An automatic driving model training method
The autonomous driving simulation training system solves the problem that existing technologies cannot effectively simulate traffic flow and emergencies, thereby improving the actual driving safety and response capabilities of autonomous vehicles.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANTONG UNIV
- Filing Date
- 2026-01-13
- Publication Date
- 2026-06-09
Smart Images
Figure CN122172639A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of autonomous driving simulation training technology, and more specifically to an autonomous driving model training method. Background Technology
[0002] Autonomous vehicles, also known as driverless cars, computer-driven cars, or wheeled mobile robots, are intelligent vehicles that achieve driverless operation through an autonomous driving system. They are commonly found in new energy vehicles or hybrid vehicles. The autonomous driving system is a cutting-edge technology that relies on computer and artificial intelligence technology to complete a complete, safe, and efficient driving process without human intervention. With the rapid development of deep learning and in-depth research on artificial intelligence, autonomous driving simulation training has also become a major research direction.
[0003] Existing autonomous driving simulation training involves the autonomous driving system driving according to predetermined simulated road data. This simulated road data consists of common road data and does not include unexpected situations that occur in actual road driving. Furthermore, the probability of traffic flow and unexpected events varies at different times in real-world driving. Therefore, it is impossible to test whether the autonomous driving system can drive normally under different traffic volumes and unexpected events, or whether its emergency avoidance functions can be used properly during actual operation. This hinders subsequent improvements to these aspects. In summary, existing autonomous driving model training lacks the ability to simulate different traffic volumes and unexpected events, thus reducing the safety of autonomous vehicles in actual driving. Summary of the Invention
[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide an autonomous driving model training method to solve the technical problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: an autonomous driving model training method, comprising an autonomous driving simulation training system, wherein the autonomous driving simulation training system further comprises a control center, an identification unit, a scoring unit, a decision-making module, a road condition information collection unit, a storage unit, and an emergency incident unit, wherein the identification unit, the scoring unit, the decision-making module, the road condition information collection unit, the storage unit, and the emergency incident unit are controlled by the control center;
[0006] The identification unit includes a face recognition module, a voice recognition module, a fingerprint recognition module, and an electronic key. The control center consists of a processor. Data transmission between the control center, identification unit, scoring unit, decision-making module, road condition information collection unit, storage unit, and emergency unit is carried out through both wired data transmission and 5G signal transmission. Furthermore, the data transmission between the control center, identification unit, scoring unit, decision-making module, road condition information collection unit, storage unit, and emergency unit can be encrypted using TLS algorithm encryption, FTPS encryption technology, firewall friendliness, and encryption certificate configuration.
[0007] The road condition information collection unit includes three types of road data: urban roads, rural roads, and special roads. The special roads include urban elevated roads, expressways, and national highways. The road condition information collection unit mainly collects road element point data for urban roads, rural roads, and special roads through remote sensing road extraction algorithms. The remote sensing road extraction algorithms mainly use satellite remote sensing technology, UAV aerial photography technology, and vehicles with panoramic cameras for collection. At the same time, during the single-road collection process, the road condition information unit also collects the corresponding traffic law information of the collected road and stores and manages the collected data in a categorized manner.
[0008] In a preferred embodiment, the face recognition module mainly collects face data through one of four methods: Fisherfaces facial recognition algorithm, Haar Cascade, 3D recognition, and skin texture analysis. At the same time, the face recognition process requires the following preprocessing: image enhancement and face alignment processing, and the face recognition module needs to perform dynamic recognition during the face recognition process.
[0009] The face recognition process of the face recognition module is as follows:
[0010] S1. Image acquisition: Divide the face into n regions and use multiple sets of acquisition devices to acquire corresponding data from the n regions. Image acquisition includes static image acquisition, and multiple sets of static image acquisition are combined to form dynamic image acquisition.
[0011] S2. Real-time facial recognition is required during the data collection process;
[0012] S3. Extract features from the collected human images and perform 3D modeling.
[0013] S4. Compare and identify the modeled 3D image with the facial data from the facial recognition module.
