Intelligent driving function algorithm safe self-evolution method based on data closed loop

Abstract

The invention belongs to the technical field related to intelligent driving functions, and particularly relates to an intelligent driving function algorithm safe self-evolution method based on a data closed loop, which comprises the following steps of: obtaining a current vehicle state based on real-time data obtained by a vehicle sensor by adopting a current grey box model; based on the current vehicle state, a preliminary control instruction is generated through MPC optimization; performing security check and optimization on the preliminary control instruction by using a control barrier function and quadratic programming to ensure that the control instruction Usafe is close to the initial control instruction Uinitial as much as possible on the premise of meeting security constraints; the control instruction Usafe is converted into an actual execution action of the vehicle; the grey box model is the sum of a vehicle state physical model which does not contain system noise and a data driving model which is used for describing nonlinear characteristics and time-varying characteristics of the vehicle, the data driving model in the grey box model obtains parameter updating optimization after a period of time, and accurate prediction of the vehicle state can be achieved in a complex and changeable environment.

Description

Technical Field

[0001] The present invention belongs to the technical field related to intelligent driving functions, and more specifically, relates to a method for safe self-evolution of intelligent driving function algorithms based on a data closed loop. Background Art

[0002] The early stages of intelligent driving feature development often face challenges with insufficient data and limited scenarios. To quickly launch functional software methods and conduct testing and verification, systems typically begin with a minimum viable product (MVP). This approach aims to implement basic functions such as adaptive cruise control, lane keeping, and automatic parking under idealized conditions using simple and stable algorithms. This approach allows developers to quickly verify core functions and collect data feedback, providing a basis for subsequent iterations and improvements.

[0003] As vehicle systems are increasingly applied to increasingly complex and diverse real-world driving environments, initial models that rely solely on fixed rules and static parameters are no longer able to meet evolving demands. In this context, self-evolution based on a closed-data loop becomes particularly important. This closed-data loop involves collecting real-world driving data, performing model updates and parameter optimization based on this data, and then updating the system to improve performance and adaptability. Furthermore, any algorithm optimization or parameter adjustments must ensure that the optimization process does not compromise system safety. This safe self-evolution approach based on a closed-data loop helps vehicle systems maintain driving safety and robustness in the face of complex environments and uncertainties.

[0004] Against this backdrop, several new technical solutions have emerged. Patent CN116300850A proposes a humanoid safety self-evolutionary method and system for autonomous driving based on data mechanism fusion. This method extracts features from real human driving data and iteratively extracts the objective function using a maximum entropy inverse reinforcement learning algorithm. However, this method suffers from its dependence on large amounts of historical data and challenges adapting to unknown or abnormal traffic conditions. Furthermore, patent CN118529081A proposes a self-evolutionary learning autonomous driving assistance system and method based on edge scenarios. This system implements personalized automatic vehicle driving route recommendations by integrating user identification, driving data collection and analysis, self-evolutionary intelligent learning, and personalized intelligent control modules. Although this system improves system adaptability through intelligent learning modules, it faces challenges in comprehensive and real-time data collection, and cannot guarantee safety in changing traffic environments and complex driving scenarios. Summary of the Invention

[0005] In response to the above-mentioned defects or improvement needs of the prior art, the present invention provides a safe self-evolution method for intelligent driving function algorithms based on a data closed loop, the purpose of which is to ensure the safety of intelligent driving in changing traffic environments and complex driving scenarios.

[0006] To achieve the above objectives, according to one aspect of the present invention, a method for safe self-evolution of intelligent driving function algorithms based on a closed data loop is provided, comprising:

[0007] Using the current gray box model, the current vehicle state is obtained based on the real-time data obtained by the vehicle sensors. Based on the current vehicle state, preliminary control instructions are generated through MPC optimization.

