CN115081303A

CN115081303A - Laser radar virtual modeling and simulation method, electronic device and storage medium

Info

Publication number: CN115081303A
Application number: CN202210995517.3A
Authority: CN
Inventors: 王强; 翟洋; 陈硕; 王寅东; 孟佳旭; 杜志彬; 赵帅; 赵鹏超; 国建胜; 沈永旺; 张鲁; 张骁; 刘子毅; 马文霄; 孙博华
Original assignee: Automotive Data of China Tianjin Co Ltd
Current assignee: Automotive Data of China Tianjin Co Ltd
Priority date: 2022-08-19
Filing date: 2022-08-19
Publication date: 2022-09-20
Anticipated expiration: 2042-08-19
Also published as: CN115081303B

Abstract

The embodiment of the invention discloses a virtual modeling method and a simulation method for a laser radar, electronic equipment and a storage medium. The modeling method comprises the following steps: acquiring a laser radar simulation model, wherein the model comprises a simulation algorithm suitable for multiple radars and simulation parameters to be determined; constructing multiple groups of values of the simulation parameters according to the specification of a specific radar; simulating radar data of the specific radar in a specific scene according to the simulation algorithm and each group of values; clustering errors of a plurality of groups of simulation results by adopting a particle swarm algorithm, wherein different classes represent different error levels; and selecting an optimal value from the multiple groups of values according to the type with the minimum error, and forming a final laser radar simulation model by the optimal value and the simulation model together. The embodiment improves the universality of the model parameters.

Description

Laser radar virtual modeling and simulation method, electronic device and storage medium

Technical Field

The embodiment of the invention relates to the field of sensor simulation modeling, in particular to a laser radar virtual modeling method, a laser radar simulation method, electronic equipment and a storage medium.

Background

The laser radar simulation model can adopt virtual model test to replace real vehicle test, and promotes the development of vehicle intelligent evaluation technology. Particularly, in the test under the dangerous test condition, the simulation model is used for testing more safely and efficiently, the defect of insufficient practicability caused by high price is overcome, and potential safety hazards caused by uncertainty of a laser radar algorithm in the test process are eliminated.

Under a certain simulation algorithm, the value of the simulation parameter influences the error of the simulation result to a great extent. In the prior art, before simulation calculation of any type of radar is performed, multiple sets of simulation parameters are generally required to be simulated respectively, and parameter values with high simulation precision at selected positions are continuously used, so that simulation result distortion caused by improper simulation parameter selection is avoided. When the radar model changes, the parameter value may not be suitable any more, and even a large result error is generated.

Disclosure of Invention

The embodiment of the invention provides a virtual modeling method and a simulation method for a laser radar, electronic equipment and a storage medium, and improves the universality of simulation parameters.

In a first aspect, an embodiment of the present invention provides a virtual modeling method for a laser radar, including:

acquiring a laser radar simulation model, wherein the model comprises a simulation algorithm suitable for multiple types of radars and to-be-determined simulation parameters;

constructing multiple groups of values of the simulation parameters according to the specification of a specific radar;

simulating radar data of the specific radar in a specific scene according to the simulation algorithm and each group of values;

clustering errors of a plurality of groups of simulation results by adopting a particle swarm algorithm, wherein different classes represent different error levels;

and selecting an optimal value from the multiple groups of values according to the class with the minimum error, and forming a final laser radar simulation model by the optimal value and the simulation model together.

In a second aspect, an embodiment of the present invention provides a laser radar virtual simulation method, including:

and simulating the radar data of any one of the plurality of types of radars in any scene by adopting the final laser radar simulation model constructed by the method in the embodiment.

In a third aspect, an embodiment of the present invention further provides an electronic device, where the electronic device includes:

one or more processors;

a memory for storing one or more programs,

when the one or more programs are executed by the one or more processors, the one or more processors implement the lidar virtual modeling method or the lidar virtual simulation method of any of the embodiments.

In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the laser radar virtual modeling method or the laser radar virtual simulation method according to any embodiment.

The embodiment of the invention takes the errors under the combination of different value parameters as discrete particles, and adopts a machine learning algorithm to cluster the error particles; and selecting an optimal value of the simulation parameter according to the type of the particles with the minimum error, wherein the value is suitable for any radar under any scene. Compared with a certain example, one type of particles has stronger generalization capability, so that the reflectivity errors under different scenes or different types of radars are smaller, better simulation effect of the multiple types of radars under the optimal value is ensured as much as possible, and the universality of parameter value is improved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.