[0014] In a preferred embodiment, the specific operation process of the speech recognition module is as follows: collecting the user's speech data; extracting features from the collected speech data; training an acoustic model using the extracted speech features and forming a contrastive acoustic model; when the user performs speech recognition, the speech recognition module extracts the user's speech features and forms an acoustic model, which is then compared with the contrastive acoustic model to determine whether the user is authorized.
[0015] The speech data collected by the speech recognition module during the establishment of the comparative acoustic model needs to undergo audio data preprocessing such as filtering and frame segmentation.
[0016] In a preferred embodiment, the specific operation process of the fingerprint recognition module is as follows: Using bio-radio frequency fingerprint recognition technology, a small radio frequency signal is emitted by the sensor itself to penetrate the epidermis of the finger and detect the patterns in the inner layer to obtain the optimal fingerprint image. The acquired fingerprint image is normalized to form a successfully focused comparison two-dimensional image. The normalized fingerprint image needs to undergo image enhancement processing to make the fingerprint ridges clearer, connect broken ridges, and maintain the original structure. The collected fingerprint image data is stored. During user identification, the user's fingerprint features are extracted, normalized, and binarized to form a new two-dimensional image, which is then compared with the comparison two-dimensional image to facilitate user identification.
[0017] In a preferred embodiment, the time module service collects corresponding data based on the road condition information collection unit. It collects pedestrian and vehicle traffic data for three types of roads—urban roads, rural roads, and special roads—at different times throughout the day and during special times on holidays. The time module collection device is provided by the traffic information collection system.
[0018] In a preferred embodiment, the sudden accident unit refers to minor traffic accidents, major traffic accidents, non-motorized vehicle traffic accidents, and accidents caused by pedestrians not complying with traffic rules encountered during driving. The sudden accident unit appears randomly during the autonomous driving simulation training, and the probability of the sudden accident unit appearing randomly corresponds to the probability of the accident occurring in the road condition information collection unit.
[0019] In a preferred embodiment, the storage unit includes a relational database and a non-relational database, and the data retrieval of the storage unit is protected by modern cryptographic algorithms and dual strong authentication. The data inside the storage unit is backed up through the cloud.
[0020] In a preferred embodiment, the scoring unit determines whether the autonomous vehicle violates traffic rules and the overall speed of the vehicle in the autonomous driving simulation training system, and the vehicle speed is compared with the speed collected by the road condition information collection unit.
[0021] In a preferred embodiment, the decision module determines whether a violation of traffic rules occurs in the scoring unit and compares the overall vehicle speed. When a violation of traffic rules occurs during the simulation of an autonomous vehicle in the automatic simulation training system, the data provided by the road condition information collection unit, time module, and emergency accident unit that violates traffic rules can be marked to remind staff to modify and update the data. When there is a 10%-40% difference between the simulated vehicle speed and the speed collected by the road condition information collection unit, the data provided by the information collection unit, time module, and emergency accident unit within that range can be marked to remind staff to modify and update the data.
[0022] In a preferred embodiment, the specific operation flow of the autonomous driving simulation training system is as follows:
[0023] S1. The first step for a vehicle to enter the training module is to identify the driver. This can be done using a facial recognition module, a voice recognition module, a fingerprint recognition module, or an electronic key.
[0024] S2. The current vehicle's autonomous driving system is imported into the autonomous driving simulation training system. The control center inputs the road condition information collection unit data into the autonomous driving simulation training system to carry out vehicle autonomous driving simulation training.
[0025] S3, the road condition information collection unit can provide three types of road data for autonomous driving simulation training: urban roads, rural roads and special roads. When simulating the above three types of road data, different time periods of traffic flow and pedestrian flow data can be selected.
[0026] S4. When the current vehicle is conducting autonomous driving simulation training, the emergency accident unit provides emergency accident simulation data and records the vehicle's driving status during road driving simulation training.
[0027] S5. When the current vehicle is conducting autonomous driving simulation training, the scoring unit scores the vehicle based on whether there are collisions, traffic violations, and vehicle driving efficiency during the simulation training. When collisions or traffic violations occur during the vehicle simulation, they are recorded and marked by the decision module to remind staff to make corresponding corrections and updates to the current vehicle's autonomous driving system.
