CN116994086A

CN116994086A - Road model generation method, device, equipment and medium

Info

Publication number: CN116994086A
Application number: CN202310780703.XA
Authority: CN
Inventors: 刘宇辰; 朱灿
Original assignee: DeepRoute AI Ltd
Current assignee: DeepRoute AI Ltd
Priority date: 2023-06-28
Filing date: 2023-06-28
Publication date: 2023-11-03

Abstract

The invention discloses a road model generation method, a device, equipment and a medium, wherein the method comprises the steps of classifying high-precision map data to generate a plurality of road sub-modules; and fusing the plurality of road sub-modules to generate a road model. The road model is built based on the high-precision map data, so that the simulated road model is closer to the real situation; by classifying the high-precision map data, a plurality of sub-modules are respectively generated and then fused, so that the operation amount of each sub-module is reduced, and the modeling efficiency is improved.

Description

Road model generation method, device, equipment and medium

Technical Field

The present invention relates to the field of automatic driving technologies, and in particular, to a method, an apparatus, a device, and a medium for generating a road model.

Background

Autopilot simulation is the primary method of virtual roading, where scene realism is important, which can help to understand and evaluate the performance and behavior of an autopilot vehicle. But it is difficult to restore a real world scene in the virtual world due to resource limitations.

Currently, there are two general approaches to simulating road models, the other is to use some custom roads and digital assets, such as road models generated according to general protocols and building, environmental models originally used for games, etc. Another is digital twinning, which creates a real scene. The modeling of the road is relatively simple, but the scene reality cannot be guaranteed, and the real road condition cannot be covered due to the lack of support of real data, so that the simulated road has large difference from the real world, and if abnormality occurs in the simulation test, the problems of algorithm abnormality or road self cannot be distinguished, so that the simulation test is not facilitated, and the accuracy of the test result cannot be guaranteed. The simulation algorithm is improved by improving the simulation reality to a certain extent, but the technology has high operation difficulty, needs a large amount of resources for completely realizing digital reconstruction of scenes, has high cost and is not suitable for being promoted widely.

Disclosure of Invention

In order to solve at least one technical problem, the invention provides a road model generation method, a device, equipment and a medium, which can enable a simulated road model to be closer to a real situation and improve modeling efficiency.

In a first aspect, the present invention provides a road model generating method, the method including:

classifying the high-precision map data to generate a plurality of sub-modules;

and fusing the plurality of road sub-modules to generate a road model.

In one possible implementation, the road sub-modules include a road trunk module, a road traffic marking module, and a road edge module.

In one possible implementation, the generating the road trunk module includes:

acquiring road main road information from the high-precision map data, wherein the road main road information comprises road boundary information and a communication relationship between each road;

discretizing the road trunk information to generate a plurality of polygonal areas;

and generating a road trunk module according to the polygonal area.

In one possible implementation manner, the generating a road trunk module according to the polygonal area includes:

dividing the polygonal area by using a triangulation algorithm to generate a road trunk module; wherein the triangulation algorithm comprises an ear clipping algorithm.

In one possible implementation manner, after the dividing the polygonal area by using the triangulation algorithm, the method further includes:

Generating texture coordinates of regional vertexes based on the divided polygonal regions;

determining material information of a road main road according to the texture coordinates; the texture information comprises texture information;

rendering is carried out according to the material information of the road trunk.

In one possible implementation, generating the road traffic marking module includes:

acquiring point cloud data and reflection intensity information of each point in the high-precision map data;

extracting road traffic marking information in a high-precision map according to the point cloud data and the reflection intensity information of each point;

and generating the road traffic marking module according to the road traffic marking information.

In one possible implementation manner, the generating the path edge module includes:

extracting road edge information of different road sections from the high-precision map data, wherein the road edge information comprises length, width and height information of the road edge;

determining a road edge region according to the road edge information, and discretizing the road edge region to generate a plurality of polygonal regions;

and generating the road edge module according to the polygonal area.

In a second aspect, the present invention also provides a road model generating device, which includes:

The road sub-module generating unit is used for classifying the high-precision map data to generate a plurality of road sub-modules;

and the road sub-module fusion unit is used for fusing the plurality of road sub-modules to generate a road model.

In a third aspect, the present invention also provides an electronic device, including: a processor, a transmitting means, an input means, an output means and a memory for storing computer program code comprising computer instructions which, when executed by the processor, cause the electronic device to perform the method as described in the first aspect and any one of its possible implementation manners.

In a fourth aspect, the present invention also provides a computer readable storage medium having stored therein a computer program comprising program instructions which, when executed by a processor of an electronic device, cause the processor to perform a method as in the first aspect and any one of the possible implementations thereof.

Compared with the prior art, the invention has the beneficial effects that:

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

In order to more clearly describe the embodiments of the present invention or the technical solutions in the background art, the following description will describe the drawings that are required to be used in the embodiments of the present invention or the background art.

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the technical aspects of the disclosure.