Fig. 1 is a flowchart of a virtual modeling method for a laser radar according to an embodiment of the present invention.

Fig. 2 is a flowchart of a simulation algorithm in a lidar simulation model according to an embodiment of the present invention.

Fig. 3 is a flowchart of a virtual simulation method for a laser radar according to an embodiment of the present invention.

FIG. 4 is a point cloud distribution effect diagram of a simulation test when a person is 8m away from a laser radar.

Fig. 5 is a cloud distribution effect diagram of a vehicle test when a person is 8m away from a laser radar according to the present invention.

Fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplification of description, but do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

In the description of the present invention, it should also be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in a specific case to those of ordinary skill in the art.

Fig. 1 is a flowchart of a virtual modeling method for a laser radar according to an embodiment of the present invention. The method is suitable for the situation that the universal simulation parameters are determined under a certain simulation algorithm and is executed by the electronic equipment. As shown in fig. 1, the method specifically includes:

s110, a laser radar simulation model is obtained, wherein the model comprises a simulation algorithm suitable for multiple radars and to-be-determined simulation parameters.

The simulation model is used for simulating an echo signal of laser emitted by the laser radar in a certain scene. The simulation model adopted by the embodiment is suitable for various radars, has certain universality, and simulation parameters are undetermined. The embodiment aims to find an optimal simulation parameter value, and ensures that a plurality of radars can obtain a good simulation effect under the optimal value as far as possible, so that the simulation parameters have certain universality, and repeated tests and selection of the parameters in each simulation process are omitted.

Further, the parameters of the simulation model include input parameters and output parameters. For the output parameters, the laser radar has better comprehensive performance in the aspects of measurement precision, measurement distance and the like, so the radar ranging precision and the point cloud density are used as two output parameters of the simulation model; meanwhile, the reflectivity of the target is also an important index for measuring the performance of the radar, so that the reflectivity of the target is particularly increased as the output of the simulation model. The output parameters are usually set according to the technical requirements of vehicle intelligent evaluation, and the distance measurement precision error, the point cloud density error and the reflectivity error which are achieved by the laser radar are determined.

The input parameters, i.e., the simulation parameters, can be roughly classified into two types. The first type is set according to specific radar specifications, parameter values do not need to be selected particularly, and different types of radars usually correspond to different parameters. Such parameters include: the coordinates and the positions of laser radar mounting points, the number of scanning line beams, scanning modes, vertical and horizontal scanning angles, detection distances and the like. The second type can be set according to simulation requirements, and when different parameter values are selected, different simulation effects (such as simulation precision) are often corresponding to the second type. Such parameters include: the radar angular resolution, the scanning frequency, and the like, the embodiment determines the optimal values of such simulation parameters. It is worth mentioning that both the radar angular resolution and the scanning frequency have important influence on the reflectivity error of the target object, and the selection of the appropriate angular resolution and scanning frequency value is very important for improving the reflectivity error.

And S120, constructing multiple groups of values of the simulation parameters according to the specification of a specific radar.

In this embodiment, parameter setting is performed starting from a specific radar, on one hand, values of the first type of simulation parameters are set according to the specifications of the radar, and on the other hand, multiple groups of values are constructed for the second type of simulation parameters according to the angular resolution range, the scanning frequency range and the like of the radar.

In particular, angular resolution refers to the angular resolution between the laser beam and the beam. For a uniformly distributed laser beam, the angular resolution between the line beams is also uniformly distributed. For a non-uniformly distributed laser beam, the distribution of angular resolution is also non-uniform, e.g., 1 degree between the first line to the 10 th line, 2 degrees between the 11 th line to the 20 th line, etc. The method provided by the implementation is particularly suitable for the laser beams which are not uniformly distributed. The angular resolution between each two beams constitutes a vector, which together with the scanning frequency constitutes a set of simulation parameters.

And S130, simulating the radar data of the specific radar in the specific scene according to the simulation algorithm and each group of values.

Before simulation is carried out by using a simulation model, an evaluation scene for simulation is constructed firstly. Specifically, the construction of the evaluation scene comprises the following steps:

the method comprises the steps of firstly, determining the types and the number of target species of an evaluation scene according to the requirements of vehicle intelligent evaluation technology, constructing a three-dimensional model of the target according to a specified scale, and storing the whole scene in a computer hard disk or other memories in a data file mode. The target objects comprise vehicles, traffic signs, pedestrians, traffic lights and the like.

