Training method of automatic driving model and related equipment

Abstract

The invention provides a training method of an automatic driving model and related equipment. The method comprises the following steps: constructing a perception body for simulating human perception and evaluation; simulating a human growth process based on different driving scenes, and performing perception integration training on the perception body to obtain an initial automatic driving model; optimizing the initial automatic driving model based on the individual application data of the vehicle end and the set application data of the platform to obtain an automatic driving model; the platform comprises a plurality of vehicle ends.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular, to a training method for an autonomous driving model, an application method for an autonomous driving model, a training device for an autonomous driving model, an electronic device, a vehicle, and a computer-readable storage medium. Background Art

[0002] With the rapid progress of artificial intelligence technology, the research focus has gradually shifted to endowing machines with human-like learning and adaptation capabilities. Although modern autonomous driving technology has achieved remarkable results in simplified road conditions, it still faces many challenges when dealing with complex and changing real-world environments. The training of traditional autonomous driving systems highly depends on a large amount of labeled data and complex preset rules, which limits the flexibility of the system and its adaptive ability to the environment. Summary of the Invention

[0003] The present invention aims to at least solve one of the technical problems in the related art to some extent. For this reason, the first object of the present invention is to propose a training method for an autonomous driving model, which simulates the human growth process, draws on the concept of sensory integration training, and guides the learning process of the autonomous driving model, enabling it to more intelligently adapt to and process various complex driving scenarios.

[0004] The second object of the present invention is to propose an application method for an autonomous driving model.

[0005] The third object of the present invention is to propose a training device for an autonomous driving model

[0006] The fourth object of the present invention is to propose an electronic device.

[0007] The fifth object of the present invention is to propose a vehicle.

[0008] The sixth object of the present invention is to propose a computer-readable storage medium.

[0009] To achieve the above object, an embodiment of the first aspect of the present invention proposes a method for training an autonomous driving model. First, a perceptual body for simulating human perception and evaluation is constructed. Then, based on different driving scenarios to simulate the human growth process, perceptual integration training is performed on the perceptual body to obtain an initial autonomous driving model. Finally, the initial autonomous driving model is optimized based on the individual application data of the vehicle terminal and the collective application data of the platform to obtain the autonomous driving model; wherein the platform includes multiple vehicle terminals. By simulating the human growth process and drawing on the concept of perceptual integration training, the learning process of the autonomous driving model is guided, enabling it to more intelligently adapt to and handle various complex driving scenarios, not only significantly improving the flexibility and adaptability of the autonomous driving system, but also reducing the dependence on a large amount of labeled data and complex preset rules, thus opening up a new path for the development of autonomous driving technology.

[0010] In addition, the method for training an autonomous driving model according to the above embodiment of the present invention may further have the following additional technical features:

[0011] As an optional embodiment, constructing a perceptual body for simulating human perception and evaluation includes:

[0012] Obtain the comfort perception data of the driver for longitudinal and lateral acceleration and deceleration values, as well as the reaction rules for driving dangerous situations;

[0013] Combine the processing rules for abnormal situations of the vehicle to construct a pain mechanism similar to that of humans for the perceptual body;

[0014] Integrate the comfort perception data, reaction rules, and pain mechanism into an intelligent body to obtain the perceptual body.

[0015] As an optional embodiment, based on different driving scenarios to simulate the human growth process, performing perceptual integration training on the perceptual body to obtain an initial autonomous driving model includes:

[0016] Design at least one set of training systems from simple to complex and from low speed to high speed;

[0017] Based on the training system, train the ability of the perceptual body in sensory information processing to obtain the initial autonomous driving model.

[0018] As an optional embodiment, the training system includes at least one of safety collision risk identification and control, single-lane control, multi-lane control, and urban road U-turn control.

[0019] As an optional embodiment, the method further includes:

[0020] Train the perceptual body in the order from simple to complex and from low speed to high speed.

[0021] As an optional embodiment, the individual application data includes the driver's feedback; the aggregated application data includes the characteristics of different vehicles;

[0022] Optimizing the initial autonomous driving model based on the individual application data at the vehicle end and the aggregated application data of the platform to obtain an autonomous driving model, including:

[0023] Establish a driving HMI two-way interaction mechanism, collect the driver's feedback, and input the feedback into the initial autonomous driving model for learning and optimization;

[0024] Adjust the parameters of the initial autonomous driving model according to the characteristics of different vehicles to obtain an autonomous driving model.

[0025] As an optional embodiment, the method further includes:

[0026] Extract high-quality feedback with a qualified feedback accuracy rate from the collected driver feedback, and transmit the high-quality feedback to the platform so that the platform can optimize the aggregated application data based on the high-quality feedback.

[0027] To achieve the above object, an embodiment of the second aspect of the present invention proposes an application method of an autonomous driving model. The application method of the autonomous driving model includes controlling a vehicle to perform autonomous driving based on the autonomous driving model obtained by the training method of the autonomous driving model as described above.

[0028] To achieve the above object, an embodiment of the third aspect of the present invention proposes a training device for an autonomous driving model. The training device for the autonomous driving model includes:

[0029] A construction module configured to construct a perceptual body for simulating human perception and evaluation;

[0030] A training module configured to perform perceptual integration training on the perceptual body based on different driving scenarios to simulate the human growth process, and obtain an initial autonomous driving model;

[0031] A verification module configured to optimize the initial autonomous driving model based on the individual application data at the vehicle end and the aggregated application data of the platform to obtain an autonomous driving model; wherein the platform includes multiple vehicle ends.

[0032] To achieve the above object, an embodiment of the fourth aspect of the present invention proposes an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor is characterized in that when the processor executes the program, it implements the training method of the autonomous driving model as described above.

[0033] To achieve the above object, a vehicle according to an embodiment of the fifth aspect of the present invention includes the above-mentioned electronic device. The memory of the electronic device stores a program or instruction that can run on a processor. When the program or instruction is executed by the processor, the steps of the training method of the above-mentioned autonomous driving model are implemented.

[0034] To achieve the above object, a computer-readable storage medium according to an embodiment of the sixth aspect of the present invention stores computer instructions, and the computer instructions are used to cause the computer to execute the training method of the above-mentioned autonomous driving model.