FIG. 1 is a flowchart of a road model generating method according to an embodiment of the present invention;

FIG. 2 is a diagram illustrating the type of high-precision map data provided by an embodiment of the present invention;

FIG. 3 is a flowchart illustrating a sub-step of the step S10 according to an embodiment of the present invention;

FIG. 4 is a visual effect of high-precision map semantic information according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of a result of triangulating a polygonal area of a road trunk according to an embodiment of the present invention;

FIG. 6 is a schematic illustration of the rendered effect of FIG. 5;

FIG. 7 is a flowchart illustrating a sub-step of the step S10 according to another embodiment of the present invention;

FIG. 8 is a flowchart illustrating a sub-step of step S10 according to another embodiment of the present invention;

fig. 9 is a schematic structural diagram of a road model generating device according to an embodiment of the present invention;

fig. 10 is a schematic hardware structure of an electronic device according to an embodiment of the present invention.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The terms first, second and the like in the description and in the claims and in the above-described figures are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus.

The term "and/or" is herein merely an association relationship describing an associated object, meaning that there may be three relationships, e.g., a and/or B, may represent: a exists alone, A and B exist together, and B exists alone. In addition, the term "at least one" herein means any one of a plurality or any combination of at least two of a plurality, for example, including at least one of A, B, C, and may mean including any one or more elements selected from the group consisting of A, B and C.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the invention. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.

Furthermore, in the following detailed description, numerous specific details are set forth in order to provide a better illustration of the invention. It will be understood by those skilled in the art that the present invention may be practiced without some of these specific details. In some instances, well known methods, procedures, components, and circuits have not been described in detail so as not to obscure the present invention.

In the process of automatic driving simulation, the reality of a scene is very important, and the scene can help to understand and evaluate the performance and behavior of an automatic driving vehicle. Due to resource limitations, some custom roads and digital assets are currently only available, or digital twinning techniques are relied upon to restore the morphology of the real scene as much as possible. However, when the road model is customized, the scene reality will be different and the real road situation cannot be covered, so the result of the automatic driving simulation test will be affected, and the simulation result will be different from the real world situation. Because the reality of the scene cannot be ensured, whether the driving abnormality is caused by the problem of the algorithm or the problem of the scene cannot be distinguished in the simulation process, and the training and iteration of the algorithm are not facilitated. Although the digital twin technology aims at improving simulation reality, the technology is still in a theoretical exploration stage, and a large amount of resources are required to be input for completely and truly reconstructing a scene, so that the cost is quite high. In order to solve the above-described problems, the present embodiment aims to provide a road model generation method that generates a plurality of road sub-modules by performing classification processing on high-precision map data; and fusing the plurality of road sub-modules to finally generate a road model. Therefore, the simulated road model is closer to the real situation, and the modeling efficiency is improved.

In order to facilitate the understanding of the specific embodiments of the invention by those skilled in the art, the relevant terms to which the invention relates are first explained:

semantic information: meaning the semantic meaning of the elements, such as objects, terrain, landmarks, and their location and interrelationships in the scene, as well as all information that interacts with the autonomous vehicle. The meaning and effect of this information is of great importance for autonomous navigation and decision-making of the vehicle.

Digital asset: digital assets generally refer to digitized descriptions and models corresponding to various elements in a scene, including but not limited to elements of roads, vehicles, pedestrians, buildings, terrain, weather, and traffic conditions. These digital assets can be directly applied in various simulation, testing, validation, and data driven machine learning. The invention mainly refers to a map and a road model.

High precision map: the high-precision map is a map with finer and more accurate information of various element positions, attributes, forms and the like compared with a common digital map, and comprises finer road information and lane information, including road width, number of lanes, lane markings, intersection planning, topographical information and the like. In the field of autopilot, high-precision maps are usually acquired by sensors and require fine manual labeling on the order of centimeters.

Referring to fig. 1, fig. 1 is a flowchart of a road model generating method according to an embodiment (a) of the present invention. As can be seen from fig. 1, the road model generation method includes the following steps:

s10, classifying the high-precision map data to generate a plurality of channel sub-modules;

referring to fig. 2, fig. 2 provides a type of high-precision map data. As can be seen from fig. 2, the high-precision map data is generally divided into a static map layer, a real-time data layer, a dynamic data layer, and a user model layer; the static map layer comprises a road network, a lane network, traffic facilities and a positioning map layer; the real-time data layer comprises traffic restriction information, traffic flow information and service area information; the dynamic data layer comprises active sensing dynamic information and passive sensing dynamic information; the user model layer includes a driving record data set and a driving experience data set. Each data layer has its own layer, geometry, attributes and associated information.

The static map layer is mainly used for accurately describing a static driving environment and providing rich road semantic information constraint and controlling vehicle behaviors; road network refers to a road system which is formed by various roads and is mutually connected and interweaved into net-shaped distribution in a certain area so as to describe the relationship between road geometry and expression and traffic facilities, including road datum lines, road datum line connecting points, intersections and the like. The highway network is composed of all levels of highways. Urban road networks, which consist of various roads in urban areas. The lane network mainly comprises a lane-level road network, and the relevant attribute of each independent lane in the lane network is recorded to describe lane geometry, road display and the like, including lane-level roads, lane reference lines, lane marking lines and the like. The lane network map layer group is an abstract model of lane objects, is used for describing the topological geometry of high-precision road lanes, and records the related attribute of each independent lane. The lane network layer group is a key information layer group for lane-level path planning.

High precision maps generally have the following roles:

1) Auxiliary perception: if the perceptually identified object is not present in the high-precision map, it is indicated that the object may be an obstacle.

2) Auxiliary positioning: and the vehicle position is used for map matching, so that the vehicle position accuracy is improved.

3) Auxiliary path planning: the high-precision map can narrow the path selection range so as to select the optimal obstacle avoidance scheme.