And step two, sampling the three-dimensional model and converting the three-dimensional model into a two-dimensional image. Optionally, the three-dimensional model is a three-dimensional point cloud, and the three-dimensional coordinate of each data point in the point cloud in the horizontal reference coordinate system is (X, Y, Z). Suppose the data point on one edge of the three-dimensional point cloud A is S: (a ₁ ,b ₁ ,c ₁ ) And T: (A)a ₂ ,b ₂ ,c ₂ ) Connecting two data points to rotate the line segment ST around the Z axis by a certain angle

And enabling the line segment ST and the X axis to be positioned in the same plane to obtain the three-dimensional point cloud B. Wherein the angle of rotation

The calculation process of (2) is as follows:

a rotation matrix about the Z axis of

。

Then, B is rotated about the Y axis by the same method

The rotation matrix is

Obtaining a three-dimensional point cloud C; rotating the C by an angle of rotation around the X axis

The rotation matrix is

And obtaining the final three-dimensional point cloud. After the three-time axial rotation, the final three-dimensional point cloud is parallel to the horizontal reference coordinate system, and then the final three-dimensional point cloud is projected to each plane in the coordinate system to obtain a two-dimensional image.

Thirdly, generating a basic reflectivity map of the target object according to the two-dimensional image and the material of the target object; and generating the real reflectivity of the three-dimensional model according to the basic reflectivity map. Different materials correspond to different basic reflectivities, a basic reflectivity map of the target object is generated in the two-dimensional image, and the map is converted back to a three-dimensional space according to the transformation matrix, so that the real reflectivity of the three-dimensional model can be obtained. The basic reflectivity material model mainly describes the local light reflection attribute of the object surface, and the description and accurate modeling of the material can enhance the reality of the rendering result. Examples of the vehicle include a metal material, a rubber material, and a glass material. Different materials can influence the behavior of the interaction of the laser beam and the materials, the realistic material model is expressed as a bidirectional reflection distribution function, and the functions can be divided into different models, and the surface material model is adopted optionally.

The number of the constructed evaluation scenes can be multiple, a specific scene is selected from the multiple evaluation scenes, each group of values of simulation parameters are respectively substituted into the simulation algorithm, radar data of the specific radar under the specific scene are simulated, and each group of values corresponds to one group of simulation results.

And S140, clustering the errors of the multiple groups of simulation results by adopting a particle swarm algorithm, wherein different classes represent different error levels.

Since the angular resolution and the scanning frequency have a large influence on the reflectivity error, the quality of each set of parameter values is reflected according to the reflectivity errors of multiple sets of simulation results. Firstly, according to a plurality of groups of real radar data of the specific radar under the specific scene, reflectivity errors of a plurality of groups of simulation results are calculated. Then, the reflectivity errors are classified in an objective and subjective combination mode, and the multiple groups of reflectivity errors are divided into five types of large errors, medium errors, small errors and small errors. The classification method based on fuzzy evaluation reflects the error level simulated under different parameter combinations to a certain extent.

Specifically, firstly, each group of errors is taken as one discrete particle, and at least one cluster partition of all the discrete particles is determined by adopting a particle swarm algorithm, so that the discrete errors in the group are minimized. Optionally, the reflectivity error of each point cloud data is used as an error value, or the reflectivity error of each material is used as an error value, and a group of errors is formed by a plurality of error values. And taking each group of errors as a discrete particle to form a data set Q for particle swarm search, and finding a partition which minimizes the intra-class dispersion sum in Q. Wherein the sum of dispersion may be expressed as:

wherein V represents the number of clusters (= 5),

the j-th cluster center is represented,

representing particles

To the center of the cluster

The distance of (c). The clusters are divided as objective classification results.

And meanwhile, acquiring the subjective division of the user on all discrete particles, wherein the subjective division result is obtained by observing the real radar data and the point cloud display of the simulation result. And the subjective evaluation of errors adopts a questionnaire form, the simulation result is displayed as laser point cloud data through an upper computer, the laser point cloud data is compared with real point cloud data of the radar in a real scene, the subjective evaluation is carried out on the reflectivity errors, and a subjective classification result is given.