[0035] According to the training method of the autonomous driving model and related devices according to the embodiments of the present invention, first, a perceptual body for simulating human perception and evaluation is constructed. Then, based on different driving scenarios, the process of human growth is simulated, and the perceptual body is subjected to perceptual integration training to obtain an initial autonomous driving model. Finally, the initial autonomous driving model is optimized based on the individual application data of the vehicle end and the collective application data of the platform to obtain an autonomous driving model; wherein the platform includes multiple vehicle ends. By simulating the process of human growth and drawing on the concept of perceptual integration training, the learning process of the autonomous driving model is guided, enabling it to more intelligently adapt to and process various complex driving scenarios, not only significantly improving the flexibility and adaptability of the autonomous driving system, but also reducing the dependence on a large amount of labeled data and complex preset rules, thus opening up a new path for the development of autonomous driving technology.

[0036] The additional aspects and advantages of the present invention will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] To more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only those of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0038] Figure 1 It is a schematic flowchart of the training method of the autonomous driving model provided by the embodiment of the present invention.

[0039] Figure 2 It is a schematic flowchart of constructing a perceptual body for simulating human perception and evaluation provided by the embodiment of the present invention.

[0040] Figure 3 It is a schematic flowchart of the perceptual integration training provided by the embodiment of the present invention.

[0041] Figure 4A schematic diagram of a training device for an autonomous driving model provided in an embodiment of the present invention.

[0042] Figure 5 A more specific schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0044] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0045] As mentioned in the background technology section, with the rapid advancement of artificial intelligence (AI) technology, researchers are working to give machines the ability to learn and adapt similar to humans. In the cutting-edge field of autonomous driving technology, although remarkable achievements have been made under simplified road conditions, there are still many challenges in the face of the complexity and variability of the real world. The training of traditional autonomous driving systems is highly dependent on a large amount of labeled data and complex rules, which not only limits the flexibility of the system, but also affects its adaptive ability.

[0046] At present, the development of AI mainly revolves around the three core elements of data, algorithms and computing power. For connected vehicles, electrification is no longer a problem, and the focus of intelligent development has shifted to building a model training and optimization system based on data closed-loop capabilities. Although end-to-end learning methods are gradually emerging, most automakers and autonomous driving algorithm providers still rely mainly on modular end-to-end models. In cutting-edge research, only a few companies have begun to explore the possibility of building a "oneworld" world model, and this field still needs further breakthroughs.

[0047] In AI technology, neural networks based on brain-like structures perform excellently in dealing with multi-input and multi-output problems and can achieve efficient model construction through parameter tuning and training. However, integrating human-like learning and growth mechanisms into AI systems remains an urgent problem in the industry. Currently, the application of perception and cognition integration training mainly focuses on areas such as early childhood education, medical treatment for special groups, and rehabilitation of the elderly, while research and practice in complex application scenarios such as autonomous driving are still insufficient.

[0048] In addition, in the field of intelligent driving, the evaluation of software algorithms still highly relies on manual on-road vehicle testing, which leads to a disconnection between software development and real-world verification. From data collection, model training to software deployment, and then to on-road vehicle testing, the entire process takes a long time, and information loss is prone to occur during the process of problem understanding and optimization transmission, thus limiting the autonomous learning and closed-loop optimization capabilities of software models.

[0049] Therefore, the present invention proposes a training method for an autonomous driving model. First, a perceptron for simulating human perception and evaluation is constructed. Then, based on different driving scenarios to simulate the human growth process, the perceptron is subjected to perceptual integration training to obtain an initial autonomous driving model. Finally, the initial autonomous driving model is optimized based on individual application data at the vehicle end and aggregated application data on the platform to obtain an autonomous driving model; wherein the platform includes multiple vehicle ends. By simulating the human growth process and drawing on the concept of perceptual integration training to guide the learning process of the autonomous driving model, it enables the model to more intelligently adapt to and handle various complex driving scenarios, not only significantly enhancing the flexibility and adaptability of the autonomous driving system, but also reducing the dependence on a large amount of labeled data and complex preset rules, thus opening up a new path for the development of autonomous driving technology.

[0050] Hereinafter, the technical solution of the present invention will be further described in detail through specific embodiments.

[0051] Refer to Figure 1 , which is a schematic flowchart of the training method for the autonomous driving model provided by the embodiment of the present invention.

[0052] Step S101, construct a perceptron for simulating human perception and evaluation.

[0053] Specifically, an artificial intelligence "perceptron" needs to be established, enabling it to have basic somatosensory data and be able to collect human comfort feelings regarding longitudinal and lateral acceleration and deceleration values as well as basic rules for danger. At the same time, the "pain" of the perceptron is constructed in combination with vehicle situations such as collisions, fault diagnosis, and vehicle instability.

[0054] As an optional embodiment, constructing a perceptron for simulating human perception and evaluation includes:

[0055] Obtain the driver's comfort perception data for longitudinal and lateral acceleration and deceleration values, as well as the reaction rules for driving dangerous situations; combine the processing rules for abnormal situations of the vehicle to construct a pain mechanism similar to that of humans for the perception body; integrate the comfort perception data, reaction rules, and pain mechanism into the intelligent body to obtain the perception body.

[0056] Reference Figure 2 , which is the schematic flow chart of constructing a perception body for simulating human perception and evaluation provided by the embodiment of the present invention.

[0057] Constructing a perception body for simulating human perception and evaluation includes the following steps:

[0058] Step S201, somatosensory data acquisition and modeling.

[0059] In this step, the driver's somatosensory data corresponding to the vehicle's acceleration sensor, gyroscope, GPS, and HMI interaction system in different driving scenarios can be obtained.

[0060] Specifically, in the process of constructing the artificial intelligence "perception body", it is crucial to obtain the driver's somatosensory data corresponding to the vehicle's acceleration sensor, gyroscope, GPS, and HMI (Human-Machine Interface) interaction system in different driving scenarios.

[0061] Among them, the acceleration sensor can measure the acceleration change of the vehicle during driving, which is crucial for evaluating the driver's comfort. The acceleration data includes longitudinal acceleration (the acceleration of the vehicle in the front-rear direction) and lateral acceleration (the acceleration of the vehicle in the left-right direction). The data can be read through the vehicle's acceleration sensor interface, communicate with the acceleration sensor using dedicated data acquisition equipment or software, and record the data in real time. In driving scenarios such as sudden acceleration, sudden deceleration, or turning, the acceleration sensor data can help evaluate the driver's somatosensory comfort. By analyzing the change trend of the acceleration data, abnormal acceleration situations during driving can also be identified.

[0062] The gyroscope can measure the angular velocity change of the vehicle, that is, the attitude change of the vehicle. This is also important for evaluating the stability of the vehicle during driving and the driver's somatosensory comfort. The data can be read through the vehicle's gyroscope interface, communicate with the gyroscope using dedicated data acquisition equipment or software, and record the data in real time. When the vehicle turns, bumps, or is affected by external disturbances such as crosswinds, the gyroscope data can help evaluate the stability of the vehicle. By analyzing the change trend of the gyroscope data, abnormal changes in the vehicle's attitude can be identified.

[0063] The GPS system can provide information such as the real-time location, speed, and direction of a vehicle. This data is crucial for evaluating whether the driver's driving route, speed, and direction meet expectations. The data can be read through the vehicle's GPS system interface, using dedicated GPS data acquisition devices or software to communicate with the GPS system and record the data in real time. During navigation, the GPS data can help assess whether the driver is following the planned route. By analyzing the trend of changes in GPS data, it can be identified whether the driver's driving speed is too fast or too slow, and whether there are abnormal driving directions, etc.

[0064] The HMI interaction system provides an interface for interaction between the driver and the vehicle. By collecting data from the HMI interaction system, it is possible to understand the driver's operations and feedback on the vehicle during driving. The data can be read through the vehicle's HMI interaction system interface. Use dedicated data acquisition devices or software to communicate with the HMI interaction system and record the data in real time. When the driver uses the HMI interaction system for operations such as navigation, entertainment, or vehicle settings, the corresponding operation data and feedback can be collected. By analyzing the data of the HMI interaction system, it is possible to evaluate the driver's proficiency in operating the vehicle, understanding of vehicle functions, and satisfaction during use, etc.

[0065] On this basis, somatosensory detection sensors can also be set on the vehicle, such as electrocardiogram sensors, electroencephalogram sensors, and iris sensors, etc. The somatosensory detection sensors can directly obtain the driver's somatosensory data in different driving scenarios.

[0066] Specifically, setting somatosensory detection sensors on the vehicle, such as electrocardiogram sensors, electroencephalogram sensors, and iris sensors, etc., can directly obtain the driver's somatosensory data in different driving scenarios.

[0067] The electrocardiogram sensor can monitor the driver's electrocardiogram (ECG) data, reflecting the driver's heart activities and health status. By real-time monitoring of key indicators such as the driver's heart rate and rhythm, potential heart problems can be detected in a timely manner. In long-distance driving or high-intensity driving scenarios, evaluate the driver's fatigue level and attention level. Among them, the electrocardiogram sensor can be integrated in positions such as the seat, steering wheel, or seat belt, and collect data in a non-invasive manner. The data can be transmitted to the vehicle's central control system or the cloud through wireless methods such as Bluetooth and Wi-Fi for analysis.

[0068] The electroencephalogram (EEG) sensor can monitor the driver's EEG data, reflect their brain activities and cognitive states, and can evaluate the driver's attention level, emotional state, and cognitive load. In emergency situations, the reaction speed and decision-making ability of the driver can be judged through the EEG data. Among them, the EEG sensor can be designed as a head-mounted device and worn on the driver's head for data collection. The data can also be transmitted wirelessly to the vehicle's central control system or the cloud for analysis.

[0069] The iris sensor can identify the driver's iris characteristics for identity verification and fatigue monitoring. It is necessary to ensure that only authorized drivers can start and operate the vehicle. By monitoring the dynamic changes of the iris, the fatigue level and attention level of the driver can be evaluated. Among them, the iris sensor can be integrated near the vehicle's dashboard or steering wheel, and the iris image is captured by a camera for analysis. The data can be processed inside the vehicle or transmitted to the cloud for more advanced identity verification and fatigue monitoring.

[0070] Step S202, environmental and vehicle monitoring.

[0071] In this step, relevant data between the vehicle and the environment can be obtained through collision detection sensors, function triggers, diagnostic dashboards, and vehicle stability control systems.

[0072] Specifically, by comprehensively using the above monitoring means, relevant data between the vehicle and the environment can be obtained, including but not limited to vehicle state data, such as engine state, brake system state, tire pressure, etc. Environmental perception data, such as collision information from collision detection sensors, driving state information from vehicle stability control systems, etc. Fault diagnosis data, fault codes and diagnostic information read through the OBD system.

[0073] The collision detection sensor is mainly used to monitor the collision situation between the vehicle and the external environment in real time, and can accurately detect the collision between the vehicle and external objects (such as pedestrians, other vehicles, obstacles, etc.), and provide information such as the time, location, and intensity of the collision. When a collision occurs to the vehicle, the collision detection sensor can respond quickly and trigger safety devices such as airbags to protect the safety of the driver and passengers.

[0074] In the vehicle monitoring system, the function trigger, that is, the function trigger, refers to a specific event trigger mechanism associated with the vehicle system, such as data integrity check, business logic execution, etc. Setting triggers in the vehicle system can automatically execute predefined operations when specific events occur, such as data recording, status update, etc. For example, when a key component of the vehicle (such as the engine, brake system) fails, the trigger can automatically record the fault information.

[0075] OBD (On-Board Diagnostic system) is an important function on the vehicle dashboard, which is used to monitor the working conditions of various vehicle systems in real time. It can monitor the working states of multiple key components such as the engine electronic control system, emission system, and fuel system, and trigger an alarm when a fault occurs. The vehicle stability control system (such as ESP) is an important technology in the field of vehicle safety, which is used to maintain the stability of the vehicle in emergency situations. It can monitor the driving state of the vehicle (such as speed, direction, side slip, etc.) in real time through sensors, and automatically adjust the engine output and braking system when detecting unstable situations to restore vehicle stability. In scenarios such as emergency braking, rapid turning, or avoiding sudden obstacles, the vehicle stability control system can significantly improve driving safety and prevent the vehicle from getting out of control.

[0076] Step S203, comfort and comfort level analysis.

[0077] In this step, the data obtained in Step S201 (driver somatosensory data collection) and Step S202 (environment and vehicle monitoring) will be comprehensively used to deeply analyze and evaluate parameters such as acceleration and deceleration comfort, direction control rationality, and driving smoothness during the driving process.

[0078] For the analysis of acceleration and deceleration comfort, the vehicle acceleration and deceleration data obtained from the acceleration sensor and the physiological response data feedback by the driver through somatosensory devices such as electrocardiogram sensors and electroencephalogram sensors can be combined. By comparing the relationship between the actual vehicle acceleration and deceleration and the driver's somatosensory comfort, the impact of acceleration changes on the driver can be identified, and the physiological response data (such as heart rate changes, brain wave activities) can be used to evaluate the driver's comfort during acceleration and deceleration. Whether the acceleration change rate is within the acceptable range of the driver or whether the driver's physiological response shows tension or discomfort can be used as an evaluation index.

