Road network map construction method for vehicle steering

Abstract

The invention relates to the technical field of traffic management, and particularly discloses a road network map construction method for vehicle steering, and the method comprises the following specific steps: 1, formulating a multi-level spatial topology construction rule of a steering-oriented two-dimensional road network; 2, carrying out standardization processing on the road network data, and constructing a road network information model based on a spatial topology construction rule; 3, constructing a road network information retrieval relation based on the road network information model under a spatial topology construction rule; 4, defining a vehicle positioning expression model, and providing a conversion model of the real position and the positioning expression of the track; and step 5, realizing next-step driving direction prediction of the track based on the track positioning expression model and the road network information model. The problems that in the prior art, road network connectivity retrieval is difficult, vehicle road section selection tendency analysis is inaccurate, and fusion of multiple kinds of key information is insufficient are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic management, and specifically provides a method for constructing a road network atlas for vehicle steering. Background Art

[0002] In the field of traffic, vehicle trajectories and road network information have always been important research data. Most studies on traffic flow statistics and vehicle prediction, although constantly strengthening the consideration of factors affecting vehicles in the road network structure, still have the following defects:

[0003] First, the topological association information of the road structure has not been fully utilized. In the traditional calculation of road network connectivity, intersections are mostly used as nodes and road networks as edges for connection, ignoring other restrictive information in complex urban road networks, resulting in the inability to quickly and accurately determine the feasible paths of vehicles.

[0004] Second, there is a lack of effective analysis of vehicle section selection tendencies, making it difficult to accurately predict the turning probability of vehicles at intersections. This makes the vehicle trajectory prediction results inaccurate and unable to meet the needs of actual traffic management and planning.

[0005] In addition, existing methods often fail to fully integrate the experience of traffic management experts, the spatial topological information of road structures, and the multi-level link relationships of road networks, resulting in a significant reduction in the reliability and practicality of prediction results. Therefore, a method for constructing a road network atlas for vehicle steering is provided. Summary of the Invention

[0006] The purpose of the present invention is to address the deficiencies of the prior art and provide a method for constructing a road network atlas for vehicle steering to solve the problems raised in the above background art.

[0007] To achieve the above purpose, the present invention provides the following technical solution: A method for constructing a road network atlas for vehicle steering, the specific steps are as follows:

[0008] Step 1: Formulate multi-level spatial topological construction rules for a two-dimensional road network oriented to turning;

[0009] Step 2: Standardize the road network data and construct a road network information model based on the spatial topological construction rules;

[0010] Step 3: Under the spatial topological construction rules, construct a road network information retrieval relationship based on the road network information model;

[0011] Step 4: Define a vehicle positioning expression model and provide a conversion model for the true position and positioning expression of the trajectory;

[0012] Step 5: Based on the trajectory positioning expression model and the road network information model, predict the next driving direction of the trajectory.

[0013] As a preferred technical solution of the present invention, the specific steps for formulating the multi-level spatial topology construction rules for the two-dimensional road network oriented to turning in step 1 are as follows:

[0014] Step 1.1: According to the entity characteristics of the spatial road network, the spatial road network is split into intersection entities, road section entities, and lane entities; the intersection entity represents the turning intersection where the vehicle driving direction changes in the spatial road network, the road section entity represents the road connecting each turning intersection, and the lane entity is an important element constituting the road;

[0015] Step 1.2: Construct the spatial and attribute division rules of "road section - lane", match the spatial positions of the lane and the road section. Both the lane and the road have the expression of spatial planar information, and the relationship between the road section and the lane is "containment", that is, the road section contains the lane both in terms of spatial and attribute relationships;

[0016] Step 1.3: Construct the graph expression rules of the spatial road network of "road section - road section", without paying attention to the relationship from the intersection to the road section, only save the association relationship of the road section, and express it with "passable";

[0017] Step 1.4: Construct the spatial road network graph expression rules of "lane - road section". The relationship between the lane and the road section should be unidirectional connection, and only pay attention to the relationship from the lane to the road section. The relationship between the lane and the road section is many-to-many, without paying attention to the relationship between the lane and the intersection, and express it with "arrive at";

[0018] Step 1.5: Construct the attribute information expression rules. The intersection is represented by a name as a relationship attribute, and the name is named in the hierarchical relationship of "district - road - intersection - number of directions", and uniqueness is ensured; the road section only contains the name attribute, and the name is named in the hierarchical relationship of "district - road - road section number", and uniqueness is ensured; the lane contains a weight attribute, and this weight attribute records the weight from the lane to the road section.