4) Auxiliary decision and control: the high-precision map can provide key road information for acceleration and deceleration, lane doubling, turning and the like of the vehicle, and is complementary with other sensor data to assist in control.

In this embodiment, in order to make the generated road model more in line with the real road information, the high-precision map data is first acquired, and then the classification processing is performed on the high-precision map data. Since the semantic information contained in the high-precision map is extremely fine, the road model generated on the basis of the high-precision map can be attached to the real road condition.

Because the invention mainly builds the road model, the data of the static map layer needs to be classified, including road network, lane network, traffic facilities and the like.

In the implementation process, based on semantic information provided by the high-precision map, theoretically, data required by the road model can be extracted, and then the road model is built once, but the following problems exist in the way: 1) The data volume to be processed is large in one-time modeling, and the computational power requirement on a computer is high, so that the problems of overlong rendering time, easy error in rendering and the like easily occur. 2) If the rendered model is adapted to various scenes, the model is usually adjusted after the road model is built to obtain different road models, and the one-time road model rendering mode is used for re-rendering all data during modeling even if local road information is modified during modification, so that unnecessary calculation steps are caused, and parameter adjustment during modeling by a tester is inconvenient.

In order to solve the above problems, the present embodiment adopts a modeling concept of split modularization, and first, the high-precision map data is classified and processed to generate a plurality of sub-modules required for constructing a road model, and the calculation amount of one-time rendering is dispersed during the respective construction, so that the calculation amount of each module is not high, and the calculation amount of each module can be performed respectively and simultaneously, thereby greatly improving the modeling efficiency of each module.

S20, fusing the plurality of road sub-modules to generate a road model.

In the last step, a plurality of road sub-modules are obtained after the data of the high-precision map are respectively processed, and in the step, the road model can be generated by mainly fusing the plurality of road sub-modules.

In summary, the embodiment generates a plurality of road sub-modules by classifying and processing the high-precision map data, and then fuses the plurality of road sub-modules to finally generate the road model, so that the modeling speed can be improved, and the generated road model is more attached to a real road.

Illustratively, classifying the high-precision map data, generating the plurality of track sub-modules may be performed by:

1) Determining a point cloud model and an orthographic image corresponding to the road model to be generated;

2) Extracting point cloud data required by constructing a road model from the point cloud model along a road vector line, specifically, the point cloud data corresponding to each road sub-module in the high-precision map data;

3) Extracting geometric features of each path sub-module from the point cloud data to determine a geometric region of the path sub-module;

4) Extracting texture features corresponding to the geometric areas of the sub-modules from the orthographic image;

5) Dividing the geometric region, for example, dividing the geometric region by adopting a triangulation algorithm to generate a plurality of triangular regions;

6) Rendering is carried out according to the triangle area and the texture characteristics of the channel sub-module, and the channel sub-module is respectively generated.

Preferably, when the road sub-module is a road main road, the following procedure may be adopted when performing step 2): based on a road center line vector in preset two-dimensional data, cutting a buffer area with a preset buffer area radius, and filtering from a point cloud model to obtain a first road area point cloud; acquiring the highest point H of the first road area point cloud, gradually reducing the filtering height by taking a fixed step length d as a unit from the highest point H, and filtering the first road area point cloud to obtain a second road area point cloud; and acquiring the road surface point cloud height of the second road area point cloud, and filtering the ground object point cloud of the road surface in the second road area point cloud based on the road surface point cloud height.

By filtering the information of the main road, the road main road excluding other traffic facilities or obstacles on the road surface can be established, and interference by other information is prevented.

In a preferred embodiment, the road vector line can be further segmented, triangular network construction is carried out on each segment by adopting road geometric features, three-dimensional roads are reconstructed, texture mapping is sequentially carried out on the three-dimensional roads by adopting texture features, modeling of auxiliary facilities and modeling of semantic information are carried out, and a real-scene three-dimensional road is generated;

it should be noted that, when the semantic information modeling is performed, the related semantic information is mainly based on the semantic information marked in the high-precision map, if the related semantic information needs to be adjusted, for example, the road semantic information in the road model established according to the scene is changed, and at this time, the related semantic information can be directly adjusted to adapt to the modeling requirement.

In one possible implementation, the road sub-modules include, but are not limited to, a road trunk module, a road traffic marking module, and a road edge module.

The road main road, i.e. the road body, is mainly the whole travelable or non-travelable but visible area of the road, i.e. the part of the road normally covered with asphalt pavement; the main road is a skeleton of an urban road network and is a traffic road for connecting main subareas of cities. Urban road class divides the trunk road, secondary trunk road, branch road tertiary, red line width control at each level: the main road is 30-40 m, the secondary main road is 20-24 m, and the branch road is 14-18 m.

Road traffic marking: yellow line, white line, stop line, etc. The white dotted line is used for separating lanes running in the same direction, and can perform operations such as lane crossing, lane changing and the like; the white solid line does not allow the overtaking driving, and is often used as an extension line of the white dotted line to appear before an intersection or a parking area for separating roadsides, the yellow solid line is used for distinguishing lanes in different directions and is generally drawn in the middle of the road, vehicles on one side of the dotted line allow temporary overtaking or turning around, and vehicles on one side of the solid line cannot allow line pressing, otherwise, the vehicles on one side of the solid line belong to illegal behaviors; the deceleration lines are usually arranged on road surfaces of intersections, school gates and other places, and marked lines or marks for reminding the vehicles of decelerating often appear; the guide line is also an indication mark appearing at an intersection, a ramp or a turning lane; the guiding indication line is used for indicating the direction of the lane; zebra crossings are used for pedestrian aisles, numbers on roads are used for speed limits, etc.