[0079] For the analysis of direction control rationality, the vehicle attitude and driving trajectory data provided by the gyroscope and GPS and the driver's operation data on the steering wheel (such as steering angle, steering speed) can be combined. Analyze the deviation between the vehicle driving trajectory and the expected path, evaluate the accuracy of direction control, and combine the driver's operation on the steering wheel to judge the timeliness and smoothness of direction control. The degree of deviation between the driving trajectory and the expected path or the smoothness of the steering wheel operation and the driver's reaction speed can be used as an evaluation index.

[0080] For ride comfort analysis, data provided by the vehicle stability control system and collision detection sensors, as well as road condition information (which can be obtained by combining GPS with map data), can be integrated. The vehicle stability control system data is used to evaluate the vehicle's stability on uneven roads or in emergency situations. Combining the collision detection sensor data, the interaction between the vehicle and the external environment is analyzed to judge potential risks during driving. The vehicle's stability performance under complex road conditions or the driver's ability to identify and respond to potential risks can be used as evaluation indicators.

[0081] Step S204, Hazard perception analysis.

[0082] In this step, the data obtained in Step S201 (Driver somatosensory data collection) and Step S202 (Environment and vehicle monitoring) will also be comprehensively used to conduct in-depth analysis and evaluation of hazard warnings, fault indications, and instability detections during driving.

[0083] For hazard warning analysis, data from sensors such as electrocardiogram, electroencephalogram, and iris, which reflect the driver's physiological and psychological states, can be combined with environment and vehicle monitoring data, including external environment and vehicle state information provided by collision detection sensors, vehicle stability control systems, radars, and cameras. By combining the driver's physiological responses (such as increased heart rate and distracted attention) and environment monitoring data (such as obstacles, pedestrians, and other vehicles ahead), potential dangerous situations are identified. Machine learning or deep learning algorithms are used to establish a hazard warning model to predict possible dangerous events based on historical data and real-time data. The accuracy and timeliness of the warning, that is, whether a warning can be issued in time before a danger occurs or the false alarm rate and missed alarm rate of the warning, are used to evaluate the reliability of the warning system as evaluation indicators.

[0084] For fault indication analysis, data on the status of each vehicle system provided by the vehicle diagnostic dashboard (such as the OBD system) and sensor data, such as vehicle internal status information provided by the engine temperature sensor and oil pressure sensor, can be combined. The operating status of each vehicle system is monitored in real time to identify abnormal data, judge whether there is a fault, and give corresponding fault indications, such as warning lights and sound prompts, according to the type and severity of the fault. The accuracy and reliability of the fault indication, that is, whether the fault can be accurately identified and the correct indication given, or the timeliness and effectiveness of the fault indication, that is, whether the driver can be quickly reminded and corresponding countermeasures can be taken when the fault occurs, can be used as evaluation indicators.

[0085] For instability detection and analysis, it is possible to combine the vehicle attitude and driving state data provided by the vehicle stability control system with the vehicle dynamic information provided by gyroscopes, acceleration sensors, etc. Monitor the driving state of the vehicle in real time, including speed, acceleration, steering angle, etc., to determine whether the vehicle is in an unstable state, and use algorithms to process and analyze the vehicle attitude data to identify early signs of vehicle instability. The accuracy and sensitivity of instability detection, that is, whether it can accurately identify vehicle instability and give timely warnings, or the robustness and stability of the instability detection system, that is, it can maintain good detection performance under different road conditions and driving conditions, can be used as evaluation indicators.

[0086] Step S205, intelligent analysis.

[0087] In this step, the data obtained in step S201 (driver somatosensory data collection) and step S202 (i.e., environment and vehicle monitoring) will also be comprehensively used to deeply analyze and evaluate the takeover mileage (i.e., the mileage traveled by the autonomous driving after switching from manual driving to autonomous driving), takeover scenario (i.e., when to switch from manual driving to autonomous driving), coverage mileage (i.e., the range of autonomous driving), obstacle recognition (i.e., identifying obstacles that appear in autonomous driving), and commuting efficiency (i.e., autonomous driving efficiency) during the driving process.

[0088] Specifically, physiological response data can reflect the driver's tension, fatigue, etc. during driving, providing a reference for the activation timing of the autonomous driving system. For example, when it is detected that the driver's heart rate increases and blood pressure rises, the autonomous driving system can take over the driving task in a timely manner to relieve the driver's stress. Driving behavior data can reveal the driver's driving habits and preferences, providing a basis for the personalized settings of the autonomous driving system. At the same time, abnormal behavior data may also indicate potential safety risks, which require the autonomous driving system to intervene in a timely manner. Emotional state data is also important for evaluating the driver's driving state. When the driver shows negative emotions such as anxiety and anger, the autonomous driving system can comfort the driver's emotions through voice prompts, music playback, etc. to improve driving safety.

[0089] By analyzing the driver's somatosensory data and vehicle status data, the activation timing and takeover conditions of the autonomous driving system can be judged more accurately, thereby optimizing the allocation and efficiency of the takeover mileage. Combining environmental and vehicle monitoring data can identify the performance differences of the autonomous driving system in different scenarios, providing a basis for optimizing the triggering mechanism. By analyzing road condition and weather condition data, the applicable scope and limitations of the autonomous driving system can be evaluated, providing a reference for optimizing map data and path planning algorithms. Combining sensor data and algorithm parameters can improve the detection accuracy and recognition speed of the autonomous driving system for obstacles, thereby enhancing its ability to handle complex scenarios. By analyzing vehicle status data and the dynamic data of surrounding objects, the path planning algorithm and vehicle control strategy of the autonomous driving system can be optimized, improving commuting efficiency and user satisfaction.

[0090] Among them, road condition data is crucial for the path planning and decision-making of the autonomous driving system. By analyzing this data, the autonomous driving system can identify the optimal driving route and avoid potential road risks. Weather condition data also affects the performance and safety of the autonomous driving system. For example, in rainy or foggy weather, the autonomous driving system needs to adjust sensor parameters and algorithm strategies to improve the accuracy and reliability of obstacle recognition. Vehicle status data reflects the current control status and performance of the autonomous driving system. By analyzing this data, the autonomous driving system can detect and solve potential problems in a timely manner, such as wheel slippage and brake failure. The dynamic data of surrounding vehicles and pedestrians is also important for obstacle avoidance and path planning of the autonomous driving system. The autonomous driving system needs to perceive and predict the movement trajectories of surrounding objects in real time to ensure driving safety and efficiency.