[0028] The technical effects and advantages of this invention are as follows:
[0029] 1. This invention, by incorporating an autonomous driving simulation training system, facilitates the collection of pedestrian and vehicle traffic data from the road condition information collection unit across three types of roads—urban roads, rural roads, and special roads—at different times throughout the day and during special holiday periods. This allows for the incorporation of different traffic and pedestrian traffic data from different road conditions at different times into the autonomous driving simulation training. This enables the autonomous driving vehicles to conduct simulation experiments that more closely resemble real-life situations, facilitating updates and improvements by staff to address potential problems in actual operation and thereby enhancing the safety of autonomous driving in real-world use.
[0030] 2. By incorporating an autonomous driving simulation training system, this invention facilitates the simulation of autonomous vehicles' responses to minor traffic accidents, major traffic accidents, non-motorized vehicle traffic accidents, and accidents caused by pedestrians violating traffic rules through a sudden accident module. This enhances the ability of autonomous vehicles to take appropriate countermeasures when encountering sudden accidents on the road, thereby increasing the safety of autonomous driving in actual use. Attached Figure Description
[0031] Figure 1 This is a schematic diagram of the overall structure of the autonomous driving simulation training system of the present invention.
[0032] Figure 2 This is a schematic diagram of the overall process of the autonomous driving simulation training system of the present invention. Detailed Implementation
[0033] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The autonomous driving model training method involved in the present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0034] refer to Figure 1 The present invention provides an autonomous driving model training method, including an autonomous driving simulation training system. The autonomous driving simulation training system further comprises a control center, an identification unit, a scoring unit, a decision-making module, a road condition information collection unit, a storage unit, and an emergency accident unit. The identification unit, scoring unit, decision-making module, road condition information collection unit, storage unit, and emergency accident unit are controlled by the control center.
[0035] The identification unit includes a face recognition module, a voice recognition module, a fingerprint recognition module, and an electronic key. The control center consists of a processor. Data transmission between the control center, identification unit, scoring unit, decision-making module, road condition information collection unit, storage unit, and emergency unit is carried out through both wired data transmission and 5G signal transmission. Furthermore, the data transmission between the control center, identification unit, scoring unit, decision-making module, road condition information collection unit, storage unit, and emergency unit can be encrypted using TLS algorithm encryption, FTPS encryption technology, firewall friendliness, and encryption certificate configuration.
[0036] The road condition information collection unit includes three types of road data: urban roads, rural roads, and special roads. Special roads include urban elevated roads, expressways, and national highways. The road condition information collection unit mainly collects road element point data for urban roads, rural roads, and special roads through remote sensing road extraction algorithms. The remote sensing road extraction algorithms mainly use satellite remote sensing technology, UAV aerial photography technology, and vehicles with panoramic cameras for collection. At the same time, during the single-road collection process, the road condition information unit also collects the corresponding traffic law information of the collected road and stores and manages the collected data in a categorized manner.
[0037] Reference Figure 1 As shown, the present invention provides an autonomous driving model training method. The face recognition module mainly collects face data through one of four methods: Fisherfaces face recognition algorithm, Haar Cascade, three-dimensional recognition, and skin texture analysis. At the same time, the face recognition process requires the following preprocessing: image enhancement and face alignment processing. The face recognition module also needs to perform dynamic recognition during the face recognition process.
[0038] The face recognition process of the face recognition module is as follows:
[0039] S1. Image acquisition: Divide the face into n regions and use multiple sets of acquisition devices to acquire corresponding data from the n regions. Image acquisition includes static image acquisition, and multiple sets of static image acquisition are combined to form dynamic image acquisition.
[0040] S2. Real-time facial recognition is required during the data collection process;
[0041] S3. Extract features from the collected human images and perform 3D modeling.
[0042] S4. Compare and identify the modeled 3D image with the facial data from the facial recognition module.
[0043] In this embodiment, the TLS algorithm encryption uses Rayspeed AES-256 financial-grade encryption strength to protect user data privacy and security; FTPS encryption technology adds SSL security to the FTP protocol and data channel; firewall friendly: the Rayspeed transmission protocol only needs to open one UDP port to complete communication, which is more secure than opening a large number of firewall network ports; encryption certificate configuration: supports configuring confidential certificates to make service access more secure;
[0044] Furthermore, during data transmission, regular CVE vulnerability risk database scans are required to address risky code vulnerabilities and employ high-performance SSL VPN encryption to provide secure access services for users in various scenarios. Additionally, the automatic simulation training system of this application requires login to a corresponding account, and the login account must employ a two-factor authentication system, support multiple password authentication methods such as USBKey and terminal hardware ID binding, and the password stored by the user in the data is encrypted using a high-strength encryption algorithm of AES-256 + random salt, so even developers cannot recover the original password from the stored ciphertext.