[0008] Using the control barrier function and quadratic programming, the initial control instructions are checked and optimized for safety, ensuring that the control instructions U safe Under the premise of meeting the safety constraints, the control instruction U initial ; Among them, the safety conditions that need to be met for the vehicle to be in a safe state are: h(S) represents the control obstacle function, S represents the vehicle state, is the derivative of the control barrier function, γ is a positive number;

[0009] The control instruction U safe Translated into actual execution actions of the vehicle;

[0010] Among them, the gray box model is the sum of a vehicle state physical model that does not contain system noise and a data-driven model used to describe the nonlinear and time-varying characteristics of the vehicle. The data-driven model in the gray box model is parameter updated and optimized after a period of data collection, so as to achieve accurate prediction of the vehicle state in a complex and changing environment.

[0011] Furthermore, the initial control instruction generation method is:

[0012] Using deep learning algorithms to process real-time data, the system identifies traffic environment information, including lane markings, vehicles ahead, pedestrians, and traffic signs. Based on this traffic environment information, the system determines the target state of the vehicle in accordance with a preset driving strategy. The real-time data includes real-time environmental information data and real-time vehicle state data acquired by vehicle sensors.

[0013] The current gray box model is used to obtain the current vehicle state based on the real-time data; and the model predictive control MPC algorithm is used to generate preliminary control instructions that meet the target state based on the current vehicle state.

[0014] Furthermore, the preliminary control instructions are optimized to obtain the control instruction U safe The quadratic programming optimization model is:

[0015]

[0016] Where U initial Represents the preliminary control command, U represents the control command independent variable, and f(S, U) represents the vehicle state physical model without system noise.

[0017] Furthermore, the gray box model is represented as S k+1 =f physical (S k , U k )+g θ (D key ); where f physical (S k , U k ) represents the vehicle state physical model without system noise, g θ (D key ) is the residual function based on the neural network, θ is the parameter of the neural network; D key Key data obtained after preprocessing vehicle status, environmental information, control input and historical driving data;

[0018] The method for updating and optimizing the vehicle's gray box model is:

[0019] The vehicle state, environmental information, control input, and historical driving data are used to update the training sample set and adjust the parameters θ of the neural network. In each training sample, the neural network input data is the key data obtained by preprocessing the vehicle state, environmental information, control input, and historical driving data. The neural network output data is the difference between the actual vehicle state and the vehicle state predicted by the physical model.

[0020] Furthermore, the preprocessing is implemented by using edge computing methods to filter and reduce noise, extract features and compress data.

[0021] According to another aspect of the present invention, a data closed-loop-based safe self-evolution system for intelligent driving function algorithms is provided, which is characterized in that it is used to execute the above-mentioned safe self-evolution method for intelligent driving function algorithms, including: a function algorithm module, a safety supervision module, a data collection and storage module, a model update module, and an execution module;

[0022] Wherein, the functional algorithm module is used to generate preliminary control instructions;

[0023] The safety supervision module is used to use the control barrier function and quadratic programming to perform safety verification and optimization on the preliminary control instructions to obtain the control instructions U safe ;

[0024] The execution module is used to execute the control instruction U safe Translated into actual execution actions of the vehicle;

[0025] The data collection and storage module is used for data collection so as to enable the model updating module to update the model parameters of the gray box model.

[0026] Furthermore, the data collection and storage module includes a vehicle side and a cloud side, wherein the vehicle side is used to collect data and upload it to the cloud side after pre-processing, or the vehicle side is used to collect data and upload it directly to the cloud side, and the cloud side is used to pre-process the data;

[0027] The cloud is further used to update the model parameters of the gray box model based on the preprocessed data and transmit them to the functional algorithm module and the safety supervision module to provide the current vehicle status to the functional algorithm module and the safety supervision module respectively.

[0028] According to another aspect of the present invention, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the above method when executing the computer program.

[0029] According to another aspect of the present invention, a computer-readable storage medium is provided, characterized in that the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed by a processor, the device where the storage medium is located is controlled to execute the steps of the method described above.

[0030] According to another aspect of the present invention, a computer program product is provided, comprising a computer program or instructions, wherein the computer program or instructions implement the steps of the above method when executed by a processor.