[0019] As a preferred technical solution of the present invention, the specific steps for standardizing the road network data and constructing the road network information model based on the spatial topology construction rules in step 2 are as follows:

[0020] Step 2.1: Use the spatial vector road network data to construct a road network information model that only contains road sections and intersections. The specific data processing method adopts the basic attribute and spatial matching algorithms;

[0021] Step 2.2: Obtain the traffic turning sign information of different roads through block image data, remote sensing image data, and measured image data, and record the heading relationship between the traffic turning sign information and the road section;

[0022] Step 2.3: Obtain the number of lanes and the road section width of the road section according to the lane attributes in the spatial vector road network data. Calculate the lane center line and lane width in an average manner based on the road section width and the number of lanes, and correspond them to the traffic turning sign information in Step 2.2 respectively;

[0023] Step 2.4: Construct combined lanes. Construct a buffer based on the lane center line and width, calculate adjacent lanes in the same road section, and compare whether their traffic turning signs are consistent. If they are consistent, merge the two lanes, generate a new combined lane, and update the lane center line and width. This combined lane serves as the final lane information.

[0024] As a preferred technical solution of the present invention, in Step 3, under the spatial topology construction rule, based on the road network information model, the specific steps for constructing the road network information retrieval relationship are as follows;

[0025] Step 3.1: Use a reference directed graph structure to represent the two-way relationship of "road section - road section". The ids of the two edges between two reversible road section nodes must be the same. This id corresponds to the intersection attribute table, and the id naming refers to "the id of road section 1 - the id of road section 2 - the number is 1 or 2". When the number is 1, it is expressed as "road section 1 -> road section 2". If the number is 2, it is expressed as "road section 1 -> road section 2" and "road section 2 -> road section 1";

[0026] Step 3.2: Construct the intersection attribute table. Generate the transfer weight information of "road section - road section" based on historical data. The weight from road section A to road section B = the number of all vehicles from road section A to road section B / the number of all vehicles passing through road section A;

[0027] Step 3.3: Use a tree structure to represent the association relationship of "road section - lane". The road section serves as the root node, the lane serves as the child node, and the associated road sections of the root node road section serve as the final leaf nodes of the tree structure. Each road section node corresponds to a three-layer tree structure;

[0028] Step 3.4: If the retrieved information does not contain lane information, retrieve based on the directed graph in Step 3.1 and the intersection attribute table in Step 3.2. If it contains lane information, retrieve based on the tree structure in Step 3.3.

[0029] As a preferred technical solution of the present invention, in Step 4, define the vehicle positioning expression model and provide the specific steps for the conversion model between the true position of the trajectory and the positioning expression as follows;

[0030] Step 4.1: The vehicle positioning is expressed as P(x, y). Based on the spatial vector road network data in Step 2.1, using spatial position matching, the road section where the vehicle is located can be calculated, and at the same time, the corresponding lane information of this road section can be obtained;

[0031] Step 4.2: Use P(x,y) to match with the lane spatial position in Step 2.4 to obtain the lane where the vehicle is located, and finally represent the current vehicle position as "section - lane".

[0032] As a preferred technical solution of the present invention, the specific steps for predicting the next driving direction of the trajectory based on the trajectory positioning expression model and the road network information model in Step 5 are as follows:

[0033] Step 5.1: Use the section - lane information of the vehicle position obtained in Step 4. Based on the tree structure in Step 3.3, the set of sections accessible to the vehicle can be pointed to;

[0034] Step 5.2: Construct a unique identifier for the possible intersection attribute table;

[0035] Step 5.3: Use the unique identifier in Step 5.2 and find the intersection identifier with the largest weight from the intersection attribute table generated in Step 3.2 to finally obtain the section the vehicle is going to and the corresponding turning identifier information.

[0036] Compared with the prior art, the beneficial effects of the present invention are:

[0037] Through the innovative double - layer road network structure, the integration of real - time knowledge graphs, and the fusion analysis of multi - dimensional information, the present invention not only improves the accuracy and efficiency of vehicle path prediction, can provide more continuous semantic results for vehicle tracking, and can improve the accuracy of traffic condition and traffic flow prediction according to the association between turning results and sections, making a positive contribution to the development of intelligent transportation systems.

[0038] The present invention solves the problems of difficult retrieval of road network connectivity, inaccurate analysis of vehicle section selection tendency, and insufficient fusion of multiple key information in the prior art.

[0039] The present invention proposes a spatial road network topology construction rule, which is different from the traditional road network graph structures of "intersection - entity - section - edge relationship" and "both section and intersection are entities, and connectivity is the relationship". This rule consciously removes the intersection entity information, takes the section as the entity, reduces the retrieval quantity, and effectively retains the intersection information through the attribute table; adds lane entities, constructs a multi - layer graph relationship, and enriches the spatial road network information;

[0040] The present invention enriches the road network information model based on multi - source data, and samples the number of lanes based on spatial topological relationships and attribute consistency, reducing the information storage and retrieval pressure. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flowchart of the implementation process of the method for constructing a road network information model for vehicle turning in the present invention;

[0042] Figure 2 It is a flow chart for constructing multi-level spatial topology rules of a two-dimensional road network oriented to steering in the present invention;

[0043] Figure 3 It is to construct a road network information retrieval relationship graph based on the road network model in the present invention. Specific embodiments

[0044] The following elaborates on the preferred embodiments of the present invention in conjunction with the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.

[0045] Embodiment: Please refer to Figures 1-3 , the present invention provides a technical solution: A method for constructing a road network atlas for vehicle steering, the specific steps are as follows:

[0046] Step 1: Formulate multi-level spatial topology construction rules for a two-dimensional road network oriented to steering;

[0047] Step 1.1: According to the entity characteristics of the spatial road network, split the spatial road network into intersection entities, road section entities, and lane entities; the intersection entity is represented as the turning point where the vehicle driving direction changes in the spatial road network, the road section entity is expressed as the road connecting each turning point, and the lane entity is an important element that makes up the road;

[0048] Step 1.2: Construct the spatial and attribute division rules of "road section - lane", match the spatial positions of the lane and the road section, both the lane and the road should have the expression of spatial planar information, and the relationship between the road section and the lane should be "containment", that is, the road section should contain the lane both spatially and in terms of attribute relationship;

[0049] Step 1.3: Construct the graph expression rules of the spatial road network of "road section - road section", without paying attention to the relationship from the intersection to the road section, only save the association relationship of the road section, expressed by "passable", such as "road section A is passable to road section B", and the intersection C constructs an attribute correspondence table separately to record the relationship of "road section - road section";

[0050] Step 1.4: Construct the graph expression rules of the spatial road network of "lane - road section", the relationship between the lane and the road section should be unidirectionally connected, and only pay attention to the relationship from the lane to the road section, and the relationship between the lane and the road section should be many-to-many, without paying attention to the relationship between the lane and the intersection, expressed by "arrive", such as "lane L can arrive at road section R";

[0051] Step 1.5: Construct the expression rules for attribute information. An intersection is represented by a name as a relationship attribute, and the name is named in the hierarchical relationship of "district - road - intersection - number of directions", and uniqueness is ensured; a road section only contains a name attribute, and the name is named in the hierarchical relationship of "district - road - road section number", and uniqueness is ensured; a lane should contain a weight attribute, and this weight attribute records the weight from the lane to the road section.

[0052] Step 1 of the present invention proposes the rules for constructing the spatial road network topology. Different from the traditional road network graph structures of "intersection entity - road section edge relationship" and "both road sections and intersections are entities, and connectivity is the relationship", this rule consciously removes the intersection entity information, takes the road section as the entity, reduces the retrieval quantity, and effectively retains the intersection information through the attribute table; adds lane entities, constructs a multi - layer graph relationship, and enriches the spatial road network information.

[0053] Step 2: Standardize the road network data and construct a road network information model based on the spatial topology construction rules.

[0054] Step 2.1: Use the spatial vector road network data to construct a road network information model that only contains road sections and intersections. The specific data processing method uses basic attribute and spatial matching algorithms.

[0055] Step 2.2: Obtain the traffic turning sign information of different roads through block image data, remote sensing image data, and measured image data, and record the heading relationship between the traffic turning sign information and the road section, such as "taking the heading direction as due north, taking the left - most as 0, and increasing towards the right".

[0056] Step 2.3: Obtain the number of lanes and the width of the road section according to the lane attribute in the road section of the spatial vector road network data. Calculate the lane center line and lane width in an average manner based on the road section width and the number of lanes, and correspond them to the traffic turning sign information in Step 2.2 respectively.

[0057] Step 2.4: Construct combined lanes. Construct a buffer according to the lane center line and width, calculate adjacent lanes in the same road section, and compare whether their traffic turning signs are consistent. If they are consistent, merge the two lanes, generate a new combined lane, and update the lane center line and width. This combined lane is used as the final lane information.

[0058] Step 2 of the present invention enriches the road network information model based on multi - source data, and samples the number of lanes based on the spatial topology relationship and attribute consistency, reducing the information storage and retrieval pressure.