[0091] Step S205, AI perception body fusion decision-making.

[0092] In this step, the data obtained in step S203 (comfort and comfort analysis), step S204 (hazard perception analysis), and step S205 (intelligent analysis) will be integrated to perform data fusion, pain simulation, and behavior suggestion analysis.

[0093] Specifically, AI perception body fusion decision-making is a key link in the autonomous driving system. It integrates the data and information obtained in multiple steps to perform data fusion, pain simulation, and behavior suggestion analysis to make the optimal driving decision.

[0094] For data fusion, first, the data obtained in step S203 (analysis of peace of mind and comfort), step S204 (analysis of danger perception), and the early stage of step S205 (intelligent analysis, specifically referring to the part other than data fusion, pain simulation, and analysis of behavior suggestions) need to be fused. These data include, but are not limited to, peace of mind and comfort data, which reflect the psychological feelings and physiological comfort of the driver during driving, such as heart rate, breathing rate, facial expressions, etc. Danger perception data, which are potential danger information detected through sensors and algorithms, such as obstacles ahead, traffic violations, road construction, etc. Intelligent analysis data, which are in-depth analysis results of key indicators such as takeover mileage, takeover scenarios, covered mileage, obstacle recognition, and commuting efficiency during driving.

[0095] The purpose of data fusion is to integrate these data from different sources into a comprehensive and consistent driving environment model, providing a basis for subsequent pain simulation and analysis of behavior suggestions.

[0096] For pain simulation, based on data fusion, step S205 further conducts pain simulation. Pain simulation is a virtual experience technology that simulates the discomfort or sense of danger that a driver may feel in specific driving situations. Through pain simulation, the autonomous driving system can more intuitively evaluate the impact of different driving decisions on the driver, and thus make more user-friendly driving decisions.

[0097] Among them, the specific implementation methods of pain simulation can include:

[0098] Physiological feedback simulation, that is, by simulating the physiological reactions of the driver in situations such as emergency braking and sharp turns (such as increased heart rate and blood pressure), to evaluate the impact of these situations on the driver.

[0099] Psychological feedback simulation, that is, by simulating the psychological state of the driver when facing potential dangers or uncertain situations (such as anxiety and nervousness), to evaluate the impact of these situations on the driver's driving behavior.

[0100] Furthermore, after completing data fusion and pain simulation, step S205 can also conduct analysis of behavior suggestions. Analysis of behavior suggestions is based on the results of the previous steps and provides specific driving behavior suggestions for the autonomous driving system. These suggestions aim to optimize the driving process and improve driving safety, comfort, and efficiency.

[0101] Among them, the analysis of behavior suggestions can include the following aspects:

[0102] Route planning suggestions: Provide the optimal driving route suggestions for the autonomous driving system according to road conditions, traffic conditions, and driver preferences.

[0103] Speed control suggestion: Provide appropriate speed control suggestions for the autonomous driving system according to the current road conditions, traffic flow, and the physiological and psychological states of the driver.

[0104] Obstacle avoidance strategy suggestion: Provide the best obstacle avoidance strategy suggestions for the autonomous driving system when detecting potential dangers to ensure driving safety.

[0105] Driving mode switching suggestion: Provide timely driving mode switching suggestions (such as switching from autonomous driving mode back to manual driving mode) for the autonomous driving system according to the driver's somatosensory data and comfort analysis results.

[0106] In summary, the AI perception body fusion decision in step S205 is an important link in the autonomous driving system. It integrates the data and information obtained in multiple steps and provides the optimal driving decision for the autonomous driving system through data fusion, pain simulation, and behavior suggestion analysis. This process not only improves the safety and comfort of the autonomous driving system but also enhances its interactivity and humanization with the driver.

[0107] Step S206, feedback and execution.

[0108] In this step, vehicle control, output of user interface feedback content, and generation of safety warnings will be carried out based on the comprehensive results of step S205 (AI perception body fusion decision).

[0109] Specifically, feedback and execution are the key links in the autonomous driving system to transform decisions into actual actions. In this step, the system integrates the results of step S205 (AI perception body fusion decision) and performs a series of operations to control the vehicle, interact with the user, and ensure driving safety.

[0110] For vehicle control, the autonomous driving system can first directly control the vehicle according to the decision results of step S205. This includes path execution, that is, according to the recommended path planning in the decision results, the autonomous driving system adjusts parameters such as steering angle and vehicle speed to make the vehicle drive along the planned route. Speed adjustment, that is, according to the current road conditions, traffic flow, and speed control suggestions in the decision results, the autonomous driving system adjusts the vehicle speed in real time to maintain driving safety and efficiency. Obstacle avoidance operation, that is, when detecting potential dangers, the autonomous driving system will take corresponding obstacle avoidance operations according to the obstacle avoidance strategy in the decision results, such as emergency braking and steering avoidance.

[0111] For the output of user interface feedback content, in addition to vehicle control, step S206 also involves outputting interface feedback content to the user, including the display of driving information, that is, during the vehicle's driving process, the automatic driving system displays information such as the current driving speed, remaining mileage, and route planning to the user through interfaces such as in-vehicle displays. System status prompts, that is, when the automatic driving system is in a specific state (such as the activation of the automatic driving mode, the switching of the manual driving mode, etc.), clear prompts are sent to the user through the interface to ensure that the user understands the current state of the system. Safety warning information, that is, when detecting potential dangers or system anomalies, the automatic driving system issues safety warnings to the user in the form of sound, light, or text to remind the user to pay attention and take corresponding measures.

[0112] For the generation of safety warnings, it is also another important link in step S206. According to the decision result of step S205, the automatic driving system generates and issues safety warning information to remind the user or the system itself of potential dangers, including warnings of obstacles ahead, that is, when detecting obstacles ahead (such as vehicles, pedestrians, animals, etc.), the system issues warning information to remind the user or automatically take obstacle avoidance operations. Traffic violation warnings, that is, when detecting that the vehicle may violate traffic rules (such as running a red light, going in the wrong direction, etc.), the system issues warning information to remind the user to pay attention and correct. System fault warnings, that is, when the system detects its own faults or anomalies, it issues warning information to remind the user to pay attention and take corresponding measures, such as switching to the manual driving mode or seeking help.

[0113] In summary, the feedback and execution link of step S206 is a key step in the automatic driving system to convert decisions into actual actions. Through operations such as vehicle control, output of user interface feedback content, and generation of safety warnings, the automatic driving system can ensure driving safety and efficiency, while improving the user experience and satisfaction.

[0114] Step S102, based on different driving scenarios, simulates the human growth process to perform sensory integration training on the sensor to obtain an initial automatic driving model.

[0115] As an optional embodiment, based on different driving scenarios, simulating the human growth process to perform sensory integration training on the sensor to obtain an initial automatic driving model includes: designing at least one set of training systems from simple to complex and from low speed to high speed; training the sensor's ability in sensory information processing based on the training system to obtain an initial automatic driving model.

[0116] Reference Figure 3 , is a schematic diagram of the sensory integration training process provided by the embodiment of the present invention.

[0117] In the process of simulating the human growth process based on different driving scenarios and conducting sensory integration training for the perception body, the design of all training scenarios and steps should ensure safety, especially in terms of collision risk identification and control, which should be given top priority.

[0118] As an optional embodiment, the perception body can be trained in the order from simple to complex and from low speed to high speed.

[0119] Furthermore, starting from simple scenarios, gradually increase the complexity to ensure that the driver or the autonomous driving system can gradually adapt to and master various driving skills.

[0120] In addition, multi-sensory integration should be emphasized, and the training should cover the cultivation of the processing abilities of various sensory information such as vision and hearing to simulate the complex perception requirements in the real driving environment.

[0121] Finally, for human-machine interaction, introduce two-way interaction of HMI (Human-Machine Interface) during the training process to collect the driver's feedback for optimizing the training effect and the performance of the intelligent driving system.

[0122] As an optional embodiment, the training system includes at least one of safety collision risk identification and control, single-lane control, multi-lane control, and urban road U-turn control.

[0123] Among them, single-lane training can be from low-speed driving conditions to high-speed driving conditions. In low-speed driving conditions, it can include straight-line control of the vehicle from point A to point B, large-curvature bend control, car following, motorcycle following, pedestrian crossing, truck following, vehicle coming from behind, perpendicular parking, parallel parking, emergency braking response, blind spot detection and alarm, instability detection, flat tire detection, and passing through the toll station, etc. In high-speed driving conditions, it can include straight-line control of the vehicle from point A to point B, large-curvature bend control, car following, motorcycle following, pedestrian crossing, truck following, and vehicle coming from behind, etc.

[0124] Among them, after single-lane training, multi-lane training can be carried out. Multi-lane training can also be from low-speed driving conditions to high-speed driving conditions. In low-speed driving conditions, it can include straight-line control of the vehicle, driving from point A to point B, combining driving from point A to point B with right and left lane changes, large-curvature bend control, combining large-curvature bend control with right and left lane changes, car following, motorcycle following, pedestrian crossing, truck following, vehicle coming from behind, emergency braking response, blind spot detection and alarm, and passing through toll stations and other scenarios. In high-speed driving conditions, it can include straight-line control of the vehicle, driving from point A to point B, large-curvature bend control, car following, motorcycle following, pedestrian crossing, truck following, vehicle coming from behind, emergency steering and avoidance, overtaking slow vehicles and other scenarios.

[0125] Through the design of the above training system, it is possible to simulate the sensory integration training in the process of human growth, and gradually improve the driving skills of drivers or autonomous driving systems. Starting from simple scenarios, gradually increasing the complexity, ensuring safety, and paying attention to multi-sensory integration and human-machine interaction will provide strong support for the development of autonomous driving technology. At the same time, by collecting drivers' feedback and continuously optimizing the training effect, the performance and safety of the intelligent driving system can be further improved.

[0126] Step S103, optimize the initial autonomous driving model based on the individual application data of the vehicle end and the aggregated application data of the platform to obtain an autonomous driving model; wherein, the platform includes multiple vehicle ends.

[0127] As an optional embodiment, the individual application data includes the driver's opinion feedback; the aggregated application data includes the characteristics of different vehicles. Optimizing the initial autonomous driving model based on the individual application data of the vehicle end and the aggregated application data of the platform to obtain an autonomous driving model includes: establishing a driving HMI two-way interaction mechanism, collecting the driver's opinion feedback, and inputting the opinion feedback into the initial autonomous driving model for learning and optimization; adjusting the parameters of the initial autonomous driving model according to the characteristics of different vehicles to obtain an autonomous driving model.

[0128] Specifically, the in-vehicle closed-loop mainly reflects the learning and improvement of individual capabilities and is a specific manifestation of the sensory integration training paradigm. It can perform real-time learning and adjustment according to the actual driving scenarios of in-vehicle individuals, thereby optimizing driving behaviors and strategies. However, in-vehicle individuals are restricted by relatively fixed scenarios in use and are prone to problems such as local optimal solutions and even overfitting. In addition, the model training iteration and upgrade capabilities and speeds of in-vehicle individuals are relatively slow and cannot quickly adapt to new driving scenarios and changes. The platform closed-loop belongs to the collective evolution of the experiences of multiple in-vehicle individuals. Through regular upgrades and evolutions on the platform, it can feed back to each surviving in-vehicle individual and improve their driving capabilities and safety. The platform-based model training version has stronger generalization capabilities and performance and can better adapt to various complex driving scenarios. However, there may be some waste in data collection and processing in the platform closed-loop. If the data collected by the platform is of low quality or processed improperly, it will directly affect the effect and accuracy of model training.

[0129] To overcome the limitations of the in-vehicle closed-loop and the platform closed-loop respectively, the present invention combines the advantages of both to construct a dual closed-loop model training system. The specific approach is as follows:

[0130] Perform sensory integration training on the vehicle end, and use the real-time driving data and learning algorithms of in-vehicle individuals to continuously optimize driving behaviors and strategies. At the same time, in-vehicle individuals can transmit their own learning achievements and experiences to the platform to provide high-quality data support for the platform.

[0131] The platform receives data from in-vehicle individuals and performs integration, cleaning, and annotation. Then, using big data and machine learning technologies, the model is trained and iteratively upgraded. Through the regular upgrade and evolution of the platform, it can feed back to each surviving in-vehicle individual and improve their driving capabilities and safety.

[0132] A synergistic optimization relationship is formed between the in-vehicle closed-loop and the platform closed-loop. In-vehicle individuals continuously improve their own capabilities through sensory integration training and transmit their experiences to the platform; the platform, through data integration and model training, provides higher-quality feedback and guidance for in-vehicle individuals. This dual closed-loop synergistic optimization mechanism can promote the continuous development and progress of the entire autonomous driving system.

[0133] The dual closed-loop model training system combining the vehicle end and the platform is one of the important directions for the development of autonomous driving technology. Through the synergistic optimization of in-vehicle sensory integration training and platform data integration and upgrade, it can overcome the limitations of in-vehicle individuals in use and the waste in data processing on the platform, and improve the overall performance and safety of the autonomous driving system. At the same time, this dual closed-loop model training system also provides strong support for the continuous development and innovation of autonomous driving technology.

[0134] As can be seen from the above, the training method of the autonomous driving model provided by the present invention first constructs a perceptual body for simulating human perception and evaluation, and then, based on different driving scenarios, simulates the human growth process to perform perceptual integration training on the perceptual body to obtain an initial autonomous driving model. Finally, the initial autonomous driving model is optimized based on the individual application data of the vehicle end and the aggregated application data of the platform to obtain the autonomous driving model; wherein the platform includes multiple vehicle ends. By simulating the human growth process and drawing on the concept of perceptual integration training, the learning process of the autonomous driving model is guided, enabling it to more intelligently adapt to and process various complex driving scenarios, not only significantly enhancing the flexibility and adaptability of the autonomous driving system, but also reducing the dependence on a large amount of labeled data and complex preset rules, thus opening up a new path for the development of autonomous driving technology.

[0135] It should be noted that the method of the embodiment of the present invention can be executed by a single device, such as a computer or a server. The method of this embodiment can also be applied to a distributed scenario and completed by multiple devices cooperating with each other. In this case of a distributed scenario, one of the multiple devices can only execute one or more steps of the method of the embodiment of the present invention, and these multiple devices will interact with each other to complete the described method.

[0136] It should be noted that some embodiments of the present invention have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be executed in a different order from that in the above embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0137] Based on the same inventive concept, corresponding to the method provided in any of the above embodiments, the present invention further provides an application method of an autonomous driving model, including controlling a vehicle to perform autonomous driving based on the autonomous driving model obtained by the training method of the autonomous driving model as described above.

[0138] Based on the same inventive concept, corresponding to the method provided in any of the above embodiments, the present invention further provides a training device for an autonomous driving model.

[0139] Refer to Figure 4 , which is a schematic diagram of a training device for an autonomous driving model provided by an embodiment of the present invention.

[0140] The device includes:

[0141] A construction module 401, configured to construct a perceptual body for simulating human perception and evaluation;

[0142] A training module 402, configured to simulate the human growth process based on different driving scenarios, perform sensory integration training on the perception body, and obtain an initial autonomous driving model;

[0143] A verification module 403, configured to optimize the initial autonomous driving model based on the individual application data of the vehicle end and the collective application data of the platform, and obtain the autonomous driving model; wherein, the platform includes multiple vehicle ends.

[0144] Optionally, the construction module 401 is further configured to:

[0145] Obtain the comfort perception data of the driver for longitudinal and lateral acceleration and deceleration values, as well as the reaction rules for driving dangerous situations;

[0146] Combine the processing rules for abnormal situations of the vehicle to construct a pain mechanism similar to that of humans for the perception body;

[0147] Integrate the comfort perception data, reaction rules, and pain mechanism into the intelligent body to obtain the perception body.

[0148] Optionally, the training module 402 is further configured to:

[0149] Design at least one set of training systems from simple to complex and from low speed to high speed;

[0150] Based on the training system, train the ability of the perception body in sensory information processing to obtain an initial autonomous driving model.

[0151] Optionally, the training system includes at least one of safety collision risk identification and control, single-lane control, multi-lane control, and urban road U-turn control.

[0152] Optionally, the training module 402 is further configured to:

[0153] Train the perception body in the order from simple to complex and from low speed to high speed.

[0154] Optionally, the individual application data includes the driver's opinion feedback; the collective application data includes the characteristics of different vehicles;

[0155] The verification module 403 is further configured to:

[0156] Establish a two-way driving HMI interaction mechanism, collect the driver's opinion feedback, and input the opinion feedback into the initial autonomous driving model for learning and optimization;

[0157] According to the characteristics of different vehicles, adjust the parameters of the initial autonomous driving model to obtain the autonomous driving model.

[0158] Optionally, the verification module 403 is further configured to:

[0159] Extract high-quality feedback with a qualified feedback accuracy rate from the collected driver feedback, and transmit the high-quality feedback to the platform, so that the platform optimizes the aggregated application data based on the high-quality feedback.

[0160] According to a training device for an autonomous driving model provided by an embodiment of the present invention, first, a perceptron for simulating human perception and evaluation is constructed. Then, based on different driving scenarios, the process of human growth is simulated, and the perceptron is subjected to perceptual integration training to obtain an initial autonomous driving model. Finally, the initial autonomous driving model is optimized based on the individual application data of the vehicle terminal and the aggregated application data of the platform to obtain an autonomous driving model; wherein the platform includes multiple vehicle terminals. By simulating the process of human growth and drawing on the concept of perceptual integration training, the learning process of the autonomous driving model is guided, enabling it to more intelligently adapt to and process various complex driving scenarios. This not only significantly improves the flexibility and adaptability of the autonomous driving system but also reduces the dependence on a large amount of labeled data and complex preset rules, thus opening up a new path for the development of autonomous driving technology.

[0161] For the convenience of description, when describing the above system, various modules are described separately according to their functions. Of course, when implementing the present invention, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0162] The system of the above embodiment is used to implement the corresponding method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be elaborated here.

[0163] Based on the same inventive concept, corresponding to the method described in any of the above embodiments, the present invention further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the method described in any of the above embodiments.

[0164] Figure 5 FIG. shows a more specific schematic diagram of the hardware structure of the electronic device provided in this embodiment. The device may include: a processor 510, a memory 520, an input / output interface 530, a communication interface 540, and a bus 550. Among them, the processor 510, the memory 520, the input / output interface 530, and the communication interface 540 are communicatively connected to each other inside the device through the bus 550.

[0165] The processor 510 can be implemented in the form of a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0166] The memory 520 can be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, etc. The memory 520 can store an operating system and other application programs. When implementing the technical solutions provided in the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 520 and are called and executed by the processor 510.

[0167] The input / output interface 530 is used to connect to the input / output module to achieve information input and output. The input / output module can be configured as a component in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Among them, the input device can include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device can include a display, a speaker, a vibrator, an indicator light, etc.

