CN116362119A - Automatic driving simulation scene library expansion method based on GA genetic algorithm - Google Patents
Automatic driving simulation scene library expansion method based on GA genetic algorithm Download PDFInfo
- Publication number
- CN116362119A CN116362119A CN202310255344.6A CN202310255344A CN116362119A CN 116362119 A CN116362119 A CN 116362119A CN 202310255344 A CN202310255344 A CN 202310255344A CN 116362119 A CN116362119 A CN 116362119A
- Authority
- CN
- China
- Prior art keywords
- environment
- genetic algorithm
- scene
- driving simulation
- ttc
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of automatic driving simulation, and discloses an automatic driving simulation scene library expansion method based on a GA genetic algorithm, which comprises the following steps: s1, determining scene types and environment types as parameter spaces of a genetic algorithm; s2, initializing; s3, individual evaluation; s4, selecting operation; s5, performing cross operation; s6, judging termination conditions: if t=t, outputting the individual with the best fitness obtained in the evolution process as an optimal solution, and terminating the calculation; and S7, outputting and storing the optimal solutions, and cross-combining the environmental type influence with the output data of each optimal solution to form final scene data. The automatic driving simulation scene expansion is carried out through the genetic algorithm, the optimal result can be rapidly screened, the optimal offspring is selected according to the authenticity index through the environment factor intersection, massive scene data are generated through iteration, and key characteristics of scene library expansibility, infinity, batch quantity, automation and the like are better met.
Description
Technical Field
The invention relates to the technical field of automatic driving simulation, in particular to an automatic driving simulation scene library expansion method based on a GA genetic algorithm.
Background
Genetic algorithms (Genetic Algorithm, GA) were originally proposed by John holland in the united states in the 70 s of the 20 th century, designed based on the rules of evolution of organisms in nature. The method is a calculation model of the biological evolution process simulating the natural selection and genetic mechanism of the Darwin biological evolution theory, and is a method for searching the optimal solution by simulating the natural evolution process. The algorithm converts the solving process of the problem into processes like crossing, mutation and the like of chromosome genes in biological evolution by using a computer simulation operation in a mathematical mode. When solving the complex combined optimization problem, a better optimization result can be obtained faster than that of some conventional optimization algorithms. Genetic algorithms have been widely used in the fields of combinatorial optimization, machine learning, signal processing, adaptive control, and artificial life.
In an automatic driving simulation scene, in order to acquire a large number of simulation databases, the adoption of the field to acquire data and the simulation setting is definitely too slow and complicated, and the cost is greatly increased, so that a rapid and efficient method is needed to realize the rapid expansion of the simulation scene library, and the problem can be better solved by adopting a genetic algorithm.
Disclosure of Invention
The invention aims to provide an automatic driving simulation scene library expansion method based on a GA genetic algorithm, so as to solve the problems that the adoption of on-site data acquisition and simulation setting in the background technology is definitely too slow and tedious, and the cost is greatly increased.
In order to achieve the above purpose, the present invention provides the following technical solutions: the automatic driving simulation scene library expansion method based on the GA genetic algorithm comprises the following steps:
s1, determining scene types and environment types as parameter spaces of a genetic algorithm;
s2, initializing, setting an evolution algebra counter t=0, setting a maximum evolution algebra T, and randomly generating M individuals as an initial population P (0);
s3, individual evaluation, namely calculating the fitness of each individual in the group P (t);
s4, selecting operation, and determining an operation formula;
s5, performing cross operation, namely performing cross grouping on the screened high-quality individuals to generate scene data;
s6, judging termination conditions: if t=t, outputting the individual with the best fitness obtained in the evolution process as an optimal solution, and terminating the calculation;
and S7, outputting and storing the optimal solutions, and cross-combining the environmental type influence with the output data of each optimal solution to form final scene data.
Preferably, in S1, the scene types include extreme danger at the time of collision classification level, collision edges, ensuring that no collision occurs, generating an unamulation.