[0031] In general, the above technical solutions conceived by the present invention have the following beneficial effects compared with the prior art:

[0032] 1. The present invention proposes a safe self-evolution method for intelligent driving function algorithms based on a data closed loop. This method realizes real-time monitoring, dynamic optimization and intelligent control of the vehicle's operating status by integrating functional algorithms, safety supervision, data collection and storage, model updates and instruction execution. Through the data closed loop mechanism, high-quality data can be continuously acquired, supporting the iterative update of model algorithms and improving the performance and reliability of the system. The safety supervision mechanism ensures the safety of control instructions, the model update operation realizes the continuous optimization of the vehicle model, and the execution operation ensures the accurate execution of control instructions. The entire method forms a data-driven closed loop mechanism, which realizes the adaptive and reliable operation of the intelligent driving system in a complex environment, laying a solid foundation for the realization of high-level autonomous driving functions.

[0033] 2. This invention also proposes a graybox model and a method for updating model parameters. This graybox model combines the interpretability of physical models with the adaptability of data-driven models, compensating for the shortcomings of physical models and improving prediction accuracy and system robustness. It enables real-time optimization and adaptability to complex, nonlinear, and time-varying environments, enhancing the safety and decision-making capabilities of autonomous driving systems. By dynamically correcting errors in the physical model, the graybox model can effectively address various driving scenarios and provide more precise control and decision-making support. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a flow chart of a method for safe self-evolution of an intelligent driving function algorithm based on a data closed loop, provided by an embodiment of the present invention;

[0035] Figure 2 This is a flowchart of the offline design and online implementation of the security supervision mechanism provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0036] 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 with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0037] Example 1

[0038] A safe self-evolution method for intelligent driving function algorithms based on data closed loop, such as Figure 1 Shown, including:

[0039] Using the current gray box model, the current vehicle state is obtained based on the real-time data obtained by the vehicle sensors. Based on the current vehicle state, preliminary control instructions are generated through MPC optimization.

[0040] Using the control barrier function and quadratic programming, the initial control instructions are checked and optimized for safety, ensuring that the control instructions U safe Under the premise of meeting the safety constraints, the control instruction U initial ; Among them, the safety conditions that need to be met for the vehicle to be in a safe state are: h(S) represents the control obstacle function, s represents the vehicle state, is the derivative of the control barrier function, γ is a positive number;

[0041] The control instruction U safe Translated into actual execution actions of the vehicle;

[0042] Among them, the gray box model is the sum of a vehicle state physical model that does not contain system noise and a data-driven model used to describe the nonlinear and time-varying characteristics of the vehicle. The data-driven model in the gray box model is parameter updated and optimized after a period of data collection, so as to achieve accurate prediction of the vehicle state in a complex and changing environment.

[0043] It should be noted that the functional algorithm used to perform the aforementioned steps—"using the current graybox model to determine the current vehicle state based on real-time data acquired by vehicle sensors; and based on this state, generating preliminary control instructions through MPC optimization"—is responsible for implementing the most basic intelligent driving functions, namely the system's minimum viable product, such as adaptive cruise control, lane centering, and automatic parking. It receives data from vehicle sensors and generates preliminary control instructions. The functional algorithm not only provides operational data for subsequent operations and analysis, but also obtains optimized algorithm and model parameters from the model update algorithm to continuously improve functional performance. The model update algorithm is the algorithm that performs graybox model parameter update optimization.

[0044] That is, the functional algorithm is responsible for realizing the basic functions of intelligent driving, such as adaptive cruise control, lane keeping assist, etc. The algorithm mainly includes three sub-algorithms: perception, decision-making and planning. The perception sub-algorithm uses sensors such as cameras, millimeter-wave radars, and lidars on the vehicle to obtain environmental information and vehicle status data. It can be used as a preferred method to process sensor data using deep learning algorithms to identify lane lines, vehicles in front, pedestrians, traffic signs, and other important traffic environment information. The decision-making sub-algorithm determines the target state of the vehicle based on the perception results and the predetermined driving strategy. The planning sub-algorithm uses the model predictive control (MPC) algorithm to generate control instructions U that meet the target state. initial As a preferred implementation, the MPC optimization problem can be formalized as:

[0045]

[0046] Among them, s k is the vehicle state at the kth moment, U k is the control input, q and R are weight matrices, and N is the length of the prediction time domain. The function algorithm module outputs the preliminary control instruction U initial , sent to the security supervision algorithm.

[0047] The safety supervision algorithm used to perform the aforementioned step of "safety-verifying and optimizing preliminary control instructions using a control barrier function and quadratic programming" utilizes a combination of a control barrier function (CBF) and quadratic programming (QP) to determine the vehicle's safety boundaries in real time. By transforming safety constraints into an optimization problem, QP can be used to solve for optimal control inputs, ensuring optimal vehicle performance while meeting safety requirements. This algorithm monitors the control instructions output by the functional algorithm and, if potential risks are detected, adjusts the control instructions to ensure safe vehicle operation.

[0048] This safety supervision algorithm uses the control barrier function (CBF) and quadratic programming (QP) method to verify and adjust the safety of the control instructions generated by the functional algorithm. The control barrier function h(S) is set. When h(S) ≥ 0, the vehicle is in a safe state; when h(S) < 0, the vehicle is in an unsafe state. To ensure that the vehicle state always remains within the safe set, the following conditions must be met:

[0049]

[0050] in, is the derivative of the control barrier function, γ is a positive number.

[0051] As a preferred implementation, while keeping the original control instructions as much as possible, the control input u is adjusted to meet the safety constraints. The optimization problem is:

[0052]

[0053] The safety supervision algorithm solves the above QP problem and serves as a supervisory controller such as Figure 2 , get the safe control instruction U safe , ensure that the control instruction U safe While meeting the safety constraints, the initial control instruction U initial The optimized safety control instructions will be sent to the execution module to ensure that the vehicle operation meets safety standards.

[0054] To execute the above steps, "Use the control instruction safeThe execution algorithm that converts the control instruction U into the actual execution action of the vehicle receives the control instruction U generated by the safety supervision algorithm. safe Finally, the system is responsible for translating these commands into actual vehicle actions, such as acceleration, braking, and steering. The vehicle's actuators (such as the engine control unit, braking system, and steering system) precisely execute the control signals. Furthermore, the execution algorithm collects feedback from the actuators and vehicle status information in real time to support closed-loop optimization of the system, namely, updating model parameters.

[0055] The model update algorithm combines a physical model with a data-driven learning approach to build a gray-box model of the vehicle. The physical model describes the vehicle's fundamental dynamics, while the data-driven learning approach compensates for the nonlinear and time-varying characteristics that the physical model struggles to accurately describe. Through online learning and the accumulation of historical data, the gray-box model is continuously updated, improving its accuracy and adaptability. The algorithm utilizes real-time and historical data to evaluate and adjust key parameters of the vehicle's gray-box model and functional algorithms. The optimized model parameters and algorithms are then distributed to the functional and safety oversight algorithms, enabling iterative updates of both algorithms and models.

[0056] First, a gray-box model of the vehicle is built and updated. The goal of this model update algorithm is to continuously optimize the vehicle's dynamic model so that the system can adapt to the ever-changing driving environment and complex traffic conditions.

[0057] Based on the vehicle's dynamic and kinematic characteristics, establish a physical model of the vehicle:

[0058] S k+1 =f physical (S k , U k )+w k

[0059] Among them, w k System noise represents random disturbances or uncertainties that cannot be fully described by physical models. The data-driven model uses machine learning algorithms to learn the nonlinear and time-varying characteristics of real vehicles that are difficult to accurately describe by physical models, and establishes a residual model:

[0060]

[0061] And approximated by a neural network model:

[0062] r k ≈g θ (D key )

[0063] where g θ (D key) is the residual function based on the neural network, and θ is the neural network parameter. The neural network is trained offline using a large amount of historical driving data (including but not limited to past driving behavior, vehicle status, and environmental information) to learn the mapping relationship of the residual model. In addition, real-time vehicle status, environmental information, and user operation data collected during online learning continuously optimize the neural network model to adapt to different driving environments.

[0064] The physical model and the data-driven model are combined to form a gray box model:

[0065] S k+1 =f physical (S k , U k )+g θ (D key )

[0066] In order to optimize the residual model, a deep neural network is used for fitting. The training goal is to minimize the prediction error. The loss function is defined as:

[0067]

[0068] The optimization algorithm is used to minimize the loss function and update the model parameters θ. The updated gray-box model and algorithm parameters are sent to the functional algorithm module and safety supervision module on the vehicle side to support iterative updates of the algorithm.

[0069] Through the collaborative work of these modules, the present invention achieves the safe self-evolution of intelligent driving functional algorithms. Supported by the data collection and storage module and the model update module, the system continuously acquires high-quality data and optimizes models, continuously improving the performance of real-time functional algorithms. The safety monitoring module ensures system security during the optimization process, while the execution module ensures the accurate implementation of control instructions. The entire framework forms a data-driven closed-loop mechanism, enabling the adaptive and reliable operation of intelligent driving systems in complex environments.

[0070] As a preferred implementation, the above-mentioned preliminary control instruction generation method is:

[0071] Using deep learning algorithms to process real-time data, the system identifies traffic environment information, including lane markings, vehicles ahead, pedestrians, and traffic signs. Based on this traffic environment information, the system determines the target state of the vehicle in accordance with a preset driving strategy. The real-time data includes both real-time environmental information data and real-time vehicle state data acquired by vehicle sensors.

[0072] The current gray box model is used to obtain the current vehicle state based on the above real-time data; the model predictive control MPC algorithm is used to generate preliminary control instructions that meet the target state based on the current vehicle state.

[0073] As a preferred implementation, the gray box model is represented by S k+1 =f physical (S k , U k )+g θ (D key ); where f physical (S k , U k ) represents the vehicle state physical model without system noise, g θ (D key ) is the residual function based on the neural network, θ is the parameter of the neural network; D key Key data obtained after preprocessing vehicle status, environmental information, control input and historical driving data;

[0074] The method for updating and optimizing the vehicle's gray box model is:

[0075] The vehicle state, environmental information, control input, and historical driving data are used to update the training sample set and adjust the parameters θ of the neural network. In each training sample, the neural network input data is the key data obtained by preprocessing the vehicle state, environmental information, control input, and historical driving data. The neural network output data is the difference between the actual vehicle state and the vehicle state predicted by the physical model.

[0076] The implementation method of the above preprocessing is preferably: using edge computing methods to filter and reduce noise, extract features and compress data.

[0077] This embodiment proposes a method for the safe self-evolution of intelligent driving function algorithms based on a closed-loop data system, focusing on continuous optimization and safety assurance in real-world driving environments. Compared to existing technologies, this method uses a closed-loop data system to collect and utilize data, continuously optimizing and adjusting the algorithm to achieve greater adaptability in complex and changing traffic scenarios. This method prioritizes safety as the core of the evolutionary process, rigorously monitoring and providing feedback at every step of algorithm updates and model optimization to ensure high reliability in complex environments.

[0078] Example 2

[0079] A data closed-loop-based safe self-evolution system for intelligent driving function algorithms, used to execute the above-mentioned safe self-evolution method for intelligent driving function algorithms, comprising: a function algorithm module, a safety supervision module, a data collection and storage module, a model update module, and an execution module;

[0080] Among them, the function algorithm module is used to generate preliminary control instructions; the safety supervision module is used to use the control obstacle function and quadratic programming to perform safety verification and optimization on the preliminary control instructions to obtain the control instructions U safe ; The execution module is used to control the instruction U safe The data collection and storage module is used to collect data for the model updating module to update the model parameters of the gray box model.

[0081] In this system, the functional algorithm module is responsible for implementing the most basic intelligent driving functions—the system's minimum viable product—such as adaptive cruise control, lane keeping, and automatic parking. It receives data from vehicle sensors and generates preliminary control instructions. The functional algorithm module provides operational data for subsequent operations and analysis, while also obtaining optimized algorithms and model parameters from the model update module to continuously improve functional performance.