Road edge: the road edge is an important structure for ensuring the overall stability of the road bed and the road surface and removing the water on the road surface, and is also an important component for ensuring the surplus width of two sides required by temporary parking. The maintenance quality of the road edge is directly related to the strength and stability of the road surface and the smoothness of driving.

The road model is constructed based on the high-precision map data, so that the road model has multiple advantages, namely the granularity of the high-precision map data is very fine, the data required by more detailed modeling are contained, and the road model can be adaptively extracted according to the required data so as to build different road scene models.

In this embodiment, the road sub-module may be selected as a road trunk module, a road traffic marking module, and a road edge module. The road scene model is built according to the road main road, the road traffic marking and the road edge information in the current modeling scene, so that a road scene model with less semantic information and convenient test use can be built.

By way of example, assuming now that only tests are being conducted for the speed of the autonomous vehicle, for the recognition of traffic regulations, the road model can be generated solely from the road trunk module, the road traffic marking module and the road edge module. When the road model is built, the road traffic marking module and the road main road module are overlapped, so that the road surface with the traffic marking line can be generated, and then the road edge module is used for fusing, so that the road model with the road edge can be generated. During testing, only the automatic driving vehicle is required to observe whether the automatic driving vehicle can control and adjust the vehicle speed or not in the model, and whether the automatic driving vehicle can accurately recognize and obey the traffic rules and the road identification lines or not.

In one embodiment, the road sub-module is not limited to the above, but there may be a wide variety of elements in the physical road scene as road sub-modules, such as various obstacles, including static and dynamic obstacles, static generally refers to a transportation facility fixed at a certain position, or a parking-violating vehicle, etc. The dynamic obstacle refers to a walking pedestrian, a traveling vehicle, or the like. In addition, other road elements that affect the autopilot test include: traffic lights, weather, road surroundings, road intersections, switches, road and environment texture features, and the like.

Based on the types of the road sub-modules, it can be appreciated that in other embodiments, a modeler can select and generate different numbers and types of road sub-modules according to different scene requirements, and then fuse the road sub-modules to finally generate the road model. For example, if a lane change test of an autonomous vehicle is to be performed, then an obstacle module should be generated, and if a path planning capability is to be tested, a more complex road body is to be generated, including not only major roads, but also other branches, road forks, intersections, roundabout, overhead, crossing roads, etc. are to be modeled together. The specific type and number are selected and determined according to the scene requirements, and the invention is not limited in any way.

Therefore, according to the above-described embodiments, the present invention is capable of generating a plurality of road sub-blocks, respectively, through classification processing based on high-precision map data, and finally adaptively generating road models containing different information according to modeling requirements. The method can reduce the running amount and improve the modeling speed of the sub-module when the sub-module is generated, can match different modeling scenes, has high flexibility, is simple and convenient to operate, and is convenient for modeling staff to implement.

In one embodiment, before the road sub-modules are fused, the road sub-modules may be further post-processed, and then the road sub-modules are fused together according to the processed data to generate the road model.

Although the high-precision map data can express the real road scene in a refined manner, in some simulation tests, the high-precision map data is not synchronized in time because a key test is required for a certain performance of the vehicle, or the real road scene is dynamically updated. At this time, in order to match the modeling requirement, the parameters of the channel sub-module generated in step S10 may be adjusted, re-rendered, and then the corresponding model may be generated.

For example, a road main road generated from a high-precision map has 8 lanes, and without the need to test the lane-changing capability of a vehicle, it is not necessary to simulate so many lanes to increase the calculation force. Then only the parameters of the road main road module are required to be adjusted at the moment, and other road sub-modules do not need to be changed, and after the new road main road module is re-rendered, the new road main road module is fused with the other road sub-modules so as to quickly generate an adjusted road model.

Therefore, if the method of generating the road model at one time is adopted, and sub-modules are not generated respectively, then when the simulation parameters need to be adjusted, all data of the road model need to be recalculated and rendered, obviously, larger resource waste is caused, and the efficiency is quite low. In this embodiment, the generating channel sub-module is a split-module type sub-module, so that not only can the calculation force be reduced, but also local data can be processed in the later adjustment, and the whole model does not need to be repeatedly rendered, so that the model can be dynamically updated, the requirements of simulation personnel on building different simulation test scenes are facilitated, and meanwhile, the efficiency can be greatly improved.

It can be understood that after a plurality of road sub-modules are generated, the road sub-modules can be stored respectively, and the type of the road sub-module required can be determined first and then directly called from a database when a road model is required to be built in the later period, so that the modeling efficiency and the resource utilization rate are further improved when the road sub-modules are not modeled again during each building.

In one embodiment, to enable visualization of the road model, the road model may also be organized in an index based on the earth tile pyramid structure and quadtree subdivision to enable three-dimensional visualization of the road. The earth tile pyramid structure (Earth Tile Pyramid) is a hierarchical data organization that divides the earth's surface into a number of tiles of different sizes, each tile containing a range of geographic data. Under this architecture, the map application can quickly load and present geographic information for a particular region while achieving seamless switching between different zoom levels and different geographic locations. Google Maps is a typical case of using the earth tile pyramid structure.

