * 2016-10-25 2017-03-15 福州大学 One kind is based on AABB and the adaptive compound OBBs algorithms of OBB
CN107945178A (en) * 2017-12-19 2018-04-20 大昌建设集团有限公司 A kind of residual hole automatic identification of explosion and feature extracting method
CN108249307A (en) * 2018-01-17 2018-07-06 北京建筑大学 A kind of traverse measurement of goliath and feedback control system and method
CN110223308A (en) * 2019-04-15 2019-09-10 东南大学 A kind of stack position point cloud localization method increased based on edge detection and region
CN111023965A (en) * 2019-11-25 2020-04-17 上海大学 Steel coil identification and positioning method based on laser technology
CN111462275A (en) * 2019-01-22 2020-07-28 北京京东尚科信息技术有限公司 Map production method and device based on laser point cloud
CN111680542A (en) * 2020-04-17 2020-09-18 东南大学 Steel coil point cloud identification and classification method based on multi-scale feature extraction and Pointernet neural network

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106839985B (en) * 2017-03-22 2019-04-26 常熟理工学院 The automatic identification localization method of unmanned overhead traveling crane coil of strip crawl
KR102131592B1 (en) * 2018-10-25 2020-08-05 주식회사 인공지능연구원 Apparatus for Predicting 3D Original Formation
CN110097593A (en) * 2019-04-15 2019-08-06 上海海事大学 A method of identifying cylindrical surface from multi-line laser radar point cloud data
CN111027484B (en) * 2019-12-11 2023-04-28 中南大学 Tunnel steel arch identification method based on three-dimensional imaging

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106503347A (en) * 2016-10-25 2017-03-15 福州大学 One kind is based on AABB and the adaptive compound OBBs algorithms of OBB
CN107945178A (en) * 2017-12-19 2018-04-20 大昌建设集团有限公司 A kind of residual hole automatic identification of explosion and feature extracting method
CN108249307A (en) * 2018-01-17 2018-07-06 北京建筑大学 A kind of traverse measurement of goliath and feedback control system and method
CN111462275A (en) * 2019-01-22 2020-07-28 北京京东尚科信息技术有限公司 Map production method and device based on laser point cloud
CN110223308A (en) * 2019-04-15 2019-09-10 东南大学 A kind of stack position point cloud localization method increased based on edge detection and region
CN111023965A (en) * 2019-11-25 2020-04-17 上海大学 Steel coil identification and positioning method based on laser technology
CN111680542A (en) * 2020-04-17 2020-09-18 东南大学 Steel coil point cloud identification and classification method based on multi-scale feature extraction and Pointernet neural network

Also Published As

Publication number Publication date
CN112489123A (en) 2021-03-12

Similar Documents

Publication Publication Date Title
CN111680542B (en) Steel coil point cloud identification and classification method based on multi-scale feature extraction and Pointnet neural network
CN110223308B (en) Stacking site cloud positioning method based on edge detection and region growing
CN109000649B (en) Omni-directional mobile robot pose calibration method based on right-angle bend characteristics
CN107292310B (en) Visual positioning and automatic reading method for circular pointer type dial plate
CN112883878B (en) Automatic point cloud classification method under transformer substation scene based on three-dimensional grid
CN109785301B (en) Rail corrugation period evaluation method based on image processing
CN114170279A (en) Point cloud registration method based on laser scanning
CN108573511B (en) Point-distributed cooperative coding mark and identification and positioning method thereof
CN115035120B (en) Machine tool control method and system based on Internet of things
CN108009551B (en) Electric knife switch opening and closing position state identification method suitable for electric power robot
CN108846831B (en) Band steel surface defect classification method based on combination of statistical characteristics and image characteristics
CN112489123B (en) Three-dimensional positioning method for surface target of truck in steel mill reservoir area
CN112860952A (en) Method and system for positioning and repairing parts of inspection robot
CN115311629A (en) Abnormal bending precision monitoring system of bending machine
CN113989308B (en) Polygonal target segmentation method based on Hough transformation and template matching
CN114387595A (en) Automatic cargo identification method and device based on cargo splicing identification
CN113298820B (en) Method and system for stockpile warehouse
CN116843742B (en) Calculation method and system for stacking volume after point cloud registration for black coal loading vehicle
CN103871048B (en) Straight line primitive-based geometric hash method real-time positioning and matching method
CN102270306A (en) Information processing device and method, and program
CN109615603A (en) A kind of visual attention model of task based access control driving extracts the universal method of laser stripe
CN101727749B (en) Method and device for identifying abnormal vehicle speed data
CN111127638B (en) Method for realizing positioning and grabbing point of protruding mark position of workpiece by using three-dimensional template library
CN116310355A (en) Laser point cloud denoising and defect detection method for complex structural member
CN110533075B (en) Multi-model power line segmentation method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant