CN109101743B - Method for constructing high-precision road network model - Google Patents
Method for constructing high-precision road network model Download PDFInfo
- Publication number
- CN109101743B CN109101743B CN201810986942.XA CN201810986942A CN109101743B CN 109101743 B CN109101743 B CN 109101743B CN 201810986942 A CN201810986942 A CN 201810986942A CN 109101743 B CN109101743 B CN 109101743B
- Authority
- CN
- China
- Prior art keywords
- lane
- road
- road section
- linear
- point
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/18—Network design, e.g. design based on topological or interconnect aspects of utility systems, piping, heating ventilation air conditioning [HVAC] or cabling
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a construction method of a high-precision road network model, which comprises the steps of firstly defining the high-precision road network model, and then extracting road network data, including lane network layer extraction and road section network layer extraction; then, calculating the incidence relation of the road section network layer and the lane network layer; finally, constructing a high-precision road network model; the invention uses high precision road network model HDRNM (high definition road network model) to represent the road network model corresponding to HDmap, and the HDRNM is richer than road section level road network model from data content to data model.
Description
Technical Field
The invention belongs to the technical field of high-precision maps, relates to a road network model construction method, and particularly relates to a construction method of a High Definition Road Network Model (HDRNM) serving unmanned high-precision road network model.
Background
The HD Map is a Map specially used for unmanned driving, the application of a high-precision Map is more and more important for the unmanned driving, and a high-precision road network is the most important composition content in the high-precision Map. With the development of intelligent transportation and the development of heat in ADAS, HD map (High Definition map) has attracted great attention in both academic and industrial circles (documents 1 to 5). The HD map can provide refined map information to assist the intelligent vehicle in realizing high-precision positioning (documents 6 to 8), can solve the problem of sensor failure under specific conditions, makes up for the deficiency of environmental sensing equipment, and effectively reduces the difficulty of intelligent vehicle sensing (documents 9 to 11); according to the prior knowledge of the map and the dynamic traffic information, an optimal driving path and a reasonable driving strategy are given based on global path planning (documents 12 to 14), active safety of driving vehicles is effectively realized, and the complexity of vehicle driving is reduced (document 15). Therefore, the generation of the HD map becomes very important, and the HD map is currently in a great demand stage (document 16). Road network data is an expression of a real-world road model, and high definition road network is an important component of the HD map.
However, currently, studies on automatic generation of a high-precision road network are not many, which focus on both road extraction and link extraction.
The research on the generation of high-precision road networks mainly focuses on the extraction of road networks (documents 17 to 19), the generation of high-precision road segments (documents 20 and 21), the extraction of high-precision intersections (document 22), and the like, by using crowdsourcing or smart car equipment. In the research on the model of the high-precision road network, research is mainly focused on the format of the expression of the high-precision road network (document 23), the expression of the high-precision intersection (document 24), the road model (documents 25 and 26) and the like, but few people pay attention to the automatic generation of the topological correlation between the lane and the road section. Some studies have been made on automatic generation of automatic generation topological relations of a road network, and the main methods include merging intersections of different layers (document 27), using point correlation (document 28), and using a Hidden Markov Model (HMM) map matching (document 29). However, these studies are not based on topology extraction of a lane-level road network, and much manual methods are used for topology extraction of the lane-level road network.
HD map data is generally centimeter-level localization of approximation (document 30), and automatic generation of a high-precision road network is receiving attention from students more and more. The objects served by the high-precision map include not only the existing ADAS system but also an unmanned system, an auxiliary safety system, a vehicle road coordination system, and the like.
On the aspect of model expression of a high-precision road network, a plurality of relevant researches are carried out on a learner who has already learned lane extraction and model at present. And Gi-Poong et al uses piece wise polymers to simulate the lanes, so that the storage efficiency of the road network is improved. The lanes are expressed by an a third-order concrete similar to an approximated faster chochochochochochochochochoid spline, and the curve of the turn at the intersection is expressed by the cubic Catmull-Rom spline (document 17), both of which can be modeled quickly. The lane centreline was modelled using the Cubic Hermite spline (document 31), and a series of lane and road section modelling was performed using software adapted to the GIS database. The road network of the lane road is expressed three-dimensionally by using a B-spline curve (document 25), and the shape and accuracy of the three-dimensional road network are ensured. This type of study focuses on the geometric representation of the lane model. Tao et al, a description model of lanes of a high-precision road network is defined, and a road network is composed of lane arcs, lane attributes, intersections, intersection attributes and the like (document 23), so that the expression of the lane models in the high-precision road network is solved, but the expression of the information of the lanes in the high-precision road network is lacked, and the corresponding association relationship between the lanes and the road network is not expressed.
The content of the expression of the high-precision road network becomes finer and finer. In the united states, the Federal Highway Administration and National Highway Traffic and Safety Administration conceptually enrich detailed information on lanes of a high-precision road network, with lanes as road network contents (document 32). Further, the geometric shape and topological connection relationship of the lane are expressed, and the expression of the accuracy and content of the road network is more complete (document 4). Tao et al adds a virtual lane as content of expression at an intersection (document 24), solving the problem of lack of detailed information at the intersection. However, these studies do not provide an attribute expression of dynamic multidimensional, and have a certain limitation on the real-time property of unmanned driving.
Reference documents:
document 1. Tommamps.
Document 2. Depmapp.
Document 3.Nedevschi S, popescu V, danescu R, marita T, origin F. Accurate Ego-Vehicle Global Localization at interactions Through Alignment of Visual Data With Digital map. IEEE Transactions on Intelligent transport systems.2013;14 (2):673-87.
Document 4. Beta. D, toledo-Moreo R. Creating enhanced maps for land-level vehicle navigation. IEEE Transactions on Intelligent transfer systems.2010;11 (4):786-98.
Document 5.Rohani M, gingras D, gruyer D.A. Novel Approach for Improved temporal Positioning Using Cooperative Map Matching and Dynamic Base Station DGPS concept. IEEE Transactions on Intelligent transfer systems.2016;17 (1):230-9.
Document 6.Suganuma N, uozumi T, editors. Precision position estimation of autonomus vehicle based on map-matching. Intelligent Vehicles Symposium;2011.
document 7.Aeberhard M, rauch S, bahram M, tanzmeister G.Experience, results and Lessons Learned from Automated Driving on Germany's Highways. IEEE Integrated transport Systems Magazine.2015;7 (1):42-57.
Document 8.Toledo-Moreo R, betaille D, peyret F, laneurit J. Furing GNSS, dead-Reckinning, and Enhanced Maps for Road Vehicle Lane-Level navigation. IEEE Journal of Selected Topics in Signal processing.2009;3 (5):798-809.
Literature 9.Driankov d, saffizotti a. Fuzzy logic techniques for autonomous vehicle navigation; 2013.
document 10.Cao G, damerow F, flade B, helmling M, eggert J, editors. Camera to map alignment for access low-cost lane-level scene interpretation. Intelligent Transmission Systems (ITSC), 2016IEEE 19th International Conference on;2016 of the computer.
Document 11.Gruyer D, belaroussi R, revillound M.Accurate relative positioning from map data and rod marking detection; 2016.1-8p.
Document 12.Li H, nashashibi F, toulmine G, editors for an interactive vehicle by vulcanizing mono-camera, low-cost GPS and map data, international IEEE Conference on Intelligent transfer Systems;2011.
document 13.Tang B, khokhar S, gupta R, editors. Turn prediction at generated interaction. Intelligent vectors Symposium (IV), 2015IEEE;2015, IEEE.
Documents 14.Kim J, jo K, chu K, sun wood M.road-model-based and graph-structure-based scientific path-planning approach for autonomous vehicles, proceedings of the institute of Mechanical Engineers, part D: journal of Automobile engineering.2014;228 (8):909-28.
Document 15, lozano-Perez T.Autonomous robot vehicles, springer Science & Business Media;2012.
document 16, progress and thinking of a sheep Cheng Liu Guo Zhang left, high-precision road navigation map, chinese engineering science, 2018;20 (2):99-105.
Document 17.Guo C, kidono K, meguro J, kojima Y, ogawa M, naito T.A low-cost solution for automatic land-level map generation using comparative in-car sensors. IEEE Transactions on Intelligent transfer systems.2016;17 (8):2355-66.
Document 18. Matern n, schubert r, wanielik g, editors, high-acid vessel localization using digital maps and coherent images, intellectual vessel Symposium (IV), 2010IEEE;2010, IEEE.
Gwon G-P, hur W-S, kim S-W, seo S-W.Generation of a precision and effect land-level map for interactive vehicle systems IEEE Transactions on Vehicular technology 2017;66 (6):4517-33.
Gikas V, stratakos J.A novel genetic engineering method for acid and automated road/road genetic geological extraction based on the bearing diagram and actual bearing IEEE transactions on organic transfer system.2012; 13 (1):115-26.
Documents 21.M a ttyus G, wang S, fidler S, urtasun R, editors. Hd maps Fine-grained row segmentation by matching group and actual images, proceedings of the IEEE Conference on Computer Vision and Pattern Recognition;2016.
document 22, yang X, tang L, niu L, zhang X, li Q.Generation of road-based interaction maps from crown source big data.transfer Research Part C, emerging technologies.2018;89:168-87.
Document 23.Zhang T, arrigoni S, garozzo M, yang D-g, cheli F. A lane-level road network model with global connectivity. Transport Research Part C, emerging technologies.2016;71:32-50.
Document 24.Zhang T, yang D, li T, li K, lian X.an enhanced visual interaction model for a temporal interaction at interactions C; 19 (3):413-23.
Document 25.Jo K, lee M, kim C, sun wood M.construction process of a three-dimensional raw geometry map for automatic distillation. Proceedings of the organization of Mechanical Engineers, part D: journal of automatic engineering.2017;231 (10):1414-34.
Document 26.Chen A, raman A, farrell JA, editors, high-precision land-level road map building for position Location and Navigation Symposium (PLANS), 2010IEEE/ION; 2010.
Document 27.Karagiorgou S, pfiser D, skoutas D.A layered improvement for more robust generation of road network maps from mobile tracking data.ACM transformations on Spatial Algorithms and Systems (TSAS). 2017;3 (1):3.
Xie X, wong KB-Y, aghajan H, veelaert P, philips W.road network in reference through multiple track alignment Research Part C, emerging technologies.2016;72:93-108.
Document 29. Qiau J, wang R.automatic extraction of road networks from GPS channels. Photogrammetric engineering and remote sensing.2016;82 (8):593-604.
Document 30.Du J, barth MJ. Next-Generation automatic Location Systems: location at the Lane level. IEEE Transactions on Intelligent transfer systems.2008;9 (1):48-57.
Document 31.Chen A, ramanandan A, farrell JA, editors, high-precision land-level road map building for position Location and Navigation Symposium (PLANS); 2010.
document 32.Enhanced Digital Mapping Project Final Report, technical Report, united States Department of transfer, federal high way Administration and National high way Traffic and Safety Administration, http:// ww-nrd. Nhtsa. Dot. Gov/pdf/nrd12/CAMP/EDMAP%20Final 20report/Main%20 Report/FinalRept-111904. Pdf,189p. Last accessed Dec.8,2008.
Document 33. Qia J, wang R.road Map Imperence; 5 (8):130.
Document 34. Navigation electronic map framework data exchange format.
Disclosure of Invention
In order to solve the problems in the background and further enrich the fineness degree of a high-precision road network, only pay attention to a road section network layer and a lane network layer in the road network, and express the corresponding relation between the road section network layer and the lane network layer in the high-precision road network, the invention provides a high-precision road network model HDRNM. The model divides the HD road network into a road section network layer and a road network layer, and not only expresses the geometric, topological and attribute information of road sections and lanes in detail, but also defines the incidence relation between the road section network layer and the lane network layer. The invention defines the place where the attribute changes in the road section as the linear event point, takes the road section as the linear measurement reference, maps the linear event position in the road section to the linear event point of the lane formed in the lane through the relative position, and expresses the lane in sections at the linear event point of the lane.
The technical scheme adopted by the invention is as follows: a method for constructing a high-precision road network model is characterized by comprising the following steps:
step 1: defining a high-precision road network model;
the model is as follows:
wherein, in formula 1, W represents a road network, C is an intersection set, and R is a set of road segments; in formula 2, a road segment index set is represented by {1,2, \8230;, N }, r represents a road segment, and r represents 1 ，r 2 ，…，r N Respectively representing each road segment in the set; in equation 3, for a certain segment r, S r Is a shape point of the road section, SN r Is the starting point node of the road section, EN r Is a road section end node, Q r Is an attribute of the road section, R L Is the number of the road section connection, L S A corresponding set of lanes on the road segment; in equation 4, a set of lane indexes is represented by {1,2, \8230;, i }, where l represents a lane, and l represents a lane 1 ，l 2 ，…，l i Representing associated lanes under a certain road section; in equation 5, for a certain lane l, S l Is a shape point of a lane, SN l Is the starting node of the roadway, EN l Is a terminal node of a lane, Q l Is an attribute of the lane, L L Is the number of lane continuation; in equation 6, Q represents a dynamic attribute value, yes or no, t represents time, and Q represents an attribute value of an indication class of a lane or a link in equations 3 and 4, and this value is paired withThe corresponding value range is enumerated;
step 2: extracting road network data, including road section network layer extraction and lane network layer extraction;
and step 3: calculating the incidence relation between the road section network layer and the lane network layer;
the correlation between the road section r and the lane l corresponding to the road section in the formula 3 is:
C＝f(M) (7)
in formula 7, C is the correlation corresponding to the road section and the lane, M is the set of linear event points, and the position of the attribute change on a certain road section is represented by the linear event points;
the functional relation C = f (M) of the lane corresponding to the road section is defined as:
wherein i represents a sequence of lane morphology data sets perpendicular to the lane direction; j represents a sequence of linear segments along the direction of the road segment; a totalLaneShapefunction representing a set of lane morphology data parallel to a road segment under a current road segment; totalMNum represents the total number of linear time points M on the current road section; i is n Is a unit array; l denotes a lane, l 1,j Expressing an expression in the jth linear section of the ith lane parallel to the road under the road section; LS represents an abbreviation of the lane morphology dataset LSFunction; LS (least squares) i,j An expression of a lane shape data set LSfunction parallel to the road section at the ith under the road section in the jth linear section; x is the number of j Representing the value range of coordinates in the j-1 th to j-th linear segments along the direction of the road segment; x i Representing the coordinate value range of the ith lane form data set parallel to the road section in the road direction;
and 4, step 4: and constructing a high-precision road network model.
Compared with the prior art, the invention has the beneficial effects that: an HDRNM (high definition road network model) is provided, the model divides an HD road network into a road section network layer and a road network layer, and besides the detailed expression of the geometry, topology and attribute information of the road section and the lane, the model also defines the association relationship between the road section and the lane.
The invention defines the place where the attribute changes in the road section as the linear event point, and uses the road section as the linear measurement reference, maps the linear event position in the road section to the linear event point forming the lane in the lane through the relative position, and expresses the lane in sections through the linear event point of the lane, and expresses the position of the attribute change on a certain road section through the linear event point. The mapping relation fully considers the requirement of unmanned driving, enriches the fineness degree of the high-precision road network, meets the requirement of unmanned driving on the fineness of the high-precision road network, and can be applied to automatic checking of the production result of the high-precision lane road network. The HDRNM is richer than a road network model at a road section level from data content to a data model.
Drawings
Fig. 1 is a schematic block diagram of model construction according to an embodiment of the present invention.
Detailed Description
In order to facilitate the understanding and implementation of the present invention for those of ordinary skill in the art, the present invention is further described in detail with reference to the accompanying drawings and examples, it is to be understood that the embodiments described herein are merely illustrative and explanatory of the present invention and are not restrictive thereof.
Referring to fig. 1, the present invention provides a method for constructing a high-precision road network model, including the following steps:
step 1: defining a high-precision road network model;
the model is as follows:
wherein, in formula 1, W represents a road network, C is an intersection set, and R is a set of road segments; in formula 2, a road segment index set is represented by {1,2, \8230;, N }, r represents a road segment, and r represents 1 ，r 2 ，…，r N Respectively representing each road segment in the set; in equation 3, for a certain segment r, S r Is the shape point of the road section，SN r Is the starting point node of the road section, EN r Is a road section end point, Q r Is an attribute of a link (specific attribute, see national standard (document 34)), R L Is the number of the road section connection, L S A corresponding set of lanes on the road segment; in equation 4, a set of lane indexes is represented by {1,2, \8230;, i }, i represents a lane, and l represents a lane 1 ，l 1 ，…，l i Representing associated lanes under a certain road section; in equation 5, for a certain lane l, S l Is a shape point of a lane, SN l Is the starting node of the roadway, EN l Is a terminal node of a lane, Q l Is the attribute of the lane (including but not limited to the length of the lane, the width of the lane, the slope of the lane, the radius of curvature of the lane, etc., which may be expanded in content with the requirements of the unmanned application), L L Is the number of lane continuation; in formula 6, Q represents a dynamic attribute value, which indicates yes or no, t represents time, Q represents an attribute value of an indication class of a lane or a road segment in formulas 3 and 4, and a value range corresponding to the value is an enumeration type;
step 2: extracting road network data, including road section network layer extraction and lane network layer extraction;
extracting the road section network layer by adopting PCA Algorithm to extract the direction of the road section; principal Component Analysis (PCA) is a commonly used data analysis method (36) to describe the characteristic distribution of points. The purpose of PCA is to approximate n-dimensional raw data in k-dimension with as little loss to the data as possible after the approximation, that is, PCA finds unit vectors during the process of transforming the data from the original coordinate system to the new coordinate system so that the variance of the projection of the data in that direction can be maximized. The PCA algorithm calculates a feature matrix, which is a sample distribution that can represent most feature dimensions and is used for measuring the correlation between a point and its surrounding points. The feature vector is calculated by the feature matrix. The eigenvalues X1, X2 are obtained by calculating eigenvectors. The K = MAX { X1, X1}/{ X1+ X2} value represents the degree of linearity. Where K >0.9 represents that the clustering points are a straight line (document 33), therefore, the present embodiment determines whether the point sets in a certain point range belong to a linear relationship by the degree of linearity.
In this embodiment, a search radius searchR is first defined, and gaussian projection is performed on all coordinate points. And secondly, starting from any point, normalizing the point in the point search radius and carrying out PCA projection, and projecting the two-dimensional coordinate into a one-dimensional space. And calculating a linear degree point K value through the characteristic value. Thirdly, the embodiment screens out all point sets with K > 0.9. And merging the point sets with intersection larger than 0.9 to form a maximum linear point set, wherein the maximum linear point set respectively corresponds to a set of all lane central line points in each road section. And fourthly, carrying out PCA projection on the maximum linear point set to obtain the projection direction of the main direction of each road section, wherein the projection direction is the direction of the road section.
Lane network layer extraction, which is to establish a lane network based on multidirectional constraint; the concrete implementation process is that after the road section is extracted, the coordinate points are further classified and extracted, and different lane point sets in the road section are found. In this process, the embodiment uses the principal direction of PCA and the angle threshold σ as the constraint directions to realize lane extraction. The angle threshold σ represents a difference between the current coordinate point and a projection direction value of the next point of the coordinate point and the principal direction. From the results of the experiments, the angular threshold σ ranges between 0,30 °. An empirical value of 15 ° is generally selected as the link shape is closer to a straight line, and an angle threshold value of 30 ° is selected as the road shape is more curved.
The specific extraction process is as follows:
through the result of the last process, the point sequence sorted according to the direction and the angle of the main direction in the same road section are obtained. First, in this embodiment, each road segment is tracked according to the first point in the main direction, traversal is performed according to the principle of main direction priority, and a point in a range where an included angle between the current point and the traversal point is σ degrees is found and regarded as a point on the same lane. And repeating the loop tracking until all the points are traversed. And calculating the length of all lanes according to the direction of the road section, and finding out the linear measurement value of the linear event point on the road section.
And step 3: calculating the incidence relation between the road section network layer and the lane network layer;
the correlation between the road section r and the lane l corresponding to the road section in formula 3 is:
C＝f(M) (7)
in formula 7, C is the correlation corresponding to the road section and the lane, M is the set of linear event points, and the position of the attribute change on a certain road section is represented by the linear event points;
the functional relationship C = f (M) for the lane corresponding to the road segment is defined as:
wherein i represents a sequence of lane morphology data sets perpendicular to the lane direction; j represents the sequence of linear segments along the direction of the road segment; a totalLaneShapefunction representing a set of lane morphology data parallel to a road segment under a current road segment; totalMNum represents the total number of linear time points M on the current road section; i is n Is a unit array; l denotes a lane, l 1,j Expressing an expression in a jth linear section of an ith lane parallel to the road under the road section; LS represents an abbreviation of the lane morphology dataset LSFunction; LS (least squares) i,j The expression of the ith lane form data set LSfunction parallel to the road section under the road section in the jth linear section is represented; x is the number of j Representing the value range of coordinates in the j-1 th to j-th linear segments along the direction of the road segment; x i Representing the coordinate value range of the ith lane form data set parallel to the road section in the road direction;
in the embodiment, a place where an attribute change occurs in a road section is defined as a linear event point, the road section is used as a linear measurement reference, the linear event position in the road section is mapped to a lane linear coordinate system which uses a lane as the linear measurement reference through a relative position, the linear event point of the lane is formed in the lane linear coordinate system, and then the lane is expressed in a segmented manner through the linear event point of the lane;
wherein the piecewise expression implementation process is to map the road segment to the lane by using a linear reference system; the mapping method of the linear reference system from the road section to the lane comprises the following steps: in the embodiment, a linear reference system is selected according to the road section direction, linear event points on the lanes are mapped to the corresponding lanes, the lanes are used as objects of line measurement, and linear positions corresponding to the selected lane change attributes are used as bases for dividing the lanes on the same road section. And after mapping, according to the actual condition of the Chinese road, the error between the linear event point and the original linear metric value of M in the lane direction is less than 10 meters.
And 4, step 4: and constructing a high-precision road network model.
The invention uses high precision road network model HDRNM (high definition road network model) to represent the road network model corresponding to HDmap, and the HDRNM is richer than road section level road network models from data content to data model.
It should be understood that parts of the specification not set forth in detail are well within the prior art.
It should be understood that the above description of the preferred embodiments is given for clarity and not for any purpose of limitation, and that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (4)
1. A method for constructing a high-precision road network model is characterized by comprising the following steps:
step 1: defining a high-precision road network model;
the model is as follows:
wherein, in formula 1, W represents a road network, C is an intersection set, and R is a set of road segments; in formula 2, a road segment index set is represented by {1,2, \8230;, N }, a road segment is represented by r 1 ，r 2 ，…，r N Respectively representing each road segment in the set; in equation 3, for a certain segment r, S r Is a shape point of the road section, SN r Is the starting point node of the road section, EN r Is a road section end node, Q r Is an attribute of the road section, R L Is the number of the road section connection, L S A corresponding set of lanes on the road segment; in equation 4, a set of lane indexes is represented by 1,2, \8230;, i, representing lanes, 1 ，l 2 ，…，l i representing associated lanes under a certain road section; in equation 5, S for a certain lane l Is a shape point of a lane, SN l Is the starting node of the roadway, EN l Is a terminal node of a lane, Q l Is the attribute of the lane, L L Is the number of lane continuation; in formula 6, Q represents a dynamic attribute value, which indicates yes or no, t represents time, Q represents an attribute value of an indication class of a lane or a road segment in formulas 3 and 4, and a value range corresponding to the value is an enumeration type;
step 2: extracting road network data, including road section network layer extraction and lane network layer extraction;
and step 3: calculating the incidence relation between the road section network layer and the lane network layer;
the correlation between the road section r and the lane l corresponding to the road section in formula 3 is:
C＝f(M) (7)
in formula 7, C is the correlation corresponding to the road section and the lane, M is the set of linear event points, and the position of the attribute change on a certain road section is represented by the linear event points;
the functional relationship C = f (M) for the lane corresponding to the road segment is defined as:
wherein i represents a sequence of lane morphology data sets perpendicular to the lane direction; j represents the sequence of linear segments along the direction of the road segment; a totalLaneShapefunction representing a set of lane morphology data parallel to a road segment under a current road segment; totalMNum represents the total number of linear time points M on the current road section; i is n Is a unit array; l denotes a lane, l 1,j Representation in jth linear segment of i-th lane parallel to road under road segmentFormula (I); LS represents an abbreviation of the lane morphology dataset LSFunction; LS (least squares) i,j The expression of the ith lane form data set LSfunction parallel to the road section under the road section in the jth linear section is represented; x is the number of j Representing the value range of coordinates in the j-1 th to j-th linear segments along the direction of the road segment; x i Representing the coordinate value range of the ith lane form data set parallel to the road section in the road direction;
and 4, step 4: and constructing a high-precision road network model.
2. The construction method of the high-precision road network model according to claim 1, wherein the road section network layer extraction in the step 2 is to extract a track point set of a road section and a direction of the road section by PCA; the specific implementation process comprises the following substeps:
step 2.1.1: defining a search radius searchR, and carrying out Gaussian projection on all coordinate points;
step 2.1.2: starting from any point, normalizing the point in the point search radius and carrying out PCA projection, projecting a two-dimensional coordinate into a one-dimensional space, and calculating a linear degree point K value through a characteristic value;
step 2.1.3: screening out all point sets with K being more than 0.9; merging the point sets with intersection and larger than 0.9 to form a maximum linear point set, wherein the maximum linear point set respectively corresponds to a set of all lane center line points in each road section;
step 2.1.4: and performing PCA projection on the maximum linear point set to obtain the projection direction of the main direction of each road section, wherein the projection direction is the direction of the road section.
3. The method for constructing the high-precision road network model according to claim 1, wherein the lane network layer extraction in the step 2 is to establish a lane network based on multidirectional constraint; the specific implementation process comprises the following substeps:
step 2.2.1: the main direction and the angle threshold value sigma of the PCA are used as constraint directions to realize lane extraction;
step 2.2.2: tracking each road section according to a first point in the main direction, traversing according to the principle of priority of the main direction, and finding out a point in a range of the included angle sigma degrees between the current point and the traversal point as a point on the same lane;
step 2.2.3: repeating the loop tracking until all the points are traversed;
step 2.2.4: and calculating the length of all lanes according to the road direction, and finding out the linear metric of the linear event point on the road section.
4. The method for constructing the high-precision road network model according to any one of claims 1-3, characterized in that: in step 2, defining a place where the attribute changes in the road section as a linear event point, taking the road section as a linear measurement reference, mapping the linear event position in the road section to a lane linear coordinate system taking a lane as the linear measurement reference through a relative position, forming the linear event point of the lane in the lane linear coordinate system, and then performing segmented expression on the lane through the linear event point of the lane;
wherein the piecewise expression implementation process is to map the road segment to the lane by using a linear reference system; the mapping method of the linear reference system from the road section to the lane comprises the following steps: and selecting a linear reference system according to the direction of the road section, mapping the linear event points on the lane to the corresponding lane, taking the lane as a line measurement object, and taking the linear position corresponding to the selected lane change attribute as a basis for dividing the lane on the same road section.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810986942.XA CN109101743B (en) | 2018-08-28 | 2018-08-28 | Method for constructing high-precision road network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810986942.XA CN109101743B (en) | 2018-08-28 | 2018-08-28 | Method for constructing high-precision road network model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109101743A CN109101743A (en) | 2018-12-28 |
| CN109101743B true CN109101743B (en) | 2023-01-17 |
Family
ID=64851592
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810986942.XA Active CN109101743B (en) | 2018-08-28 | 2018-08-28 | Method for constructing high-precision road network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109101743B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109976332A (en) * | 2018-12-29 | 2019-07-05 | 惠州市德赛西威汽车电子股份有限公司 | One kind being used for unpiloted accurately graph model and autonomous navigation system |
| CN111383450B (en) * | 2018-12-29 | 2022-06-03 | 阿里巴巴集团控股有限公司 | Traffic network description method and device |
| CN109614966B (en) * | 2019-03-07 | 2019-07-26 | 纽劢科技(上海)有限公司 | It is a kind of based on information fusion Lidar sensor efficient road surface and curb detection method |
| CN110415314B (en) * | 2019-04-29 | 2020-04-03 | 当家移动绿色互联网技术集团有限公司 | Method and device for constructing road network at road intersection, storage medium and electronic equipment |
| CN110275883A (en) * | 2019-05-17 | 2019-09-24 | 浙江吉利控股集团有限公司 | A kind of high-precision map storage system and method |
| CN115100880B (en) * | 2022-06-17 | 2024-05-14 | 南京莱斯信息技术股份有限公司 | Bus signal priority control method for realizing balanced distribution of buses |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108010316A (en) * | 2017-11-15 | 2018-05-08 | 上海电科智能系统股份有限公司 | A kind of road traffic multisource data fusion processing method based on road net model |
| CN108009666A (en) * | 2016-10-28 | 2018-05-08 | 武汉大学 | The preferential optimal path computation method of level based on dynamic road network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9286793B2 (en) * | 2012-10-23 | 2016-03-15 | University Of Southern California | Traffic prediction using real-world transportation data |
-
2018
- 2018-08-28 CN CN201810986942.XA patent/CN109101743B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108009666A (en) * | 2016-10-28 | 2018-05-08 | 武汉大学 | The preferential optimal path computation method of level based on dynamic road network |
| CN108010316A (en) * | 2017-11-15 | 2018-05-08 | 上海电科智能系统股份有限公司 | A kind of road traffic multisource data fusion processing method based on road net model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109101743A (en) | 2018-12-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109101743B (en) | Method for constructing high-precision road network model | |
| CN109256028B (en) | Method for automatically generating unmanned high-precision road network | |
| US11175145B2 (en) | System and method for precision localization and mapping | |
| Yang et al. | Generating lane-based intersection maps from crowdsourcing big trace data | |
| CN108171131B (en) | Improved MeanShift-based method for extracting Lidar point cloud data road marking line | |
| CN112184736B (en) | Multi-plane extraction method based on European clustering | |
| CN110021072B (en) | Holographic mapping-oriented multi-platform point cloud intelligent processing method | |
| Mi et al. | Automated 3D road boundary extraction and vectorization using MLS point clouds | |
| Zhang et al. | A cognitively inspired system architecture for the Mengshi cognitive vehicle | |
| Lu et al. | Graph-embedded lane detection | |
| Gao et al. | Fine-grained off-road semantic segmentation and mapping via contrastive learning | |
| Liang et al. | NetTraj: A network-based vehicle trajectory prediction model with directional representation and spatiotemporal attention mechanisms | |
| Verma et al. | Vehicle detection, tracking and behavior analysis in urban driving environments using road context | |
| CN116071722A (en) | Lane geometric information extraction method, system, equipment and medium based on road section track | |
| Li et al. | Localization for intelligent vehicles in underground car parks based on semantic information | |
| Lu et al. | A lightweight real-time 3D LiDAR SLAM for autonomous vehicles in large-scale urban environment | |
| Xia et al. | DuARUS: Automatic Geo-object Change Detection with Street-view Imagery for Updating Road Database at Baidu Maps | |
| Chen et al. | Improving Autonomous Vehicle Mapping and Navigation in Work Zones Using Crowdsourcing Vehicle Trajectories | |
| Chiang et al. | Bending the curve of HD maps production for autonomous vehicle applications in Taiwan | |
| Kim et al. | A geodetic normal distribution map for long-term LiDAR localization on earth | |
| Pang et al. | FLAME: Feature-likelihood based mapping and localization for autonomous vehicles | |
| Yang et al. | An optimization-based selection approach of landing sites for swarm unmanned aerial vehicles in unknown environments | |
| Wang et al. | A 64‐Line Lidar‐Based Road Obstacle Sensing Algorithm for Intelligent Vehicles | |
| Kong et al. | UAV LiDAR data-based lane-level road network generation for urban scene HD Maps | |
| Kim et al. | Cloud Update of Geodetic Normal Distribution Map Based on Crowd‐Sourcing Detection against Road Environment Changes |
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 | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Wang Xin Inventor before: Zheng Ling Inventor before: Li Bijun Inventor before: Wang Xin |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230607 Address after: 430072 Hubei Province, Wuhan city Wuchang District of Wuhan University Luojiashan Patentee after: WUHAN University Address before: 430223 7 floors, Block B, R&D Building No. 1, Huaengineering Science Park, Donghu New Technology Development Zone, Wuhan City, Hubei Province Patentee before: WUHAN ZHONGXIANG TECHNOLOGY Co.,Ltd. |