Quadtree splitting (quad) is a method of dividing two-dimensional data into four quadrants, each of which can be further recursively divided into four sub-quadrants. The method can be used in the fields of data compression, map rendering, image processing and the like. In map applications, quadtree subdivision may be used to organize geographical data at different levels, e.g., for altitude data, fast retrieval and rendering of the elevation map may be accomplished through the quadtree subdivision.

Through the respective use or the combined use of the two methods, the road model can be visualized, the modeling visualization is convenient, the appearance, structure, function and other information of objects, scenes or environments can be clearly presented, better understanding and communication are facilitated, and the automatic driving test efficiency is further improved.

Referring to fig. 3, fig. 3 provides a flow chart of a road generation sub-module, which is mainly used for generating a road main road module. As can be seen from fig. 3, the generation of the road main module mainly includes the following steps:

s101, acquiring road main road information from high-precision map data, wherein the road main road information comprises road boundary information and a communication relation between every two roads;

the road boundary information generally includes information of a start point, end point coordinate information, a road shape, a width, a name, and the like of a road. The information can be used for making road maps and can also be used for road planning applications such as navigation.

On the other hand, the communication relationship between each road refers to the topological relationship between the roads, including the proximity relationship, the intersection relationship, the priority relationship, the steering restriction, and the like. The information can be used for analyzing road network structures, planning vehicle paths, predicting traffic flows and the like.

S102, discretizing road main road information to generate a plurality of polygonal areas;

the discretization processing is carried out on the road main road information, and a plurality of polygonal areas can be generated by splitting the road network into a plurality of small areas. This process is commonly referred to as road meshing or road discretization.

In the road discretization, the whole road network is firstly required to be divided into a plurality of small areas, and the small areas can be square, rectangular, triangular and the like, and can also adopt any other closed shape. Then, for each small area, it is necessary to examine the roads present in the small area in the road network, to compose the roads into an area composed of polygonal boundaries, and to regard the area as a representation of the small area.

It should be noted that, in the process of discretizing, there are several key problems to be solved: first, how to divide the small area involves the problems of selecting what closed shape and the size of the division. And secondly how to determine which small area the road belongs to and to convert the road into a representation of the small area. This typically requires matching the road network data with small areas using techniques such as spatial topology analysis. Finally, how to combine the multiple small areas into a whole generates a polygonal representation of the whole city or region.

Therefore, there may be various implementations of how to discretize to generate a small road area, and the discretized area of this embodiment is merely an exemplary implementation of the discretizing to generate a plurality of polygonal areas, and other geometric areas may be used in other embodiments, which are not limited in any way herein.

S103, generating a road main road module according to the polygonal area.

And finally, generating a road trunk module through rendering according to the road information contained in the polygonal area. Therefore, the road main road module can be quickly and effectively generated through discretization processing.

To aid understanding, in one embodiment, the generation principle of the road arterial module is also provided. Referring to fig. 4, fig. 4 provides a visual effect of semantic information of a high-precision map, and as can be seen from fig. 4, it is possible to clearly extract road boundary information and a communication relationship between each road in the high-precision map. According to the road main road sequence, the road main road can be blocked, specifically: in the case of one-to-one correspondence of left and right boundaries, the block may be approximated as a quadrilateral; when a single boundary is exceeded, the block approximates a triangle. And finally, rendering according to the blocked road area to obtain a road trunk module.

In one possible implementation manner, the generating a road trunk module in step S103 according to the polygonal area specifically includes:

It should be noted that, the triangulation algorithm is an algorithm for dividing an irregular polygon into a plurality of triangles, and there are many practical applications, such as three-dimensional modeling in computer graphics, grid generation in finite element analysis, map making, and the like. In general, triangulation algorithms can be divided into two categories: one is based on triangle partitioning, such as Delaunay triangulation, and the other is based on convex polygon partitioning, such as the Ear Clipping algorithm. Wherein Delaunay triangulation is widely used in real-world scenarios.

The principle of Delaunay triangulation is to find triangles of a set of connection points in a set of points such that the circumscribed circles of these triangles do not contain any other points in the set of points. Therefore, it is called Delaunay triangulation.

Delaunay triangulation can be achieved by both delta and divide-and-conquer methods. The incremental method is to add points into a triangle set one by one, then break and update newly added points through a certain rule, and finally obtain a corresponding Delaunay triangulation result; the dividing and controlling rule adopts a recursion idea to divide and control the point set, respectively processes the left subset and the right subset, and then combines the Delaunay triangulation of the left subset and the right subset to obtain the Delaunay triangulation result of the whole point set.

In a practical application scenario, due to the huge data volume, how to ensure the calculation efficiency and the output quality of the algorithm is very critical. At present, many optimization methods have been proposed, such as a Bowyer-Watson algorithm, a sweep algorithm, a discretization algorithm, etc., in an incremental method, which can greatly improve the calculation efficiency and accuracy of Delaunay triangulation.

In this embodiment, an ear clipping algorithm is preferably used to divide the polygon into non-overlapping and seamless triangles to generate a closely-laid flat pavement mesh. The result of triangulating the polygonal area is shown in fig. 5, and the result of rendering fig. 5 is shown in fig. 6.

Therefore, when rendering is performed, compared with the polygonal area, the triangular area is divided into the triangular areas, so that better rendering effect is achieved, the triangular area in the step S103 is divided by adopting the triangulation algorithm, and finally rendering is performed according to the triangular area generated after division, so that the rendering result of the road trunk road is more attached to the actual effect, and the scene reality can be restored.

In one possible implementation, after the polygon area is divided by using a triangulation algorithm, the method further includes:

1) Generating texture coordinates of regional vertexes based on the divided polygonal regions;

2) Determining material information of a road trunk according to the texture coordinates; the texture information includes texture information;

3) Rendering is carried out according to the material information of the road trunk.

In the above-described embodiment, the polygonal area of the road main road is determined mainly by discretization processing to determine the extent of the rendered road area. However, in order to make the road in the simulation closer to the real value, in one embodiment, texture coordinates of the vertices of the region are generated according to the divided polygonal region, and the paving material of different regions is determined according to the texture coordinates, so as to determine the material information.

Preferably, for the convenience of paving materials, the materials can be preset in a universal way, and the materials can be cut and generated according to road acquisition information, so that the generated materials are more reliable in view of visual effect and reflection.

Note that, the texture information includes texture information first.

The purpose of determining texture information is to texture map the road. In computer graphics, texture Mapping (Texture Mapping) is a widely used technique for enhancing the realism and detail of the surface of objects in a real scene. In this technique, a two-dimensional image (texture) is mapped onto a three-dimensional object surface to provide more detail and realism.

In performing texture mapping, it is necessary to select an appropriate texture sample on the three-dimensional object surface for sampling. Bilinear interpolation sampling is one of the common interpolation methods in texture sampling, and four adjacent texture samples are used in sampling, and interpolation calculation is performed according to the distance and direction relation between the four adjacent texture samples, so that a required texture sample value is obtained. This interpolation method is based on the concept of linear interpolation and weighted averaging, and the sampling result has higher accuracy and better smoothness, which makes it one of the standard interpolation methods in many computer graphics. In real-time graphics processing, textures are sampled by using a bilinear interpolation sampling method, so that a more real image can be generated, and the effect quality of the rendered image is improved.

Further, the texture information includes color, glossiness, transparency, reflectivity, and the like in addition to the texture information.

Color refers to the color of the object surface. In computer graphics, RGB (red green blue) or CMY (cyan Huang Yanggong) models are typically used to represent colors.

The glossiness refers to the smoothness and reflection property of the surface of an object, and can be expressed as the surface effect of different materials such as metal, leather, plastic and the like.

Transparency refers to the degree of transparency of an object, such as a transparent substance like glass, water, etc.

Reflectivity refers to the degree of reflection of light by the surface of an object and may describe the degree of gloss of the surface of the object or the metallic characteristics of the object.

By jointly acting the material information in the modeling and rendering process, the road scene with low quality can be converted into a more real and lifelike picture, and the simulation effect of the road model is improved.

In one embodiment, the texture information of the road surface in a real scene varies with the degree of road wear: for example, the lane-texture change value may be a color change and a normal direction change. It can be understood that the real use condition of the lane can cause the paving colors of different positions of the lane to change, the surface roughness can also change correspondingly, and the capability of reflecting and scattering light of the surface can also change. Due to the fact that the surface smoothness is changed, the surface of the lane is recessed or protruding, and therefore when the light source enters the surface of the lane, the position of reflected light is changed. The normal direction also fluctuates in this way. Therefore, by taking the color change and the normal direction change into consideration, the lane texture change value in the lane texture can be reflected.

Thus, in constructing the texture coordinates, it is considered that the frequently used positions in each lane are similar, i.e., the abrasion position, texture distribution of each lane should be approximately the same. A texture coordinate system can thus be constructed for a lane of the road as a unit. Firstly, selecting any one of all lanes, then taking any point on the left edge line of the lane as an origin, taking a certain vertex of a real lane as the origin, establishing a texture coordinate system, and then establishing texture coordinates (u, v) of the real lane in the texture coordinate system. The value range of u is (0, 1), so that the coordinates of u on the right edge line of the current lane are 1, i.e. the width of the current lane is 1. The u value of other lanes is (0, 1), and the coordinate establishment mode of the current lane can be referred to, so that the width of each lane is set to be 1 preferentially.

As for the v value, in constructing texture coordinates, it is often difficult to construct a complete lane length directly in the extending direction of a single lane, and thus it is often the case that the lane length is segmented, with the intercept point z of each segment as the origin of the coordinate system. Thus, the v value of the lane takes 1 over a range, and the remaining extended portion is re-established. For example, in the v-axis direction, only a partial lane can be taken, at the edge of this region, it is ensured that the v-value reaches 1, and for the later-extending portion, the texture coordinate system is again reestablished from the cut-off position, so that the v-axis continues to elongate in the direction in which the individual lane extends.

After the texture coordinate system and the construction mode of the texture coordinates of the current lane are determined, the rest of the other lanes are repeated according to the mode of the lane until the texture coordinates of each lane are constructed.

After establishing the coordinates in the above manner, a correction function for u can be established. The correction is mainly a function constructed according to the real semantic information of the lane, parameters affecting the texture can be determined according to the actual texture condition of the lane, and fitting is carried out through discrete points so as to find out the rule of the parameters and the texture change. For example, when the texture information is selected as a color, the independent variable affecting the texture parameter can be selected as a position (distance), the texture variation at different positions can be different, the color information of a plurality of positions on the lane can be collected, then the plurality of points are fitted to obtain the function of the color information changing along with the position, further the correction is generated, and finally the lane texture variation value in the lane texture is rapidly calculated by adopting the correction.