And finally, selecting one clustering partition with the largest contact ratio with the subjective partitions as a final clustering partition for further parameter selection.

S150, selecting an optimal value from the multiple groups of values according to the class with the minimum error, and forming a final laser radar simulation model by the optimal value and the simulation model together.

In this embodiment, the type with the smallest error among the five types of reflectivity errors, that is, the type with the smallest error is calibrated, and the optimal value is selected from multiple groups of values corresponding to the type of particles and used as the optimal parameter of the laser radar simulation model. Specifically, determining the clustering center of the class with the minimum error; selecting at least one group of errors closest to the cluster center; and selecting the values corresponding to the at least one group of errors from the plurality of groups of values as optimal values. And the optimal value and the simulation model jointly form a final laser radar simulation model.

In this step, the particle with the smallest accumulated error is not directly selected from the plurality of reflectivity error particles, and the corresponding parameter combination is used as the optimal parameter. This is because the particle has the smallest error among several sets of simulation results of a specific radar in a specific scene, and when the simulation scene or the radar type changes, the parameter combination corresponding to the particle does not necessarily correspond to the smallest reflectivity error, and even a larger error may be generated. Therefore, in the step, the clustering center of the particle with the smallest error is selected, the particle closest to the clustering center is selected, the parameter combination corresponding to the particle is used as the optimal value, and the value is suitable for any radar in any scene. The clustering center has the strongest generalization capability and can represent the characteristics of the particles, so that the probability that the reflectivity errors under different scenes or different types of radars can be kept to be the maximum in the class is ensured, and the good simulation effect of the multiple types of radars under the optimal values is ensured as far as possible.

Optionally, there are multiple groups of the optimal values; after selecting the optimal value from the multiple groups of values according to the class with the minimum error, the method further comprises the following steps: simulating radar data of the specific radar under a plurality of scenes according to each group of optimal values, and verifying whether the error of the simulation result under each scene falls into the class with the minimum error; and selecting a group of optimal values with the largest number of scenes in the class, and forming a final laser radar simulation model by the optimal values and the simulation model.

Selecting a plurality of particles nearest to the clustering center, and taking a plurality of groups of parameter values corresponding to the particles as optimal values to be selected. Then inputting any group of values into a simulation algorithm, and simulating radar data under multiple scenes to obtain multiple error particles; and according to the classification boundary obtained in S140, the number of particles U falling in the class with the smallest error is counted. And after the particle number U corresponding to each group of values is obtained, taking the group of values with the minimum U as the final optimal value, and forming a final laser radar simulation model together with the simulation model.

For the convenience of understanding the present embodiment, several typical data structures of the test scenario and the error particle are given below, and the data structures of the test scenario and the error particle in practical applications are not limited thereto:

in a first scene, the laser radar is installed right ahead the host vehicle, the host vehicle is always in a static state, the target vehicle is used as a detection target, and the target vehicle is respectively located at positions 5m, 10m, 20m, 30m, 40m, 50m, 60m, 80m, 100m, 120m and 150m away from the laser radar installation position of the host vehicle. Before simulation calculation, real radar data of not less than 5s are collected at each test point respectively. After simulation calculation is performed on each test point, simulation result errors corresponding to each test point are fused (for example, spliced into a group of vectors) to form a scene of simulation result errors.

And in the second scenario, the laser radar is installed right ahead the main vehicle, the main vehicle is always in a static state, the target pedestrian is used as a detection target, and the target pedestrian is respectively positioned at positions 5m, 10m, 20m, 30m, 40m, 50m, 60m, 80m, 100m, 120m and 150m away from the right ahead of the laser radar installation position of the main vehicle. Before simulation calculation, real radar data of not less than 5s are collected at the test points respectively. And after simulation calculation is carried out on each test point, corresponding simulation result errors of each test point are fused (for example, spliced into a group of vectors) to form simulation result errors in a second scene.

And thirdly, the laser radar is installed right ahead of the main vehicle, the main vehicle is always in a static state, a plane plate (or a wall or a metal vehicle) with known material reflection intensity is used as a detection target, and the plane plate (or the wall or the metal vehicle) with known material reflection intensity is respectively positioned at positions 5m, 10m, 20m, 30m, 40m, 50m, 60m, 80m, 100m, 120m and 150m right ahead of the installation position of the laser radar of the main vehicle. Before simulation calculation, real radar data of not less than 5s are collected at the test points respectively. And after simulation calculation is carried out on each test point, corresponding simulation result errors of each test point are fused (for example, spliced into a group of vectors) to form simulation result errors under the third scene.