[0045] Reference Figure 1 As shown, the present invention provides an autonomous driving model training method. The specific operation process of the speech recognition module is as follows: collecting the user's speech data; extracting features from the collected speech data; training an acoustic model using the extracted speech features and forming a contrastive acoustic model; when the user performs speech recognition, the speech recognition module extracts the user's speech features and forms an acoustic model, which is then compared with the contrastive acoustic model to determine whether the user is authorized.
[0046] The speech data collected by the speech recognition module when building the comparative acoustic model needs to undergo audio data preprocessing such as filtering and frame segmentation.
[0047] In this embodiment of the application, image enhancement refers to supplementing the illumination of images with low brightness, and using super-resolution, noise reduction, and motion blur removal to enhance the image recognition of low-quality images. By providing an image enhancement method, face data can be collected, and the face recognition rate can be enhanced, making it easier for users to quickly perform face recognition.
[0048] Simultaneously, super-resolution can be used during the process of acquiring facial data, especially for low-resolution images, where facial attributes can be utilized to enhance the realism of high-resolution facial images.
[0049] In this application example, the filtering is as follows: the silence at the beginning and end is removed to reduce the interference to subsequent steps; the sound is framed: the collected sound data is cut into a small segment, each segment is called a frame, which is implemented using a moving window function. It is not a simple cutting, and there is generally overlap between the frames.
[0050] Reference Figure 1 As shown, this invention provides an autonomous driving model training method. The specific operation process of the fingerprint recognition module is as follows: Using bio-radio frequency fingerprint recognition technology, a small amount of radio frequency signal is emitted by the sensor itself to penetrate the epidermis of the finger and detect the inner ridges to obtain the best fingerprint image. The collected fingerprint image is normalized to form a successfully focused comparison two-dimensional image. The normalized fingerprint image needs to undergo image enhancement processing to make the fingerprint ridges clearer, connect broken ridges, and maintain the original structure. The collected fingerprint image data is stored. During user identification, the user's fingerprint features are extracted, normalized, and binarized to form a new two-dimensional image, which is then compared with the comparison two-dimensional image to facilitate user identification.
[0051] Reference Figure 1 As shown, the present invention provides an autonomous driving model training method. The time module service collects corresponding data based on the road condition information collection unit. It collects pedestrian and vehicle traffic data for three types of roads—urban roads, rural roads, and special roads—at different times throughout the day and during special times on holidays. The time module collection equipment is provided through a traffic information collection system.
[0052] Reference Figure 1 As shown, the present invention provides an autonomous driving model training method. The sudden accident unit refers to minor traffic accidents, major traffic accidents, non-motorized vehicle traffic accidents, and accidents caused by pedestrians not complying with traffic rules encountered during driving. The sudden accident unit appears randomly during the autonomous driving simulation training process, and the probability of random occurrence in the sudden accident unit corresponds to the probability of accident occurrence in the road condition information collection unit.
[0053] Reference Figure 1 As shown, the present invention provides an autonomous driving model training method. The storage unit includes a relational database and a non-relational database. When the storage unit extracts data, it needs to be protected by modern cryptographic algorithms and dual strong authentication. The data inside the storage unit will be backed up through the cloud.
[0054] Reference Figure 1As shown, the present invention provides an autonomous driving model training method. The scoring unit determines whether the autonomous vehicle violates road traffic rules and the overall driving speed of the vehicle in the autonomous driving simulation training system. The vehicle driving speed is compared with the speed collected by the road condition information collection unit.