In a preferred embodiment, the correction function is a trigonometric function with u as an argument, such as a cos function or a sin function, by means of which the lane texture can be corrected to more closely match the real road scene.

Referring to fig. 7, fig. 7 provides a flow chart for generating a road traffic marking module. As known from fig. 7, in one embodiment, this step mainly includes the steps of:

s104, obtaining point cloud data in the high-precision map data and reflection intensity information of each point;

the laser radar scans the earth surface: lidar scans the earth's surface by transmitting and receiving optical signals, and returns a reflected signal each time a beam of light strikes a surface. Because the high frequency and the speed of the laser are fast, the surface of a large area can be rapidly scanned, and a large amount of point cloud data can be acquired.

And (3) processing point cloud data: the point cloud data measured by the lidar needs to be processed to extract valuable information therein. This typically requires processing such as ground filtering, data registration, data encryption, etc., to convert the point cloud data into a corresponding three-dimensional spatial representation.

And (3) extracting reflection intensity information: the reflection intensity information refers to the reflection intensity of each point cloud data. In general, this requires calculation in combination with the transmit power, receiver sensitivity and reflection surface characteristics of the lidar. In general, the reflected intensity data may be improved in accuracy by collecting and processing multiple lidar data and fusing with image data for multiple perspectives.

S105, extracting road traffic marking information in the high-precision map according to the point cloud data and the reflection intensity information of each point;

preprocessing point cloud data: firstly, preprocessing point cloud data, ground filtering, removing speckle noise, registering the data, and integrating the point cloud data to form a three-dimensional map.

And (5) marking point position prediction: the predicted positions of the marking points are calculated by using the data of the reflected intensity images through a machine learning technology, and are optimized by using methods such as cross validation and the like. This step is most critical and the road marking position can be determined by reflection intensity analysis.

And (3) extracting a marking point: and registering the predicted position and the point cloud data to extract the point cloud data containing the marking points.

Reticle analysis and classification: the reticle points are analyzed and classified by techniques such as machine learning and computer vision. For example, the marking lines are classified according to the characteristics of width, color and the like, and clustering, optimizing and the like are performed.

And (5) reconstructing the marking morphology: and reconstructing and optimizing the marked lines according to the positions and the classification information of the marked lines, extracting the image information of the road, and forming the final road traffic marked line information.

In the above steps, the analysis of the reflection intensity image data is very critical, and the operations of identifying and extracting road and non-road areas, screening noise points and different types of road marks, extracting road boundaries and the like can be performed by utilizing the reflection intensity.

S106, generating a road traffic marking module according to the road traffic marking information.

In particular, road traffic markings are typically polylines that are made by connecting successive points in space. Continuous smooth interpolation is carried out on the point set, so that a smoother reference line can be obtained; if the line is not a single solid line (such as a variable lane white dashed line and a steerable Huang Xuxian), the corresponding point is screened and removed or supplemented by referring to the national standard; finally, the expansion point pair is quadrilateral, and is obliquely split to generate a triangle set, and the road traffic marking module can be obtained through rendering.

Where continuous smooth interpolation is a curve-based function fitting method, the principle is to generate a smooth, interconnected curve from a set of known data points. The curve can be used for representing various shapes of roads, buildings and other ground features. In this embodiment, by continuously and smoothly interpolating the original point cloud data, a more accurate and smooth road traffic marking module can be generated.

Referring to fig. 8, fig. 8 provides a flow chart of a generating route edge module. As known from fig. 8, in one embodiment, this step mainly includes the steps of:

s107, extracting road edge information of different road sections from the high-precision map data, wherein the road edge information comprises length, width and height information of the road edge;

s108, determining a road edge region according to the road edge information, and performing discretization on the road edge region to generate a plurality of polygonal regions;

s109, generating a road edge module according to the polygonal area.

It can be understood that the method used in generating the road edge can refer to the method for generating the road main road module, and the difference between the two methods is mainly that the road edge has a certain height, which is equivalent to converting the road surface grid used in generating the road main road module into a three-dimensional road edge grid. The main means is to generate a polygonal area by discretizing the data to lock the scope of the road edge, and then generate a road edge module according to the polygonal area.

As for the rendering process, in the generating of the road edge, the present embodiment may also divide the polygon by the triangulation algorithm, and then render based on the multiple triangle areas, so as to finally improve the rendering effect, so that the generated road edge is closer to the real scene. The specific segmentation algorithm and segmentation step may refer to the content described in the above embodiments, and will not be further described herein.

Referring to fig. 9, in one embodiment, the present invention further provides a road model generating device, which according to fig. 9 includes the following units:

a road sub-module generating unit 100 for classifying the high-precision map data to generate a plurality of road sub-modules;

and the road sub-module fusion unit 200 is configured to fuse the plurality of road sub-modules and generate a road model.

In one embodiment, the road sub-modules include a road trunk module, a road traffic marking module, and a road edge module.

In one embodiment, the road sub-module generating unit 100 is further configured to generate the road main road module, including:

and generating a road trunk module according to the polygonal area.