Preferably, the collision time TTC is used as a key scene type selection index, the TTC represents extreme danger at 0.7-0.9s, the TTC represents collision edge at 0-0.7s, the TTC is larger than 0.9s, the situation that collision cannot occur is ensured, and the TTC is negative number, so that simulation cannot be generated;
the TTC calculation formula and fitness function formula are as follows:
TTC＝s/(v ego -v front )；
in the formula, v ego The speed of the main vehicle; v front The vehicle speed is the front vehicle speed; s is the initial relative position of the main vehicle and the front vehicle; TTC is the calculated collision time; f (f) fit The calculated fitness value is used for guiding the genetic algorithm searching process.
Preferably, in S5, the quality characteristics are established according to the operation results of the population individuals, and the individuals with poor quality, that is, the individuals with too low fitness value are deleted; a scene with too low fitness is defined as a bad individual, which is directly rejected in the field Jing Ku.
Preferably, in S1, the environment type includes a foggy environment, a windy environment, a rainy environment, a snowy environment;
the four environment types of a foggy environment, a windy environment, a rainy environment and a snowy environment are respectively classified into three grades: low, medium, high.
Preferably, the three levels of low, medium and high are divided according to the environment types, each level is combined with the optimal solution output in a crossing way, the fitness calculation results of the optimal solution output combined with the levels are added with 0.001, 0.0025 and 0.003 respectively, the final calculation result is obtained, the data with the fitness lower than 0.04 are removed, and the obtained data are output.
Compared with the prior art, the technical scheme provided by the invention has the following technical effects:
the automatic driving simulation scene expansion is carried out through the genetic algorithm, the optimal result can be rapidly screened, the optimal offspring is selected according to the authenticity index through the environment factor intersection, massive scene data are generated through iteration, and key characteristics of scene library expansibility, infinity, batch quantity, automation and the like are better met.
Detailed Description
The following description of the embodiments of the present invention will clearly and fully describe the technical solutions of the embodiments of the present invention, and 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.
Examples
The invention provides a technical scheme that: the automatic driving simulation scene library expansion method based on the GA genetic algorithm comprises the following steps:
s1, determining scene types and environment types as parameter spaces of a genetic algorithm; scene types include extreme hazards at the impact time scale, impact edges, ensure no impact, generate no simulation;
extremely dangerous, dangerous scenes of collision cannot be avoided;
collision edges, which can optimize an improved dangerous boundary scenario;
the method ensures that collision can not occur, and has low value for algorithm improvement in general safety scenes;
generating unreasonable scenes in which the motion does not accord with a physical rule or the simulation result is absolutely safe;
the environment type comprises a foggy environment, a windy environment, a rainy environment and a snowy environment;
the four environment types of a foggy environment, a windy environment, a rainy environment and a snowy environment are respectively classified into three grades: low, medium, high.
A foggy environment,
Low level fog environment: the visibility is more than or equal to 500 meters; medium mist environment: 300 meters < visibility < 500 meters; high level fog environment: the visibility is less than or equal to 300 meters;
has a wind environment,
Low wind environment: the wind speed is less than or equal to 6.0m/s; moderate wind environment: wind speed is less than 15.0m/s and less than 6.0m/s; high wind environment: the wind speed is more than or equal to 15.0m/s;
a rainy environment,
Low level rain environment: the rainfall is less than or equal to 15.0 mm after 24 hours; moderate rain environment: 15.0 mm < 24 hours rainfall < 30.0 mm; high level rain environment: the rainfall is more than or equal to 30.0 mm after 24 hours;
a snow environment,
Low-level snow environment: the visibility is more than or equal to 500 meters; moderate snow environment: 300 meters < visibility < 500 meters; high snow environment: the visibility is less than or equal to 300 meters;
s2, initializing, setting an evolution algebra counter t=0, setting a maximum evolution algebra T, and randomly generating M individuals as an initial population P (0);
s3, individual evaluation, namely calculating the fitness of each individual in the group P (t);
s4, selecting operation, and determining an operation formula; taking the collision time TTC as a key scene type selection index, wherein the TTC represents extreme danger at 0.7-0.9s, the TTC represents a collision edge at 0-0.7s, the TTC is larger than 0.9s, so that collision is avoided, and the TTC is negative and can not be simulated;
and establishing four stages of the fitness function according to the characteristics of the four collision scenes, wherein different numerical values are set to guide the scenes to converge towards dangerous boundaries in different stages.