[0055] Reference Figure 1 As shown, this invention provides an autonomous driving model training method. The decision module determines whether a violation of road traffic rules occurs in the scoring unit and compares the overall vehicle speed. When a violation of road traffic rules occurs during the simulation of an autonomous driving vehicle in the automatic simulation training system, the data provided by the road condition information collection unit, time module, and emergency accident unit that violates the traffic rules can be marked to remind staff to modify and update the data. When there is a 10%-40% difference between the simulated vehicle speed and the speed collected by the road condition information collection unit, the data provided by the information collection unit, time module, and emergency accident unit within that range can be marked to remind staff to modify and update the data.
[0056] Reference Figure 2 As shown, this invention provides an autonomous driving model training method, and the specific operation flow of the autonomous driving simulation training system is as follows:
[0057] S1. The first step for a vehicle to enter the training module is to identify the driver. This can be done using a facial recognition module, a voice recognition module, a fingerprint recognition module, or an electronic key.
[0058] S2. The current vehicle's autonomous driving system is imported into the autonomous driving simulation training system. The control center inputs the road condition information collection unit data into the autonomous driving simulation training system to carry out vehicle autonomous driving simulation training.
[0059] S3, the road condition information collection unit can provide three types of road data for autonomous driving simulation training: urban roads, rural roads and special roads. When simulating the above three types of road data, different time periods of traffic flow and pedestrian flow data can be selected.
[0060] S4. When the current vehicle is conducting autonomous driving simulation training, the emergency accident unit provides emergency accident simulation data and records the vehicle's driving status during road driving simulation training.
[0061] S5. When the current vehicle is conducting autonomous driving simulation training, the scoring unit scores the vehicle based on whether there are collisions, traffic violations, and vehicle driving efficiency during the simulation training. When collisions or traffic violations occur during the vehicle simulation, they are recorded and marked by the decision module to remind staff to make corresponding corrections and updates to the current vehicle's autonomous driving system.
[0062] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired connection. A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium. A semiconductor medium can be a solid-state drive.
[0063] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0064] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0065] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0066] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0067] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
[0068] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for training an autonomous driving model, characterized in that: The system includes an autonomous driving simulation training system, which further comprises a control center, an identification unit, a scoring unit, a decision-making module, a road condition information collection unit, a storage unit, and an emergency response unit. The identification unit, scoring unit, decision-making module, road condition information collection unit, storage unit, and emergency response unit are controlled by the control center. The identification unit includes a face recognition module, a voice recognition module, a fingerprint recognition module, and an electronic key. The control center consists of a processor. Data transmission between the control center, identification unit, scoring unit, decision-making module, road condition information collection unit, storage unit, and emergency unit is carried out through both wired data transmission and 5G signal transmission. Furthermore, the data transmission between the control center, identification unit, scoring unit, decision-making module, road condition information collection unit, storage unit, and emergency unit can be encrypted using TLS algorithm encryption, FTPS encryption technology, firewall friendliness, and encryption certificate configuration. The road condition information collection unit includes three types of road data: urban roads, rural roads, and special roads. The special roads include urban elevated roads, expressways, and national highways. The road condition information collection unit mainly collects road element point data for urban roads, rural roads, and special roads through remote sensing road extraction algorithms. The remote sensing road extraction algorithms mainly use satellite remote sensing technology, UAV aerial photography technology, and vehicles with panoramic cameras for collection. At the same time, during the single-road collection process, the road condition information unit also collects the corresponding traffic law information of the collected road and stores and manages the collected data in a categorized manner.
2. The autonomous driving model training method according to claim 1, characterized in that: The face recognition module mainly collects face data through one of four methods: Fisherfaces facial recognition algorithm, Haar Cascade, 3D recognition, and skin texture analysis. At the same time, the face recognition process requires the following preprocessing: image enhancement and face alignment processing, and the face recognition module needs to perform dynamic recognition during the face recognition process. The face recognition process of the face recognition module is as follows: S1. Image acquisition: Divide the face into n regions and use multiple sets of acquisition devices to acquire corresponding data from the n regions. Image acquisition includes static image acquisition, and multiple sets of static image acquisition are combined to form dynamic image acquisition. S2. Real-time facial recognition is required during the data collection process; S3. Extract features from the collected human images and perform 3D modeling. S4. Compare and identify the modeled 3D image with the facial data from the facial recognition module.