In one embodiment, the road sub-module generating unit 100 is further configured to divide the polygonal area by using a triangulation algorithm to generate a road trunk module; wherein the triangulation algorithm comprises an ear clipping algorithm.

In one embodiment, the road submodule generating unit 100 is further configured to, after the dividing the polygonal area by using the triangulation algorithm, further include:

In one embodiment, the road sub-module generating unit 100 is further configured to generate the road traffic marking module, and includes:

In one embodiment, the road submodule generating unit 100 is further configured to generate the road edge module, and includes:

determining a road edge region according to the road edge information, discretizing the road edge region to generate a plurality of polygonal regions,

And generating the road edge module according to the polygonal area.

It may be appreciated that, in some embodiments, functions or modules included in an apparatus provided by the embodiments of the present disclosure may be used to perform a method described in the foregoing method embodiments, and specific implementations thereof may refer to descriptions of the foregoing method embodiments, which are not repeated herein for brevity.

The invention also provides a processor for performing the method of any one of the possible implementations described above.

The invention also provides an electronic device, comprising: a processor, a transmitting means, an input means, an output means and a memory for storing computer program code comprising computer instructions which, when executed by the processor, cause the electronic device to perform a method as any one of the possible implementations described above.

The invention also provides a computer readable storage medium having stored therein a computer program comprising program instructions which, when executed by a processor of an electronic device, cause the processor to perform a method as any one of the possible implementations described above.

Referring to fig. 10, fig. 10 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the invention.

The electronic device 2 comprises a processor 21, a memory 22, input means 23, output means 24. The processor 21, memory 22, input device 23, and output device 24 are coupled by connectors including various interfaces, transmission lines or buses, etc., as are not limited by the present embodiments. It should be appreciated that in various embodiments of the invention, coupled is intended to mean interconnected by a particular means, including directly or indirectly through other devices, e.g., through various interfaces, transmission lines, buses, etc.

The processor 21 may be one or more graphics processors (graphics processing unit, GPUs), which may be single-core GPUs or multi-core GPUs in the case where the processor 21 is a GPU. Alternatively, the processor 21 may be a processor group formed by a plurality of GPUs, and the plurality of processors are coupled to each other through one or more buses. In the alternative, the processor may be another type of processor, and the embodiment of the invention is not limited.

Memory 22 may be used to store computer program instructions as well as various types of computer program code for performing aspects of the present invention. Optionally, the memory includes, but is not limited to, a random access memory (random access memory, RAM), a read-only memory (ROM), an erasable programmable read-only memory (erasable programmable read only memory, EPROM), or a portable read-only memory (compact disc read-only memory, CD-ROM) for associated instructions and data.

The input means 23 are for inputting data and/or signals and the output means 24 are for outputting data and/or signals. The output device 23 and the input device 24 may be separate devices or may be an integral device.

It will be appreciated that in embodiments of the present invention, the memory 22 may not only be used to store relevant instructions, but embodiments of the present invention are not limited to the specific data stored in the memory.

It will be appreciated that fig. 10 shows only a simplified design of an electronic device. In practical applications, the electronic device may further include other necessary elements, including but not limited to any number of input/output devices, processors, memories, etc., and all video parsing devices capable of implementing the embodiments of the present invention are within the scope of the present invention.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, and are not repeated herein. It will be further apparent to those skilled in the art that the descriptions of the various embodiments of the present invention are provided with emphasis, and that the same or similar parts may not be described in detail in different embodiments for convenience and brevity of description, and thus, parts not described in one embodiment or in detail may be referred to in description of other embodiments.

In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.

The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.

In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, produces a flow or function in accordance with embodiments of the present invention, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in or transmitted across a computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital subscriber line (digital subscriber line, DSL)), or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a digital versatile disk (digital versatile disc, DVD)), or a semiconductor medium (e.g., a Solid State Disk (SSD)), or the like.

Those of ordinary skill in the art will appreciate that implementing all or part of the above-described method embodiments may be accomplished by a computer program to instruct related hardware, the program may be stored in a computer readable storage medium, and the program may include the above-described method embodiments when executed. And the aforementioned storage medium includes: a read-only memory (ROM) or a random access memory (random access memory, RAM), a magnetic disk or an optical disk, or the like.

Claims

1. A method of generating a road model, the method comprising:

classifying the high-precision map data to generate a plurality of sub-modules;

and fusing the plurality of road sub-modules to generate a road model.

2. The method of claim 1, wherein the road sub-modules comprise a road trunk module, a road traffic marking module, and a road edge module.

3. The road model generation method according to claim 2, wherein generating the road main road module includes:

and generating a road trunk module according to the polygonal area.

4. The method of generating a road model according to claim 3, wherein the generating a road trunk module according to the polygonal area comprises:

5. The road model generation method according to claim 4, further comprising, after the dividing the polygonal region by a triangulation algorithm:

6. The road model generation method according to claim 2, wherein generating the road traffic marking module includes:

7. The method of generating a road model of claim 2, wherein generating the road edge module comprises:

and generating the road edge module according to the polygonal area.

8. A road model generation apparatus, characterized in that the apparatus comprises:

9. An electronic device, comprising: a processor and a memory for storing computer program code comprising computer instructions which, when executed by the processor, the electronic device performs the road model generation method of any one of claims 1 to 7.

10. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a computer program comprising program instructions which, when executed by a processor of an electronic device, cause the processor to perform the road model generation method of any one of claims 1 to 7.