[0082] That is, this functional algorithm module is responsible for realizing the basic functions of intelligent driving, such as adaptive cruise control, lane keeping assist, etc. This module mainly includes three sub-modules: perception, decision-making and planning. The perception sub-module uses sensors such as cameras, millimeter-wave radars, and lidars on the vehicle to obtain environmental information and vehicle status data, and uses deep learning algorithms to process sensor data to identify lane lines, vehicles in front, pedestrians, traffic signs, and other important traffic environment information. The decision-making sub-module determines the target state of the vehicle based on the perception results and the predetermined driving strategy. The planning sub-module uses the model predictive control (MPC) algorithm to generate control instructions U that meet the target state. initial As a preferred implementation, the MPC optimization problem can be formalized as:

[0083]

[0084] Among them, S k is the vehicle state at the kth moment, U k is the control input, Q and R are weight matrices, and N is the length of the prediction time domain. The function algorithm module outputs the preliminary control instruction U initial , sent to the security supervision module.

[0085] The safety supervision module utilizes a combination of control barrier functions (CBFs) and quadratic programming (QP) to determine the vehicle's safety boundaries in real time. By transforming safety constraints into an optimization problem and using QP to solve for optimal control inputs, the module ensures optimal vehicle performance while meeting safety requirements. This module monitors the control commands output by the functional algorithm module. If potential risks are detected, the safety supervision module adjusts the control commands to ensure safe vehicle operation.

[0086] The safety monitoring module of the present invention uses the control barrier function (CBF) and quadratic programming (QP) method to perform safety verification and adjustment on the control instructions generated by the functional algorithm module. The control barrier function h(S) is set. When h(S) ≥ 0, the vehicle is in a safe state; when h(S) < 0, the vehicle is in an unsafe state. To ensure that the vehicle state always remains within the safe set, the following conditions must be met:

[0087]

[0088] in, is the derivative of the control barrier function, and γ is a positive number. As a preferred implementation, while maintaining the original control instruction as much as possible, the control input U is adjusted to meet the safety constraint. The optimization problem is:

[0089]

[0090] The safety supervision module solves the above QP problem and obtains the safe control instruction U safe , ensure that the control instruction U safe While meeting the safety constraints, the initial control instruction U initial The optimized safety control instructions will be sent to the execution module to ensure that the vehicle operation meets safety standards.

[0091] The execution module receives the control instruction U generated by the security supervision module safe Finally, the safety monitoring module is responsible for translating these commands into actual vehicle actions, such as acceleration, braking, and steering. This means it receives final control commands from the safety monitoring module and directly controls vehicle movement. The vehicle's actuators (such as the engine control unit, braking system, and steering system) precisely execute these control signals. The execution module collects feedback from the actuators and vehicle status information in real time and sends this information to the data collection and storage module, supporting closed-loop optimization of the system.

[0092] Data collection and storage module: A variety of sensors are deployed on the vehicle side, including cameras, radars, and lidars, to collect vehicle status, environmental information, and user operation data in real time. The data collection and storage module uses edge computing technology to pre-process the raw data, such as filtering, compression, and feature extraction, to reduce the data volume and improve the data quality. The processed key data can be securely and efficiently uploaded to the cloud through high-speed wireless communication networks such as 5G or LTE-V. The cloud can clean, annotate, and integrate the collected data to eliminate abnormal and redundant information. This module provides high-quality data support for the model update module, and at the same time obtains operation data from the functional algorithm module, safety supervision module, and execution module to form a complete data closed loop.

[0093] The model update module combines physical models with data-driven learning methods to build a gray-box model of the vehicle. The physical model describes the vehicle's fundamental dynamics, while the data-driven learning method compensates for the nonlinear and time-varying characteristics that the physical model struggles to accurately describe. Through online learning and the accumulation of historical data, the gray-box model is continuously updated to improve its accuracy and adaptability. This module utilizes real-time and historical data to evaluate and adjust key parameters of the vehicle model and functional algorithms. The optimized model parameters and algorithms are then distributed to the functional algorithm module and the safety monitoring module, enabling iterative updates of the algorithms and models.

[0094] The collaborative work of these modules enables the safe self-evolution of intelligent driving algorithms. Supported by the data collection and storage module and the model update module, the system continuously acquires high-quality data and optimizes models, continuously improving the performance of real-time algorithms. The safety monitoring module ensures system safety during the optimization process, while the execution module ensures the accurate implementation of control commands. The entire framework forms a data-driven closed-loop mechanism, enabling the adaptive and reliable operation of intelligent driving systems in complex environments.

[0095] As a preferred implementation method, the data collection and storage module includes a vehicle side and a cloud side. The vehicle side is used to collect data and upload it to the cloud side after preprocessing, or the vehicle side is used to collect data and upload it directly to the cloud side, and the cloud side is used to preprocess the data; the cloud side is also used to update the model parameters of the gray box model based on the preprocessed data and transmit it to the functional algorithm module and the safety supervision module to provide the current vehicle status to the functional algorithm module and the safety supervision module respectively.

[0096] The data collection and storage module is responsible for collecting, preprocessing, transmitting and storing data on the vehicle side and in the cloud. The vehicle side obtains vehicle status and environmental information in real time through sensors such as cameras, radars, and lidars. vehicle , and obtain the recorded driver operation behavior data D from the execution module userAnd vehicle execution feedback data D execution Edge computing technology can be used as an optimal method to filter and reduce noise, extract features and compress data on the original data to obtain key data D key Through high-speed wireless communication network, the pre-processed key data D key Upload to the cloud securely and efficiently. The cloud cleans and annotates the uploaded data, removes abnormal and redundant information, and stores the processed data in the cloud database for use by the model update module.

[0097] Other related technical solutions are the same as those in embodiment 1 and will not be described in detail here.

[0098] It should be noted that the data collection and storage module and edge computing technology can solve the real-time performance issues in intelligent driving systems. Vehicle and environmental data are collected in real time through a variety of sensors, and edge computing is used for preprocessing, reducing the amount of data transmitted and thus improving the real-time performance of data processing. Furthermore, by combining a gray-box model with a physical model and a data-driven model, the system's robustness in complex traffic environments is addressed, enabling continuous system optimization and improving performance in changing environments. The safety monitoring module ensures that the vehicle remains within a safe range of motion during the optimization process, thereby improving vehicle control accuracy. The data collection and storage module supports comprehensive and real-time data collection through a variety of sensors and high-speed communication technologies, ensuring that the system can obtain real-time data, supporting subsequent model updates and algorithm optimization, thereby enhancing the overall intelligence level and execution efficiency of the system.

[0099] In summary, the system framework proposed in this embodiment aims to address the challenges of insufficient data and limited scenarios in the early stages of intelligent driving system development, as well as the inability of fixed rules and static parameter models to meet requirements in complex and changing real-world driving environments. By introducing a closed-loop data mechanism and a safety oversight module, the framework enables continuous optimization and adaptive evolution of functional algorithms while ensuring system safety and robustness. This means that while ensuring safety, the framework enables continuous optimization and adaptive evolution of system functional algorithms, improving the vehicle's adaptability and reliability in complex traffic environments.

[0100] Example 3

[0101] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0102] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The processor may be a central processing unit (CPU), or other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The memory may be used to store computer programs and / or modules, and the processor may perform various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory.

[0103] The relevant technical solutions are the same as above and will not be repeated here.

[0104] Example 4

[0105] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0106] Specifically, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0107] The relevant technical solutions are the same as above and will not be repeated here.

[0108] Example 5

[0109] A computer program product or computer program includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of the method of the above-mentioned embodiment of the present application.

[0110] The relevant technical solutions are the same as above and will not be repeated here.

[0111] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A safe self-evolution method for intelligent driving function algorithm based on data closed loop, characterized in that: include: Using the current gray box model, the current vehicle status is obtained based on the real-time data obtained by the vehicle sensors; Based on the current vehicle status, preliminary control instructions are generated through MPC optimization; Using the control obstacle function and quadratic programming, the initial control instructions are checked and optimized for safety, ensuring that the control instructions U safe Under the premise of meeting the safety constraints, the control instruction U initial ; Among them, the safety conditions that need to be met for the vehicle to be in a safe state are: h(S) represents the control obstacle function, S represents the vehicle state, is the derivative of the control barrier function, γ is a positive number; The control command U safe Translated into actual execution actions of the vehicle; Among them, the gray box model is the sum of a physical model of the vehicle state that does not contain system noise and a data-driven model used to describe the nonlinear and time-varying characteristics of the vehicle. The data-driven model in the gray box model obtains parameter update optimization after a period of data collection so as to achieve accurate prediction of the vehicle state in a complex and changing environment.

2. The method for safe self-evolution of intelligent driving function algorithm according to claim 1, characterized in that: The initial control instruction generation method is: Using deep learning algorithms to process real-time data, identify traffic environment information, including lane lines, vehicles ahead, pedestrians, and traffic signs; based on the traffic environment information, determine the target state of the vehicle according to a preset driving strategy; wherein the real-time data includes real-time environment information data and real-time vehicle state data acquired by vehicle sensors; The current gray box model is used to obtain the current vehicle state based on the real-time data; the model predictive control MPC algorithm is used to generate preliminary control instructions that meet the target state based on the current vehicle state.

3. The method for safe self-evolution of intelligent driving function algorithm according to claim 1, characterized in that: Optimize the preliminary control instructions to obtain the control instruction U safe The quadratic programming optimization model is: IN min ≤U≤U max Where U initial represents the preliminary control command, U represents the control command independent variable, and f(S, U) represents the vehicle state physical model without system noise.

4. The method for safe self-evolution of intelligent driving function algorithm according to claim 1, characterized in that: The gray box model is represented by S k+1＝ f physical (S k , U k )+g θ (D key ), where f physical (S k , U k ) represents the physical model of the vehicle state without system noise, g θ (D key ) is the residual function based on the neural network, θ is the parameter of the neural network; D key Key data obtained after preprocessing vehicle status, environmental information, control input and historical driving data; The method for updating and optimizing the gray box model of the vehicle is: The vehicle state, environmental information, control input and historical driving data are used to update the training sample set and adjust the parameter θ of the neural network; wherein in each training sample, the key data obtained after preprocessing the vehicle state, environmental information, control input and historical driving data are used as the input data of the neural network; and the difference between the actual vehicle state and the vehicle state predicted by the physical model is used as the output data of the neural network.

5. The method for safe self-evolution of intelligent driving function algorithm according to claim 1, characterized in that: The preprocessing is implemented by using edge computing methods to filter and reduce noise, extract features, and compress data.

6. A safe self-evolution system of intelligent driving function algorithm based on data closed loop, characterized in that: A method for executing the safe self-evolution of an intelligent driving function algorithm as claimed in any one of claims 1 to 5, comprising: a function algorithm module, a safety supervision module, a data collection and storage module, a model update module and an execution module; Wherein, the functional algorithm module is used to generate preliminary control instructions; The safety supervision module is used to use the control obstacle function and quadratic programming to perform safety verification and optimization on the preliminary control instructions to obtain the control instructions U safe ; The execution module is used to convert the control instruction U safe Translated into actual execution actions of the vehicle; The data collection and storage module is used for data collection so as to provide the model updating module with data for updating the model parameters of the gray box model.

7. The intelligent driving function algorithm safe self-evolution system as claimed in claim 6, characterized in that: The data collection and storage module includes a vehicle side and a cloud side, wherein the vehicle side is used to collect data and upload it to the cloud side after preprocessing, or the vehicle side is used to collect data and upload it directly to the cloud side, and the cloud side is used to preprocess the data; The cloud is also used to update the model parameters of the gray box model based on the preprocessed data and transmit them to the functional algorithm module and the safety supervision module to provide the current vehicle status to the functional algorithm module and the safety supervision module respectively.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed by a processor, the device where the storage medium is located is controlled to perform the steps of any one of the methods according to claims 1 to 5.

10. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images