[0168] The communication interface 540 is used to connect to a communication module (not shown in the figure) to achieve communication interaction between this device and other devices. Among them, the communication module can achieve communication through a wired method (such as USB, network cable, etc.) or through a wireless method (such as a mobile network, WIFI, Bluetooth, etc.).

[0169] The bus 550 includes a path for transmitting information between various components of the device (such as the processor 510, the memory 520, the input / output interface 530, and the communication interface 540).

[0170] It should be noted that although the above device only shows the processor 510, the memory 520, the input / output interface 530, the communication interface 540, and the bus 550, in the specific implementation process, this device may also include other components necessary for normal operation. In addition, those skilled in the art can understand that the above device may also only include the components necessary to implement the solutions of the embodiments of this specification, and do not necessarily include all the components shown in the figure.

[0171] The electronic device in the above embodiment is used to implement the corresponding method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be elaborated here.

[0172] Based on the same inventive concept, corresponding to the method described in any of the above embodiments, the present invention further provides a vehicle, including the electronic device described above, wherein the memory of the electronic device stores a program or instruction that can run on a processor, and when the program or instruction is executed by the processor, the steps of the training method of the above-described autonomous driving model are implemented.

[0173] Based on the same inventive concept, corresponding to the method described in any of the above embodiments, the present invention further provides a computer-readable storage medium, and the computer-readable storage medium stores computer instructions for causing the computer to execute the method described in any one of the above embodiments.

[0174] The above computer-readable storage medium may be any available medium or data storage device that can be accessed by a computer, including but not limited to magnetic memories (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical memories (such as CDs, DVDs, BDs, HVDs, etc.), and semiconductor memories (such as ROMs, EPROMs, EEPROMs, non-volatile memories (NANDFLASH), solid-state drives (SSD)).

[0175] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the method described in any one of the above exemplary method parts, and have the beneficial effects of the corresponding method embodiments, which will not be elaborated here.

[0176] In addition, although the operations of the method of the present invention are described in a specific order in the drawings, this does not require or imply that these operations must be performed in that specific order, or that all the shown operations must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart can be changed in the execution order. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step for execution, and / or one step may be decomposed into multiple steps for execution.

[0177] It should be understood that each part of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following well-known technologies in the art can be used: discrete logic circuits having logic gate circuits for implementing logic functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0178] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present invention should have the ordinary meanings understood by those of ordinary skill in the art to which the present invention pertains. The "first", "second" and similar terms used in the embodiments of the present invention do not denote any order, quantity or importance, but are only used to distinguish different components. Words such as "comprising" or "including" mean that the elements or items appearing before the word cover the elements or items listed after the word and their equivalents, without excluding other elements or items. Words such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Upper", "lower", "left", "right", etc. are only used to indicate relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0179] Although the spirit and principles of the present invention have been described with reference to several specific embodiments, it should be understood that the present invention is not limited to the specific embodiments disclosed, and the division of each aspect does not mean that the features in these aspects cannot be combined for benefits. Such division is only for convenience of expression. The present invention aims to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims. The scope of the appended claims is to be accorded the broadest interpretation so as to encompass all such modifications and equivalent structures and functions.

Claims

1. A training method for an autonomous driving model, characterized in that: include: Constructing perceptrons to simulate human perception and evaluation; Based on different driving scenarios to simulate the human growth process, the perception integration training is performed on the perception body to obtain an initial autonomous driving model; The initial autonomous driving model is optimized based on the individual application data of the vehicle end and the collective application data of the platform to obtain the autonomous driving model; wherein the platform includes multiple vehicle ends.

2. The training method of the automatic driving model according to claim 1, characterized in that: The perceptron constructed to simulate human perception and evaluation includes: Obtain the driver's comfort perception data on longitudinal and lateral acceleration and deceleration values, as well as the reaction rules to dangerous driving situations; Combined with the processing rules for abnormal situations of the vehicle, a human-like pain mechanism is constructed for the sensor; The comfort perception data, the reaction rules and the pain mechanism are integrated into an intelligent agent to obtain the perception agent.

3. The training method of the automatic driving model according to claim 2, characterized in that: The method of simulating the human growth process based on different driving scenarios, performing perception integration training on the sensor, and obtaining an initial autonomous driving model includes: Design at least one training system from simple to complex, from low speed to high speed; The initial autonomous driving model is obtained by training the sensory information processing capability of the sensor based on the training system.

4. The training method of the automatic driving model according to claim 3, characterized in that: The training system includes at least one of safety collision risk identification and control, single lane control, multi-lane control, and urban road U-turn control.

5. The training method of the automatic driving model according to claim 4, characterized in that: The individual application data includes feedback from drivers; the collective application data includes characteristics of different vehicles; The step of optimizing the initial autonomous driving model based on the individual application data of the vehicle and the collective application data of the platform to obtain the autonomous driving model includes: Establishing a driving HMI two-way interactive mechanism to collect feedback from drivers, and inputting the feedback into the initial autonomous driving model for learning and optimization; According to the characteristics of different vehicles, the parameters of the initial autonomous driving model are adjusted to obtain the autonomous driving model.

6. The training method of the automatic driving model according to claim 5, characterized in that: The method further comprises: High-quality feedback with a satisfactory feedback accuracy is extracted from the collected feedback from drivers, and the high-quality feedback is transmitted to the platform, so that the platform optimizes the aggregate application data based on the high-quality feedback.

7. An application method of an automatic driving model, characterized in that: The invention comprises controlling a vehicle to perform autonomous driving based on an autonomous driving model obtained by the training method of the autonomous driving model as described in any one of claims 1 to 6.

8. A training device for an autonomous driving model, characterized in that: include: A building module configured to build a perceptron for simulating human perception and evaluation; A training module is configured to simulate the human growth process based on different driving scenarios, perform perception integration training on the sensor, and obtain an initial autonomous driving model; The verification module is configured to optimize the initial autonomous driving model based on the individual application data of the vehicle end and the collective application data of the platform to obtain the autonomous driving model; wherein the platform includes multiple vehicle ends.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the training method of the autonomous driving model as described in any one of claims 1 to 6 are implemented.

10. A vehicle, characterized in that: An electronic device comprising the electronic device of claim 9, wherein the memory of the electronic device stores programs or instructions executable on a processor, and the programs or instructions, when executed by the processor, implement the steps of the training method of the autonomous driving model as described in any one of claims 1 to 6.

11. A computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable the computer to implement the steps of the training method of the autonomous driving model described in any one of claims 1 to 6.

Images