[0059] Step 3: Under the spatial topology construction rules, construct the road network information retrieval relationship based on the road network information model.

[0060] Step 3.1: Use a reference directed graph structure to represent the two-way relationship of "road section - road section". The ids of the two edges between two reversible road section nodes must be the same. This id corresponds to the intersection attribute table. The id naming refers to "id of road section 1 - id of road section 2 - number (1 or 2)". When the number is 1, it represents "road section 1 -> road section 2". If the number is 2, it represents "road section 1 -> road section 2" and "road section 2 -> road section 1".

[0061] Step 3.2: Construction of the intersection attribute table. Generate the transfer weight information of "road section - road section" based on historical data. The weight from road section A to road section B = the number of all vehicles from road section A to road section B / the number of all vehicles passing through road section A.

[0062] Step 3.3: Use a tree structure to represent the association relationship of "road section - lane". The road section is the root node, and the lanes are the child nodes. The associated road sections of the root node road section are the final leaf nodes of the tree structure. Each road section node corresponds to a three-layer tree structure.

[0063] In step 3.4, if the retrieved information does not contain lane information, retrieve based on the directed graph in step 3.1 and the intersection attribute table in step 3.2. If it contains lane information, retrieve based on the tree structure in step 3.3.

[0064] Step 3 of the present invention constructs a road network information retrieval relationship based on multiple attribute relationships to accelerate the lane retrieval speed and recognition.

[0065] Step 4: Define a vehicle positioning expression model and provide a conversion model between the true position of the trajectory and the positioning expression.

[0066] Step 4.1: The vehicle positioning is expressed as P(x, y). Based on the spatial vector road network data in step 2.1, the road section where the vehicle is located can be calculated using spatial position matching, and at the same time, the lane information corresponding to this road section can be obtained.

[0067] Step 4.2: Use P(x, y) to match with the lane spatial position in step 2.4 to obtain the lane where the vehicle is located. Finally, the current vehicle positioning is expressed as "road section - lane".

[0068] Step 4 of the present invention combines the spatial position values of the vehicle with the vector road network data to quickly obtain the lane information of the vehicle.

[0069] Step 5: Predict the next driving direction of the trajectory based on the trajectory positioning expression model and the road network information model.

[0070] Step 5.1: Using the road-segment and lane information where the vehicle is located obtained in Step 4, based on the tree structure in Step 3.3, the set of road segments accessible to the vehicle can be pointed to. For example, if the vehicle's location is "Road Segment A - Lane 3", the corresponding result is {Road Segment C, Road Segment D};

[0071] Step 5.2: Construct unique identifiers for the possible intersection attribute tables, such as {"Road Segment A - Road Segment C - 1", "Road Segment A - Road Segment C - 2", "Road Segment A - Road Segment D - 1", "Road Segment A - Road Segment D - 2"};

[0072] Step 5.3: Using the unique identifiers in Step 5.2 and finding the intersection identifier with the largest weight from the intersection attribute table generated in Step 3.2, the road segment the vehicle is going to and the corresponding turning identifier information are finally obtained.

[0073] Step 5 of the present invention can uniquely calibrate the road segment the vehicle is going to and the turning identifier information.

[0074] The above embodiments only express the implementation manners of the present invention, and their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several variations and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. A method for constructing a road network map for vehicle steering, characterized in that: The specific steps are as follows: Step 1: Formulate a multi-level spatial topology construction rule for the turning-oriented two-dimensional road network; Step 2: Standardize the road network data and build a road network information model based on spatial topology construction rules; Step 3: Under the spatial topology construction rules, based on the road network information model, construct the road network information retrieval relationship; Step 4: Define the vehicle positioning expression model and provide a conversion model between the actual position of the trajectory and the positioning expression; Step 5: Predict the next driving direction of the trajectory based on the trajectory positioning expression model and the road network information model.