3. The autonomous driving model training method according to claim 1, characterized in that: The specific operation process of the speech recognition module is as follows: collecting the user's speech data; extracting features from the collected speech data; training an acoustic model using the extracted speech features and forming a contrastive acoustic model; when the user performs speech recognition, the speech recognition module extracts the user's speech features and forms an acoustic model, which is then compared with the contrastive acoustic model to determine whether the user is authorized. The speech data collected by the speech recognition module during the establishment of the comparative acoustic model needs to undergo audio data preprocessing such as filtering and frame segmentation.
4. The autonomous driving model training method according to claim 1, characterized in that: The specific operation process of the fingerprint recognition module is as follows: Using bio-radio frequency fingerprint recognition technology, the sensor emits a small amount of radio frequency signal to penetrate the epidermis of the finger and detect the patterns in the inner layer to obtain the optimal fingerprint image. The acquired fingerprint image is normalized to form a successfully focused comparison two-dimensional image. The normalized fingerprint image needs to undergo image enhancement processing to make the fingerprint ridges clearer, connect broken ridges, and maintain the original structure. The collected fingerprint image data is stored. During user identification, the user's fingerprint features are extracted, normalized, and binarized to form a new two-dimensional image, which is then compared with the comparison two-dimensional image to facilitate user identification.
5. The autonomous driving model training method according to claim 1, characterized in that: The time module service collects data based on the road condition information collection unit. It collects pedestrian and vehicle traffic data for three types of roads—urban roads, rural roads, and special roads—at different times of the day and during special times of holidays. The time module's collection equipment is provided through the traffic information collection system.
6. The autonomous driving model training method according to claim 1, characterized in that: The sudden accident unit refers to minor traffic accidents, major traffic accidents, non-motorized vehicle traffic accidents, and accidents caused by pedestrians not complying with traffic rules encountered during the driving process. The sudden accident unit appears randomly during the autonomous driving simulation training, and the probability of the sudden accident unit appearing randomly corresponds to the probability of the accident occurring in the road condition information collection unit.
7. The autonomous driving model training method according to claim 1, characterized in that: The storage unit includes relational databases and non-relational databases. When the storage unit extracts data, it needs to be protected by modern cryptographic algorithms and dual strong authentication. The data inside the storage unit will be backed up through the cloud.
8. The autonomous driving model training method according to claim 1, characterized in that: The scoring unit determines whether the autonomous vehicle violates traffic rules and the overall speed of the vehicle in the autonomous driving simulation training system, and the vehicle speed is compared with the speed collected by the road condition information collection unit.
9. The autonomous driving model training method according to claim 8, characterized in that: The decision-making module determines whether a violation of traffic rules occurs in the scoring unit and compares the overall vehicle speed. When a violation of traffic rules occurs during the simulation of an autonomous vehicle in the automatic simulation training system, the data provided by the road condition information collection unit, time module, and emergency unit that violates traffic rules can be marked to remind staff to modify and update the data. When there is a 10%-40% difference between the simulated vehicle speed and the speed collected by the road condition information collection unit, the data provided by the information collection unit, time module, and emergency unit within that range can be marked to remind staff to modify and update the data.
10. The autonomous driving model training method according to claim 1, characterized in that: The specific operation procedure of the autonomous driving simulation training system is as follows: S1. The first step for a vehicle to enter the training module is to identify the driver. This can be done using a facial recognition module, a voice recognition module, a fingerprint recognition module, or an electronic key. S2. The current vehicle's autonomous driving system is imported into the autonomous driving simulation training system. The control center inputs the road condition information collection unit data into the autonomous driving simulation training system to carry out vehicle autonomous driving simulation training. S3, the road condition information collection unit can provide three types of road data for autonomous driving simulation training: urban roads, rural roads and special roads. When simulating the above three types of road data, different time periods of traffic flow and pedestrian flow data can be selected. S4. When the current vehicle is conducting autonomous driving simulation training, the emergency accident unit provides emergency accident simulation data and records the vehicle's driving status during road driving simulation training. S5. When the current vehicle is conducting autonomous driving simulation training, the scoring unit scores the vehicle based on whether there are collisions, traffic violations, and vehicle driving efficiency during the simulation training. When collisions or traffic violations occur during the vehicle simulation, they are recorded and marked by the decision module to remind staff to make corresponding corrections and updates to the current vehicle's autonomous driving system.