After the initial range of the optimal value is limited through the machine learning algorithm, the selection range of the optimal value is further narrowed through changing different scenes, the adaptability of the optimal value of the simulation parameter to the use scene is improved, and the simulation precision is further guaranteed.

On the basis of the above embodiment and the following embodiment, the simulation algorithm in the simulation model is refined. Optionally, the electronic device in the above embodiment integrates a central processing unit and a graphics processing unit, and the embodiment adopts a mode of combining the central processing unit and the graphics processing unit to realize the calculation of the simulation algorithm. The central processing unit is mainly responsible for preparing and updating simulation calculation data of scenes, target objects and the like and outputting simulation calculation results; and the graphic processor is mainly responsible for realizing the transmission calculation of the physical behavior generated when the laser collides with the target object according to the physical formula of the laser radar.

Optionally, the image processor comprises a laser generator and a collision renderer. And the graphics processor simulates the radar data of the specific radar in a specific scene according to the simulation algorithm and each group of values. As shown in fig. 2, the simulation process specifically includes the following steps:

step one, a laser generator simulates laser emitted by the specific radar under the specific scene according to any group of values of the radar angular resolution and the scanning frequency. Specifically, the graphic processor loads an input data stream into the memory, wherein the input data stream comprises key parameters of the laser radar, coordinates and positions of a laser radar mounting point, the number of scanning beams, scanning frequency, scanning mode, vertical and horizontal scanning angles, angular resolution, motor speed, detection distance and the like, and further comprises data such as atmospheric effect texture, evaluation scene target object material definition, vertex texture and the like. The laser generator defines a laser beam emitted by the radar according to the loaded data stream.

And step two, judging whether the laser collides with the target object in the specific scene by a collision renderer. If the laser does not collide with the object, no rendering is required.

And step three, if collision occurs and is the nearest collision point, determining the physical behaviors of the laser by the collision renderer according to the material of the target object, wherein the physical behaviors comprise reflection, transmission and refraction. Specifically, if the laser and the target object have a collision behavior and are the closest collision point, the collision renderer determines whether the target object is transparent, and if the target object is not a transparent material, the physical behaviors of the laser, including reflection, transmission, refraction and the like, are determined according to the material of the target object until the laser beam traverses all the target objects. The target object is classified into a metal material, a semiconductor material, and a dielectric material (transparent material), and PBR (physical-Based Rendering) material is used. Different materials correspond to different physical behaviors.

And step four, selecting a corresponding rendering model by the collision renderer according to the material of the target object. The rendering model is used for describing rendering behaviors when the laser collides with the target object, and the target objects made of different materials correspond to different rendering models.

Fifthly, determining the absorption rate, the reflection rate and the transmission rate of the collision point by a collision renderer according to the rendering model and the roughness, the transparency and the texture of the target object; and substituting the improved bidirectional reflection distribution function model to simulate the physical behavior. Specifically, the material of the target object and the position information of the scene are used as input, the physical behavior of the laser at the collision point is expressed by a physical formula, and if the collision point is directly reflected back, the intensity is regarded as the primary echo intensity. If the echo in the second reflection direction does not consider diffuse reflection and refraction, the collision point is directly used as a trigger position, and the reflection direction is used as the direction of the second emission.

Particularly, the improved bidirectional reflection distribution function model can reflect the effect of continuous change from pure diffuse reflection to specular reflection, and the physical behavior is simulated by adopting the reflection distribution function model, so that higher fidelity can be obtained. Specifically, first, the intermediate variables are calculated:

，

，

，

，

wherein, in the step (A),

，

，

，

，

m represents a normal vector of the surface of the target object, N represents a tangent vector of the surface of the target object, P represents an incident direction of the laser, Q represents an emitting direction of the laser, K represents an angular bisector of P and Q, and K represents a normal angle ₁ Representing the projection of K onto the target object plane.