2. The method for constructing a road network map for vehicle steering according to claim 1, characterized in that: The specific steps for formulating the multi-level spatial topology construction rules for the turn-oriented two-dimensional road network in step 1 are as follows: Step 1.1: According to the entity characteristics of the spatial road network, the spatial road network is divided into intersection entities, road segment entities and lane entities; the intersection entity represents the turning point where the vehicle driving direction changes in the spatial road network, the road segment entity represents the road connecting each turning point, and the lane entity is an important element of the road; Step 1.2: Construct the spatial and attribute division rules of "road section-lane", match the spatial positions of lanes and road sections, and both lanes and roads have spatial surface information expression, and the relationship between road sections and lanes is "included", and road sections contain lanes in terms of spatial and attribute relationships; Step 1.3: Construct the graph expression rules of the spatial road network of "road section-road section", ignoring the relationship between intersections and road sections, and only save the association relationship between road sections, expressed as "passable"; Step 1.4: Construct the spatial road network expression rules of "lane-road section". The relationship between lanes and road sections should be one-way connection, and only the relationship from lanes to road sections is concerned. The relationship between lanes and road sections is many-to-many, and the relationship between lanes and intersections is not concerned, which is expressed by "arrival"; Step 1.5: Construct attribute information expression rules. The intersection is represented by a name as a relational attribute, and the name is named in the form of a hierarchical relationship of "district-road-intersection-number of directions" and ensures uniqueness; the section only contains the name attribute, and the name is named in the form of a hierarchical relationship of "district-road-section number" and ensures uniqueness; the lane contains a weight attribute, which records the weight of the lane to the section.

3. The method for constructing a road network map for vehicle steering according to claim 2, characterized in that: The specific steps of standardizing the road network data in step 2 and constructing the road network information model based on the spatial topology construction rules are as follows: Step 2.1: Use the spatial vector road network data to construct a road network information model that only includes road sections and intersections. The specific data processing method uses basic attributes and spatial matching algorithms; Step 2.2: Obtain traffic turn sign information of different roads through block image data, remote sensing image data, and measured image data, and record the relationship between the traffic turn sign information and the heading of the road section; Step 2.3: Obtain the number of lanes and the width of the road section according to the lane attributes in the road section of the spatial vector road network data, and calculate the lane centerline and lane width in an average manner according to the road section width and the number of lanes, which correspond to the traffic turn sign information in step 2.2 respectively; Step 2.4: Construct a combined lane, build a buffer zone based on the lane centerline and width, calculate adjacent vehicle sections in the same section, and compare whether their traffic turn signs are consistent. If they are consistent, merge the two lanes to generate a new combined lane and update the lane centerline and width. The combined lane is used as the final lane information.

4. The method for constructing a road network map for vehicle steering according to claim 3, characterized in that: In step 3, under the spatial topology construction rules, based on the road network information model, the specific steps of constructing the road network information retrieval relationship are as follows; Step 3.1: Use the reference directed graph structure to represent the bidirectional relationship of "road segment-road segment". The ids of the two edges between two reversible road segment nodes must be consistent. The id corresponds to the intersection attribute table. The id naming refers to "road segment 1 id-road segment 2 id-number 1 or 2". When the number is 1, it is represented as "road segment 1->road segment 2". If the number is 2, it is represented as "road segment 1->road segment 2" and "road segment 2->road segment 1"; Step 3.2: Construct the intersection attribute table, generate the "segment-segment" transfer weight information based on historical data, and the weight from segment A to segment B = the number of all vehicles from segment A to segment B / all vehicles passing through segment A; Step 3.3: Use a tree structure to represent the "road segment-lane" association relationship, with the road segment as the root node, the lane as the child node, and the associated road segment of the root node as the final leaf node of the tree structure. Each road segment node corresponds to a three-layer tree structure; Step 3.4: If the retrieval information does not contain lane information, the retrieval is performed based on the directed graph in step 3.1 and the intersection attribute table in step 3.

2. If the retrieval information contains lane information, the retrieval is performed based on the tree structure in step 3.

3.

5. The method for constructing a road network map for vehicle steering according to claim 4, characterized in that: The specific steps of defining the vehicle positioning expression model in step 4 and providing the conversion model between the real position of the trajectory and the positioning expression are as follows; Step 4.1: The vehicle positioning is expressed as P(x, y). Based on the spatial vector road network data in step 2.1, the road section where the vehicle is located can be calculated by using spatial position matching, and the lane information corresponding to the road section can be obtained at the same time; Step 4.2: Use P(x, y) to match the lane spatial position in step 2.4 to obtain the lane the vehicle is in, and finally represent the current location of the vehicle as "road section-lane".

6. The method for constructing a road network map for vehicle steering according to claim 5, characterized in that: The specific steps for predicting the next driving direction of the trajectory based on the trajectory positioning expression model and the road network information model in step 5 are as follows: Step 5.1: Using the road segment and lane information of the vehicle obtained in step 4, the tree structure in step 3.3 can point to the road segment set accessible to the vehicle; Step 5.2: Construct the unique mark of possible intersection attribute table; Step 5.3: Use the unique sign in step 5.2 and find the intersection sign with the largest weight from the intersection attribute table generated in step 3.2, and finally obtain the road section to which the vehicle is heading and the corresponding turn sign information.

Images