The TTC calculation formula and fitness function formula are as follows:
TTC＝s/(v ego -v front )；
in the formula, v ego The speed of the main vehicle; v front The vehicle speed is the front vehicle speed; s is the initial relative position of the main vehicle and the front vehicle; TTC is the calculated collision time; f (f) fit The calculated fitness value is used for guiding the searching process of the genetic algorithm;
s5, performing cross operation, namely performing cross grouping on the screened high-quality individuals to generate scene data; establishing good and bad characteristics according to the running results of population individuals, and deleting bad individuals, namely individuals with too low fitness values; in order to avoid the bad individuals from influencing the population evolution process and interfering the subsequent genetic operator operation, a bad individual screening module is designed, a scene with low adaptability (below 0.04) is defined as a bad individual, the bad individual is directly rejected from the field Jing Ku, the search operator is prevented from searching for the value in the subsequent iteration process, and the local optimization capacity and the quick optimization capacity of the algorithm are improved;
s6, judging termination conditions: if t=t, outputting the individual with the best fitness obtained in the evolution process as an optimal solution, and terminating the calculation;
and S7, outputting and storing the optimal solutions, and cross-combining the environmental type influence with the output data of each optimal solution to form final scene data.
And dividing the low, medium and high levels according to the environment types, carrying out cross combination on each level and the optimal solution output, adding 0.001, 0.0025 and 0.003 to the fitness calculation result of the optimal solution output combined with the level, obtaining a final calculation result, removing data with fitness lower than 0.04, and outputting the obtained data.
It should also be noted that in the embodiments of the present invention, unless otherwise known, numerical parameters in the present specification and attached claims are approximations that may vary depending upon the desired properties sought to be obtained by the present invention. In particular, all numbers expressing dimensions, range conditions, and so forth, used in the specification and claims are to be understood as being modified in all instances by the term "about". In general, the meaning of expression is meant to include a variation of + -10% in some embodiments, a variation of + -5% in some embodiments, a variation of + -1% in some embodiments, and a variation of + -0.5% in some embodiments by a particular amount.
Those skilled in the art will appreciate that the features recited in the various embodiments of the invention and/or in the claims may be combined in various combinations and/or combinations, even if such combinations or combinations are not explicitly recited in the invention. In particular, the features recited in the various embodiments of the invention and/or in the claims can be combined in various combinations and/or combinations without departing from the spirit and teachings of the invention. All such combinations and/or combinations fall within the scope of the invention.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the invention thereto, but to limit the invention thereto, and any modifications, equivalents, improvements and equivalents thereof may be made without departing from the spirit and principles of the invention.
Claims (6)
1. The automatic driving simulation scene library expansion method based on the GA genetic algorithm is characterized by comprising the following steps of: the method comprises the following steps:
s1, determining scene types and environment types as parameter spaces of a genetic algorithm;
s2, initializing, setting an evolution algebra counter t=0, setting a maximum evolution algebra T, and randomly generating M individuals as an initial population P (0);
s3, individual evaluation, namely calculating the fitness of each individual in the group P (t);
s4, selecting operation, and determining an operation formula;
s5, performing cross operation, namely performing cross grouping on the screened high-quality individuals to generate scene data;
s6, judging termination conditions: if t=t, outputting the individual with the best fitness obtained in the evolution process as an optimal solution, and terminating the calculation;
and S7, outputting and storing the optimal solutions, and cross-combining the environmental type influence with the output data of each optimal solution to form final scene data.
2. The automated driving simulation scene library extension method based on GA genetic algorithm of claim 1, wherein: in S1, scene types include extreme danger, collision edges, ensuring no collisions, generating an incapacitation, at the collision time division level.
3. The automated driving simulation scene library extension method based on GA genetic algorithm according to claim 2, wherein: taking the collision time TTC as a key scene type selection index, wherein the TTC represents extreme danger at 0.7-0.9s, the TTC represents a collision edge at 0-0.7s, the TTC is larger than 0.9s, so that collision is avoided, and the TTC is negative and can not be simulated;
the TTC calculation formula and fitness function formula are as follows:
TTC＝s/(v eg o-vf r o n t)；
in the formula, v ego The speed of the main vehicle; v front The vehicle speed is the front vehicle speed; s is the initial relative position of the main vehicle and the front vehicle; TTC is the calculated collision time; f (f) fit The calculated fitness value is used for guiding the genetic algorithm searching process.