Then, the improved two-way reflection distribution function model function value is calculated:

wherein the content of the first and second substances,

representing the wavelength of incident light as a spectral parameter

When the spectrum radiation intensity in the reflection direction is in proportion to the incident spectrum radiation intensity,

wherein, in the process,

indicating wavelength

The reflection coefficient of the incident light at the incident angle of 0 degree is within the range of

；

And the direction function is expressed, the direction characteristic of the improved bidirectional reflection distribution function model is expressed, and the direction function is composed of diffuse reflection and specular reflection.

Optionally, the transmitted or refracted laser beam continues to traverse other target objects in the scene, and a reflection, transmission or refraction action occurs. For most homogeneous media, the sine ratio of the angle of incidence and the angle of refraction of the laser is constant. The material with different densities has different influence rules on the refraction degree of the laser, and the refraction angle is generally in negative correlation with the density. And determining the refraction angle of the laser according to the definition of the material density in the scene, and continuously traversing the target object in the scene by the refracted light. The number of lines and the incident angle defined by the laser qualitatively know the transmission condition of the geometric dimension laser ray of the target object according to the physical behavior of the intersection of the laser and different objects and the propagation path of the laser.

After the physical behavior of the laser is simulated by the graphic processor, the obtained simulated radar data (including the distance, the echo intensity, the direction of the laser radar and the like) are stored in the shared memory in a point cloud mode. Then the graphics processor prepares a virtual screen, and processes and displays the point cloud in the memory; a large amount of data is converted into graphs or images, and the graphs or the images are displayed and interacted on a screen, so that a quick visualization function is realized.

At present, a laser radar simulation method is mostly realized based on a physical engine, and the problems of slow starting, overlarge resource occupation in computer equipment and the like generally exist, so that the efficiency of the existing simulation technology is low. The embodiment provides a general virtual modeling algorithm for a laser radar, which is implemented by software on a simulation model of the laser radar based on a central processing unit and a graphic processor. The parallelism of laser radar simulation data is embodied through a flow calculation model of a graphic processor, and the simulation calculation speed is accelerated at a hardware level by utilizing a computer graphic acceleration card (a display card).

Meanwhile, the embodiment provides a rendering method based on laser tracking, which is simple in calculation, easy to parallelize, suitable for virtual simulation of laser radars with different specifications under various environmental conditions, and high in universality. Meanwhile, a simulation model of the laser radar is established according to the detection mechanism of the laser radar and the physical shielding and reflection properties of laser, the mechanism and the influence factors of point cloud generation of the laser radar are reflected, the simulation model can be applied to a simulation environment, and the vehicle intelligent driving related application based on the laser radar model in the simulation environment can be further completed.

Fig. 3 is a flowchart of a virtual simulation method for a laser radar according to an embodiment of the present invention. As shown in fig. 3, the method specifically includes the following steps:

s210, simulating radar data of any one of the plurality of types of radars in any scene by adopting the final laser radar simulation model constructed by the method in any embodiment.

As described in the above embodiments, the lidar simulation model is applicable to a universal model of a plurality of types of radars. The optimal parameters determined in the above embodiments are also applicable to any one of the plurality of radars. In order to verify the simulation accuracy of the final lidar simulation model, in the embodiment, in the static target detection test, the actual scene is compared with the lidar point cloud generated in the virtual scene, as shown in fig. 4 and 5. The actual vehicle testing distance between the person and the laser radar is about 8.116m, and the simulation testing distance is 8.239 m. In addition, the embodiment also tests the reflection intensity of the target object at a fixed distance, and the result shows that the point clouds of the same material attribute of the same object have the same reflection intensity and the same reflectivity; and comparing the calculated intensities of different materials at the same distance, wherein the reflection intensity of the steel is greater than that of the wall surface. In addition, as the distance between the target vehicle and the radar is gradually increased, the reflectivity of the radar is regulated by the same rule.

Fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, as shown in fig. 6, the electronic device includes a processor 60, a memory 61, an input device 62, and an output device 63; the number of processors 60 in the device may be one or more, and one processor 60 is taken as an example in fig. 6; the processor 60, the memory 61, the input device 62 and the output device 63 in the apparatus may be connected by a bus or other means, as exemplified by the bus connection in fig. 6.

The memory 61 is a computer-readable storage medium for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the virtual modeling method for lidar in the embodiment of the present invention. The processor 60 executes various functional applications of the device and data processing by executing software programs, instructions and modules stored in the memory 61, that is, implements the laser radar virtual modeling method described above.

The memory 61 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory 61 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, the memory 61 may further include memory located remotely from the processor 60, which may be connected to the device over a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The input device 62 may be used to receive entered numeric or character information and to generate key signal inputs relating to user settings and function controls of the apparatus. The output device 63 may include a display device such as a display screen.

Embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the virtual modeling method for a lidar according to any of the embodiments.

Computer storage media for embodiments of the present invention may take the form of any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + +, or the like, as well as conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the technical solutions of the embodiments of the present invention.

Claims

1. A virtual modeling method of laser radar is characterized in that,

acquiring a laser radar simulation model, wherein the model comprises a simulation algorithm suitable for multiple radars and simulation parameters to be determined;

2. The method of claim 1, wherein the simulation parameters include radar angular resolution and scan frequency, and wherein the radar data includes object reflectivity.

3. The method of claim 2, wherein before the simulating the radar data of the specific radar in the specific scene according to the simulation algorithm and each set of values, further comprising:

constructing a three-dimensional model of a target object in a specific scene;

sampling the three-dimensional model and converting the three-dimensional model into a two-dimensional image;

generating a basic reflectivity map of the target object according to the two-dimensional image and the material of the target object;

generating the real reflectivity of the three-dimensional model according to the basic reflectivity map;

the clustering of the errors of the multiple groups of simulation results by adopting the particle swarm algorithm comprises the following steps:

and calculating the reflectivity error of each group of simulation results according to the real reflectivity.

4. The method of claim 1, wherein simulating the radar data of the specific radar in the specific scene according to the simulation algorithm and each set of values comprises:

simulating laser emitted by the specific radar in the specific scene according to any group of values of the angular resolution and the scanning frequency of the radar;

judging whether the laser collides with a target object in the specific scene;

if the collision occurs and is the nearest collision point, determining the physical behaviors of the laser, including reflection, transmission and refraction, according to the material of the target object;

selecting a corresponding rendering model according to the material of the target object;

and determining the absorptivity, reflectivity and transmissivity at the collision point according to the rendering model and the roughness, transparency and texture of the target object, substituting the determined absorptivity, reflectivity and transmissivity into an improved bidirectional reflection distribution function model, and simulating the physical behavior.

5. The method of claim 1, wherein clustering the errors of the plurality of sets of simulation results using a particle swarm algorithm comprises:

calculating errors of a plurality of groups of simulation results according to a plurality of groups of real radar data of the specific radar in the specific scene, wherein each group of values of the simulation parameters corresponds to one group of real radar data and one group of errors;

taking each group of errors as a discrete particle, and determining at least one cluster division of all discrete particles by adopting a particle swarm algorithm to minimize the discrete errors in the cluster;

acquiring subjective division of all discrete particles by a user, wherein the subjective division result is obtained by observing point cloud display of real radar data and a simulation result;

selecting one clustering partition with the largest coincidence degree with the subjective partitions as a final clustering partition;

wherein each cluster partition and the subjective partition define the following fuzzy types: the error is big, the error is medium, the error is little and the error is little.

6. The method of claim 1, wherein selecting the optimal value from the plurality of sets of values according to the class with the smallest error comprises:

determining the clustering center of the class with the minimum error;

selecting at least one group of errors closest to the cluster center;

and selecting the values corresponding to the at least one group of errors from the plurality of groups of values as optimal values.

7. The method of claim 1, wherein there are multiple sets of the optimal values;

after selecting the optimal value from the multiple groups of values according to the class with the minimum error, the method further comprises the following steps:

simulating radar data of the specific radar under a plurality of scenes according to each group of optimal values, and verifying whether the error of the simulation result under each scene falls into the class with the minimum error;

and selecting a group of optimal values with the largest number of scenes in the class, and forming a final laser radar simulation model by the optimal values and the simulation model.

8. A virtual simulation method for laser radar is characterized by comprising the following steps:

and simulating the radar data of any one of the plurality of radars in any scene by using the final laser radar simulation model constructed by the method according to any one of claims 1-7.

9. An electronic device, comprising:

one or more processors;

a memory for storing one or more programs,

when executed by the one or more processors, cause the one or more processors to implement the lidar virtual modeling method of any of claims 1-7, or the lidar virtual simulation method of claim 8.

10. A computer-readable storage medium, having stored thereon a computer program which, when executed by a processor, implements the lidar virtual modeling method of any of claims 1-7, or the lidar virtual simulation method of claim 8.