4. The automated driving simulation scene library extension method based on GA genetic algorithm of claim 3, wherein: in S5, establishing good and bad characteristics according to the operation result of the population individuals, and deleting the poor individuals, namely the individuals with too low fitness values; a scene with too low fitness is defined as a bad individual, which is directly rejected in the field Jing Ku.
5. The automated driving simulation scene library extension method based on GA genetic algorithm of claim 1, wherein: in S1, the environment type includes a foggy environment, a windy environment, a rainy environment, and a snowy environment;
the four environment types of a foggy environment, a windy environment, a rainy environment and a snowy environment are respectively classified into three grades: low, medium, high.
6. The GA genetic algorithm-based automatic driving simulation scene library extension method of claim 5, wherein: and dividing the low, medium and high levels according to the environment types, carrying out cross combination on each level and the optimal solution output, adding 0.001, 0.0025 and 0.003 to the fitness calculation result of the optimal solution output combined with the level, obtaining a final calculation result, removing data with fitness lower than 0.04, and outputting the obtained data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310255344.6A CN116362119A (en) | 2023-03-16 | 2023-03-16 | Automatic driving simulation scene library expansion method based on GA genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310255344.6A CN116362119A (en) | 2023-03-16 | 2023-03-16 | Automatic driving simulation scene library expansion method based on GA genetic algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116362119A true CN116362119A (en) | 2023-06-30 |
Family
ID=86912949
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310255344.6A Pending CN116362119A (en) | 2023-03-16 | 2023-03-16 | Automatic driving simulation scene library expansion method based on GA genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116362119A (en) |
-
2023
- 2023-03-16 CN CN202310255344.6A patent/CN116362119A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106127121B (en) | A kind of built-up areas intelligence extracting method based on nighttime light data | |
| CN111366931B (en) | Cloud precipitation refined inversion method based on cloud radar | |
| CN110175611A (en) | Defence method and device towards Vehicle License Plate Recognition System black box physical attacks model | |
| CN111339478B (en) | Meteorological data quality assessment method based on improved fuzzy analytic hierarchy process | |
| CN111191731A (en) | Data processing method and device, storage medium and electronic equipment | |
| CN110852358A (en) | Vehicle type distinguishing method based on deep learning | |
| CN111695640A (en) | Foundation cloud picture recognition model training method and foundation cloud picture recognition method | |
| CN111507824A (en) | Wind control model mold-entering variable minimum entropy box separation method | |
| CN116186611A (en) | Unbalanced data classification method, device, terminal equipment and medium | |
| CN112141098A (en) | Obstacle avoidance decision method and device for intelligent driving automobile | |
| CN107516061A (en) | A kind of image classification method and system | |
| CN116362119A (en) | Automatic driving simulation scene library expansion method based on GA genetic algorithm | |
| CN112287468B (en) | Ship collision risk degree judging method and system | |
| CN110414585B (en) | Real-time particulate matter detection method based on improved embedded platform | |
| CN116993548A (en) | Incremental learning-based education training institution credit assessment method and system for LightGBM-SVM | |
| CN110706004A (en) | Farmland heavy metal pollutant tracing method based on hierarchical clustering | |
| CN116127360A (en) | Driving style classification method based on image recognition and TOPSIS comprehensive evaluation | |
| CN115761259A (en) | Kitchen waste target detection method and system based on class balance loss function | |
| CN107274357A (en) | A kind of optimal gray level image enhancing processing system of parameter | |
| CN115205573A (en) | Image processing method, device and equipment | |
| CN113988488A (en) | Method for predicting ETC passing probability of vehicle by multiple factors | |
| CN116881854B (en) | XGBoost-fused time sequence prediction method for calculating feature weights | |
| CN111325242A (en) | Image classification method, terminal and computer storage medium | |
| CN113361655B (en) | Differential fiber classification method based on residual error network and characteristic difference fitting | |
| CN113688950B (en) | Multi-target feature selection method, device and storage medium for image classification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |