Automatic driving control method based on elastic safety critical control

Abstract

The invention relates to an automatic driving control method based on elastic safety critical control. The method comprises the steps that the vehicle state, the reference path and the environment information of automatic driving are obtained in real time; establishing a control-oriented automatic driving vehicle dynamics controlled model, and establishing an optimization problem considering time-varying safety constraints; designing an elastic safety critical control model: estimating time-varying disturbance, converting safety and stability into constraints through an elastic control barrier function of a self-adaptive penalty parameter, establishing a unified constrained optimization framework, and integrating safety constraints, stability constraints and control input constraints into a quadratic optimization problem; and solving a secondary optimization problem of the elastic safety critical control model in real time to obtain safety control input of the automatic driving vehicle. Compared with the prior art, the method has the advantages that the vehicle can accurately track the reference path under the time-varying disturbance, and the system safety and stability are ensured.

Description

Technical Field

[0001] This invention relates to the field of safety-critical system control, and in particular to an autonomous driving control method based on flexible safety-critical control. Background Technology

[0002] The motion control system of autonomous vehicles is a typical safety-critical system, and its control performance directly affects driving safety and occupant protection. Such systems require maintaining the vehicle's state within safety boundaries and accurately tracking the expected trajectory in complex dynamic traffic environments, while effectively resisting various internal and external disturbances. To achieve this goal, safety-critical control is widely considered an effective method for building reliable autonomous driving systems. This method formalizes safety specifications into control constraints and embeds them into the design of real-time control laws, thereby theoretically ensuring that the system state never violates safety boundaries, providing stringent safety guarantees for tasks such as path tracking, obstacle avoidance, and stable driving.

[0003] Despite the theoretical advantages of safety-critical control methods, their widespread application in autonomous vehicles still faces practical challenges and implementation limitations. The main reasons can be summarized in two points: 1. Real-world vehicle systems face the combined challenges of unstructured disturbances and high-order relative safety constraints. Unstructured disturbances (such as sudden wind disturbances, complex road surface adhesion changes, and unmodeled dynamic characteristics) are difficult to describe with precise models, and together with sensor noise and actuator errors, they constitute compound disturbances that directly affect the satisfaction of safety constraints. Meanwhile, safety constraints involved in vehicle dynamics (such as lateral position error and heading angle deviation) often have a high relative order relative to the control input, making control design more complex. Traditional methods struggle to simultaneously ensure accurate constraint satisfaction and system real-time performance under disturbances.

[0004] 2. Existing adaptive safety-critical control relies on adjusting the penalty parameter to maintain feasibility, but changes in this parameter itself introduce new problems. Under strong disturbances or model mismatch, the penalty parameter may fluctuate drastically or continuously increase in order to maintain constraint satisfaction. This may not only reduce control smoothness and affect ride comfort, but may also make the underlying optimization problem numerically ill-conditioned or even unsolvable, leading to a sharp drop in control performance or failure of safety mechanisms. This severely restricts the reliability and robustness of this method in real vehicle systems.

[0005] A search revealed Chinese invention patent application publication number CN118597106A, which discloses a robust tracking control method for the safe motion state of an unmanned vehicle. The method includes the following steps: Step 1: Constructing motion state equations describing velocity and acceleration using a longitudinal motion model of the unmanned vehicle; Step 2: Designing motion state constraints and control obstacle functions based on the motion state equations, proposing a longitudinal motion safety controller, and optimizing the solution to achieve a longitudinal safe motion state for the unmanned vehicle; Step 3: Constructing a system model of the unmanned vehicle containing a path curvature extension state based on the longitudinal safe motion state and applying a lateral dynamic tracking error system; Step 4: Designing a path curvature extension state observer based on the extended state system model, and using the lateral path tracking error of the unmanned vehicle to achieve path curvature observation; Step 5: Designing a lateral path tracking robust controller based on the path curvature observation to compensate for the influence of path curvature on the motion of the unmanned vehicle, improving the robustness and safety of the tracking motion. This existing patent application suffers from problems such as a tendency for safe tracking accuracy to decrease under large disturbances, and difficulty in simultaneously balancing multiple objectives including safe tracking, stable control, and control costs.

[0006] Improving the safety tracking accuracy and stability of autonomous driving has become a technical problem that needs to be solved. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide an autonomous driving control method based on flexible safety-critical control.

[0008] The objective of this invention can be achieved through the following technical solutions: According to one aspect of the present invention, an autonomous driving control method based on flexible safety-critical control is provided, the method comprising: Real-time acquisition of vehicle status, reference path, and environmental information for autonomous driving; A control-oriented dynamic controlled model of autonomous vehicle is established based on vehicle state and reference path, and an optimization problem considering time-varying safety constraints is established. Based on environmental information, an elastic safety-critical control model is designed to solve the safety motion control problem of autonomous vehicles: time-varying disturbances are estimated by a finite-time disturbance observer, and safety and stability are transformed into safety constraints and stability constraints respectively by combining an elastic control barrier function with adaptive penalty parameters. A unified constrained optimization framework is established, integrating safety constraints, stability constraints, and control input constraints into a quadratic optimization problem. The quadratic optimization problem of the elastic safety-critical control model is solved in real time to obtain the safety control input for autonomous vehicles, so that the vehicles can accurately track the reference path under time-varying disturbances and ensure the safety and stability of the system.

[0009] As a preferred technical solution, the controlled dynamics model of the autonomous vehicle includes: Based on vehicle dynamics principles, a general second-order motion control system is introduced, as shown in the following equation: , , in, These are the position and speed of the motion control system, respectively. The time derivative; It is the desired reference signal, i.e., the target that we want to track; Control input; It is a known nonlinear term; It is a disturbance term; For parameters related to interference; t is time; The specific physical variables are defined as follows: , , , The matrix representation is as follows: , , , , , in, , These are the lateral position error and its derivative, respectively. , These are the yaw angle error and its derivative, respectively. , These are state variables, representing the lateral position error, yaw angle error, and their derivatives, respectively. It is a control input, representing the front wheel steering angle; It is a reference value, representing the expected yaw rate of the reference path; m For the overall vehicle quality, These are the lateral stiffness of the front and rear axle tires, respectively. These are the distances from the center of mass to the front and rear axles, respectively. The moment of inertia of the vehicle body during yaw. This represents the longitudinal velocity.

[0010] As a preferred technical solution, the optimization problem considering time-varying security constraints is specifically as follows: Design a reasonable This causes tracking error Asymptotic stability ensures accurate path tracking; At the same time, the resilient safety-critical control model must meet safety constraints and account for tracking errors. Limited to the prescribed scope.

[0011] As a preferred technical solution, the resilient safety-critical control model includes: Designed to provide a finite-time perturbation observer for estimating time-varying perturbations and their estimation error bounds; An elastic control barrier function with positive time-varying penalty parameters is established based on a finite-time perturbation observer, and the non-negativity of the penalty parameters is guaranteed by input-output linearization of the control barrier function of the auxiliary dynamic system. A unified constrained optimization framework is established with the goal of minimizing control costs. Safety constraints, stability constraints, and control input constraints are integrated into a quadratic optimization problem. By solving the quadratic optimization problem, an input control that simultaneously satisfies safety constraints, stability constraints, and minimizes control costs is obtained.

[0012] As a preferred technical solution, the finite-time perturbation observer is specifically: Constructing the dynamic equations for the auxiliary state and estimating the time-varying perturbation: , , in, Indicates the auxiliary state of the observer; This indicates an estimate of the time-varying perturbation. This represents the estimation error, i.e., the system disturbance. d The difference between the estimated time-varying perturbation and the actual time-varying perturbation; This represents the super-twisting term, used to achieve finite-time convergence. ; Indicates the observer parameters, Let represent the hyperparameters of the higher-order sliding mode term, and α > 0, β > 0; For a dependent system state x positive function, and The relationship is , , L min for The lower bound; u The input is the control input; f and g are known nonlinear terms.

[0013] As a preferred technical solution, the estimation error of the system disturbance The estimation converges to a bounded neighborhood within a finite time; the extent of this neighborhood is denoted as the estimation error bound. , means as follows: , , , in, This represents the neighborhood where the estimation error eventually converges; , For calculation Two intermediate variables; Let be the relaxation constant; Represents system disturbance d The upper bound of the derivative.

[0014] As a preferred technical solution, the elastic control barrier function for the time-varying penalty parameter is specifically as follows: Define a continuously differentiable function To express safety regulations, use the following formula: , , Wherein, the relative order of the above equation is 2, if there exists a control input satisfy That is, to ensure the position Not exceeding the upper limit It is not lower than the lower limit. This ensures compliance with safety standards; Introducing a time-varying penalty parameter , The sequence of high-order elastic control barrier functions is constructed as follows: , , , in, These are the basic safety constraint function, the first-order elastic safety constraint function, and the second-order elastic safety constraint function, respectively. They are respectively , The derivative; They represent about K-class functions.

[0015] As a preferred technical solution, in order to have a positive time-varying penalty parameter , Stable to a sufficiently small positive value, defined The control of the Lyapunov function makes Stabilizes to a sufficiently small positive value; because It is not infinitesimal, so that The stability condition for stabilizing to a sufficiently small positive value is directly set in optimization problems.

[0016] As a preferred technical solution, establishing a unified constrained optimization framework includes: Define a continuously differentiable Lyapunov function to describe stability constraints. : , Where the relative order of the stability constraint is 2, we can obtain: , in, This represents the unknown portion of the system disturbance effect on the Lyapunov function; Considering the unknown parts caused by concentrated disturbances If any control input satisfy: , The original higher-order control Lyapunov function condition is then guaranteed.

[0017] As a preferred technical solution, the secondary optimization problem includes: Minimize the objective function: , Safety constraints: , Stability constraints: , Control input constraints: , , Among them, the control input for secondary optimization Including system control input Virtual input of auxiliary systems Penalty parameters and relaxation factors of stability conditions , ; P This is a weight matrix used to control the cost weights of each optimization variable; Let be the coefficient vector of the first-order terms. for Stabilize to a sufficiently small target value, where T is the vector transpose; Indicates an adjustable parameter; For the safety constraint matrix; The stability constraint matrix; For the safety constraint threshold vector; For the stable constraint threshold vector; To control the input constraint matrix; To control the input constraint threshold vector.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1) The autonomous driving safety control method of the present invention accurately captures time-varying disturbances through a finite-time disturbance observer and combines it with an elastic control obstacle function with adaptive penalty parameters. This ensures that the vehicle state is always kept within the safety boundary and avoids control rigidity caused by overly strict constraints, thus ensuring safety even under complex disturbance conditions. At the same time, it integrates safety, stability and control input constraints into a secondary optimization framework. While ensuring the dynamic stability of the system, it achieves accurate tracking of the reference path and smooth output of control quantities. This not only improves driving accuracy and control efficiency, but also simplifies the adaptation process under different conditions through modular design, taking into account safety, stability and ease of engineering implementation.

[0019] 2) By constructing a second-order motion control system that integrates vehicle dynamics, core state variables such as lateral position error and yaw angle error are precisely correlated with control inputs such as front wheel steering angle. Nonlinear functions are defined in conjunction with actual physical parameters such as vehicle mass and tire lateral stiffness. This achieves accurate modeling of the dynamic characteristics of autonomous vehicles and makes the model both universal and physically interpretable. At the same time, the disturbance term is included in the model, providing an accurate dynamic basis for subsequent disturbance observation and elastic constraints. This makes the model more consistent with actual driving conditions, improves the adaptability of the control method to complex vehicle dynamics and road disturbances, and ensures the engineering practicality and control accuracy of the control strategy.

[0020] 3) Accurate estimation and error bound quantification of time-varying disturbances are achieved through a finite-time disturbance observer. The high-order sliding mode term ensures that the estimation error converges to a eventually uniformly bounded neighborhood within a finite time, which not only ensures the timeliness and reliability of disturbance sensing, but also provides a clear error buffer basis for subsequent constraint design. 4) By combining the elastic control obstacle function with positive penalty parameters and the auxiliary dynamics system, the vehicle state is always kept within the safety boundary through high-order constraints, and the control rigidity is avoided by adapting the parameters to complex conditions. Furthermore, the safety, stability and control cost constraints are integrated through a unified secondary optimization framework. While meeting the safety and stability requirements, the optimal solution of the control input is achieved. Ultimately, the autonomous driving system can accurately track the path under time-varying disturbances, and also take into account the smoothness and robustness of control, which greatly improves the driving safety and control practicality in complex scenarios. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the autonomous driving control method of the present invention; Figure 2 This is a schematic diagram of the structure of the automatic driving control system in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0023] This invention addresses the shortcomings of existing technologies by proposing an autonomous driving safety control method based on an elastic control barrier function. By utilizing software design such as the elastic control barrier function and finite-time disturbance observer, as well as hardware such as on-board sensor signals and controllers, sufficient conditions for an elastic safety system are established. This effectively reduces the range of variation of the penalty parameters, lessens the dependence on their independent adaptability, and improves the safety, stability, and robustness of the original closed-loop system under disturbances.

[0024] This embodiment relates to an autonomous driving control method based on flexible safety-critical control. By constructing a flexible safety motion control system for autonomous vehicles, establishing a control-oriented dynamic controlled model and optimization problem for autonomous vehicles, and designing a flexible safety control method, the robustness of the original closed-loop system in terms of safety and stability is improved.

[0025] like Figure 1 The method includes the following steps: S1 constructs a flexible safety motion control system for autonomous vehicles, obtaining real-time vehicle status information from onboard communication and sensors, and real-time reference path information and environmental information from perception sensors.

[0026] S2. Establish a controllable dynamics model and optimization problem for autonomous vehicles. Simplify the controlled object model based on the vehicle dynamics mechanism and, in conjunction with safety constraints, establish an optimization problem that considers time-varying constraints.

[0027] S3. Design a resilient safety control method. Estimate time-varying disturbances using a finite-time disturbance observer and combine it with a control barrier function with adaptive penalty parameters to describe the expression of time-varying safety constraints.

[0028] S4 is solved in real time to obtain the safety control input of the autonomous vehicle, so that the vehicle can accurately track the reference path under time-varying disturbances and ensure the safety and stability of the system.

[0029] In S1, the flexible safety motion control system for autonomous vehicles, such as Figure 2As shown, the flexible safety motion control system for autonomous vehicles includes perception sensors, a controlled dynamic model of the autonomous vehicle, a flexible safety-critical control module, an optimization module considering time-varying constraints, and an onboard communication and sensor module. Specifically, the reference path required by the controlled dynamic model of the autonomous vehicle is input from the perception sensors, and the required vehicle state information is input from the onboard communication and sensor module; the environmental information required by the flexible safety-critical control module is input from the perception sensors; and the safety constraints required by the optimization module considering time-varying constraints are expressed and input by the flexible safety-critical control module.

[0030] In S2, based on vehicle dynamics, a general second-order motion control system is introduced, and this general second-order motion control system is simplified to obtain an elastic safety-critical control model, specifically: , , in, These are the position and speed of the motion control system, respectively. The time derivative (i.e., the rate of change of position); It is the desired reference signal, i.e., the target that we want to track; It is a control input; It is a known nonlinear term; It is a disturbance term; t represents the interference-related parameters; t represents time.

[0031] For the sake of brevity, , , , They are respectively represented as , , ; The specific physical variables are defined as follows: , , , The matrix representation is as follows: , , , , , in, , These are the lateral position error and its derivative, respectively. , These are the yaw angle error and its derivative, respectively. , These are state variables, representing the lateral position error, yaw angle error, and their derivatives, respectively. It is a control input, representing the front wheel steering angle; It is a reference value, representing the expected yaw rate of the reference path; m For the overall vehicle quality, These are the lateral stiffness of the front and rear axle tires, respectively. These are the distances from the center of mass to the front and rear axles, respectively. The moment of inertia of the vehicle body during yaw. This represents the longitudinal velocity.

[0032] The control objective of this invention is to design a reasonable This causes tracking error Asymptotic stability ensures accurate path tracking. Simultaneously, the resilient, safety-critical control model must satisfy safety constraints to minimize tracking errors. The control objectives and safety constraints are limited to a specified range. Therefore, the control objectives and safety constraints are expressed in the following form: , , in, The above inequality represents the upper and lower limits of the tracking error limit, and represents the safety-critical specifications.

[0033] In S3, a resilient safety-critical control model is proposed to solve the safety motion control problem of autonomous vehicles. The design of the resilient safety-critical control model includes a finite-time perturbation observer with estimation error bounds, a resilient control barrier function based on the perturbation observer, and a unified constrained optimization framework. Safety constraints consist of a safety specification described by the resilient control barrier function and the non-negativity of the penalty parameters enforced by the general control barrier function, while stability constraints consist of a control objective described by a Lyapunov function based on the perturbation observer and the convergence of the penalty parameters enforced by the general Lyapunov function.

[0034] The improved sufficient condition combines the estimated disturbance and its error bound provided by the finite-time perturbation observer, forming a unified constrained optimization framework that further strengthens the traditional adaptive mechanism. Furthermore, the proposed framework provides better system resilience by reducing the range of parameter variations.

[0035] (1) For a finite-time perturbation observer, the design is as follows: , , in, Indicates the auxiliary state of the observer; This indicates an estimate of the time-varying perturbation. This represents the estimation error, the difference between the actual disturbance d and the estimated time-varying disturbance; This represents the super-twisting term, used to achieve finite-time convergence. ; Indicates the observer parameters, Let α and β represent the hyperparameters of the higher-order sliding mode term, where α and β are the design parameters of the observer, and α > 0 and β > 0. , It is an intermediate function in the derivation process. For a dependent system state x positive function, and The relationship is , , L min for The lower bound; For controlling input.

[0036] The purpose of a finite-time perturbation observer is to accurately estimate perturbations in a system within a finite time, which can then be used to estimate time-varying perturbations. Design control strategies (such as counteracting the effects of disturbances).

[0037] As can be derived, a finite-time disturbance observer can detect system disturbances. estimation error The estimation converges to a neighborhood within a finite time, and this neighborhood eventually becomes uniformly bounded. Furthermore, the estimation error bound is defined. It can be represented as follows: , , , in, This represents the neighborhood where the estimation error eventually converges; , For calculation Two intermediate variables; Let be the relaxation constant; This represents the upper bound of the d-derivative of the system disturbance.

[0038] (2) For the elastic control barrier function based on the disturbance observer, the design is as follows: Define a continuously differentiable function To express safety regulations, as shown in the following formula: , , Wherein, the relative order of the above equation is 2, which means: if there is a control input satisfy That is, to ensure the position Not exceeding the upper limit It is not lower than the lower limit. This ensures that safety standards are met.

[0039] According to the definition of a higher-order elastic control barrier function, it has a positive time-varying penalty parameter. , The function sequence is as follows: , , , in, for , The derivative; Indicates about K-class functions.

[0040] A positive time-varying penalty parameter was introduced. , By adjusting these parameters, the stringency of obstacle constraints can be flexibly controlled.

[0041] To ensure the non-negativity of the penalty parameter, Defined as a simple input-output linearized auxiliary dynamics system, the control barrier function is shown below. , in, It is a virtual state variable. It is a virtual control input, the matrix is , The auxiliary system becomes .

[0042] The general control barrier function conditions are established as follows. , Among them, K-type functions It was chosen to be in linear form; They are virtual state variables. The Lie derivatives of F and G. Furthermore, since they cannot be further differentiated, It will be set directly in the optimization problem.

[0043] Therefore, from the above derivation, we can conclude that , , , , in, , , , These are the remaining items for Director Li. Indicates system disturbance The unknown part in the derivation.

[0044] It is clear that at the boundary of the safe set, This could lead to a violation of the above rules, which traditional control barrier functions would ignore. Furthermore, although in the above equation... It is unknown, but it depends only on system disturbances. And with penalty parameters , Irrelevant. This invention will use the aforementioned finite-time perturbation observer to estimate... This counteracts the effects of the disturbance.

[0045] As can be derived, the following inequality serves as a sufficient condition to ensure that the safe set remains unchanged forward: , in, , These are the estimated values ​​of the unknown portion of the system disturbance applied to the elastic control barrier function and the estimation error bound, respectively.

[0046] (3) For the unified constrained optimization framework, the design is as follows: Define a continuously differentiable Lyapunov function to describe stability constraints, as follows: In order to achieve the control objective.

[0047] , The relative order of the aforementioned stability constraint is also 2, therefore, , in, This represents the estimated value of the unknown part of the system disturbance applied to the Lyapunov function.

[0048] Considering the estimated time-varying perturbation If any control input satisfy, , The original higher-order control Lyapunov function condition is then guaranteed.

[0049] In addition, it is hoped that the penalty parameters will be adjusted. , Keep it small enough to improve the feasibility of optimization. Before proceeding with overall optimization, it is necessary to supplement... , Stability constraints, i.e., seeking to... , Stabilize to a sufficiently small positive value , .

[0050] definition The control Lyapunov function is Therefore, , in, It is an adjustable parameter. , They are respectively Lie derivatives of F and G.

[0051] Similarly, due to It is not differentiable, therefore The stability conditions will be set directly in the optimization problem.

[0052] After analyzing and obtaining the improved safety and stability conditions described above, the motion control problem is reformulated as the following quadratic optimization form. The control quantity (i.e., input control) that satisfies the minimization objective function and the safety and stability constraints is then solved through optimization: Minimize the objective function: , Safety constraints: , Stability constraints: , Control input constraints: , The specific matrix is ​​as follows: , , , Among them, the overall optimized control input Including system control input Virtual input of auxiliary systems Penalty parameters and relaxation factors of stability conditions , ; , , , All of these indicate adjustable parameters; P This is the weight matrix (a diagonal matrix that controls the cost weights of each optimization variable). This is the vector of coefficients for the first-order terms; for Stabilize to a sufficiently small target value; The safety constraint matrix is ​​a coefficient matrix derived from safety specifications, where each row corresponds to a linear expression of a safety constraint. The stability constraint matrix is ​​a coefficient matrix derived from the system stability conditions, with each row corresponding to a linear constraint on stability. Let be the safety constraint threshold vector, representing the upper limit of the safety constraints and indicating the control input for overall optimization. All security boundary restrictions must be met; The stability constraint threshold vector is the upper limit of the stability constraint, ensuring that the optimized system meets the stability requirements. The control input constraint matrix is ​​a coefficient matrix derived from the physical constraints of the control input, corresponding to the value constraints of the control input u. The control input constraint threshold vector is the upper limit of the control input, which limits the control input u from exceeding the range allowed by the hardware and operating conditions.

[0053] The autonomous driving safety control method considers the need for the motion control system of autonomous vehicles to handle unstructured disturbances and high relative-order safety constraints. This method enhances the resilience of the penalty parameter beyond its independent adaptability by integrating a finite-time disturbance observer. For the controlled system, the general safety conditions for the adaptive control barrier function are derived, and sufficient conditions for a resilient safety system are established by introducing a finite-time disturbance observer. This effectively reduces the variation range of the penalty parameter, alleviates its dependence on independent adaptability, improves the system's performance under disturbances, and ultimately enhances the robustness of the original closed-loop system in terms of safety and stability.

[0054] This embodiment also relates to an autonomous driving control system based on flexible safety-critical control, such as... Figure 2 As shown, the system includes perception sensors, a controlled dynamics model of an autonomous vehicle, a flexible safety-critical control module, an optimization module considering time-varying constraints, and an onboard communication and sensor module. Specifically, the reference path required by the controlled dynamics model of the autonomous vehicle is input from the perception sensors, and the required vehicle state parameter information is input from the onboard communication and sensor module; the environmental information required by the flexible safety control module is input from the perception sensors; and the safety constraints required by the optimization problem module considering time-varying constraints are expressed and input by the flexible safety-critical control module.

[0055] The sufficient condition for the system to be resiliently safe is derived by introducing a finite-time perturbation observer and based on a control barrier function with adaptive penalty parameters, ultimately describing it as an inequality constraint in an optimization problem. The performance of the original closed-loop system is improved by the resiliently safe-critical control model, and the effectiveness of the proposed method is verified by comparison with traditional algorithms.

[0056] The electronic device of this invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0057] Multiple components in the device are connected to the I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0058] The processing unit performs the various methods and processes described above. For example, in some embodiments, the methods may be implemented as computer software programs tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of the methods described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute the methods by any other suitable means (e.g., by means of firmware).

[0059] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0060] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0061] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0062] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An autonomous driving control method based on flexible safety-critical control, characterized in that, The method includes: Real-time acquisition of vehicle status, reference path, and environmental information for autonomous driving; A control-oriented dynamic controlled model of autonomous vehicle is established based on vehicle state and reference path, and an optimization problem considering time-varying safety constraints is established. Based on environmental information, an elastic safety-critical control model is designed to solve the safety motion control problem of autonomous vehicles: time-varying disturbances are estimated by a finite-time disturbance observer, and safety and stability are transformed into safety constraints and stability constraints respectively by combining an elastic control barrier function with adaptive penalty parameters. A unified constrained optimization framework is established, integrating safety constraints, stability constraints, and control input constraints into a quadratic optimization problem. The quadratic optimization problem of the elastic safety-critical control model is solved in real time to obtain the safety control input for autonomous vehicles.

2. The autonomous driving control method based on flexible safety-critical control according to claim 1, characterized in that, The controlled model of autonomous vehicle dynamics includes: Based on vehicle dynamics principles, a general second-order motion control system is introduced, as shown in the following equation: ， ， in, These are the position and speed of the motion control system, respectively. The time derivative; It is the desired reference signal, i.e., the target that we want to track; Control input; It is a known nonlinear term; It is a disturbance term; For parameters related to interference; t is time; The specific physical variables are defined as follows: ， ， ， The matrix representation is as follows: ， ， ， ， ， in, , These are the lateral position error and its derivative, respectively. , These are the yaw angle error and its derivative, respectively. , These are state variables, representing the lateral position error, yaw angle error, and their derivatives, respectively. It is a control input, representing the front wheel steering angle; It is a reference value, representing the expected yaw rate of the reference path; m For the overall vehicle quality, These are the lateral stiffness of the front and rear axle tires, respectively. These are the distances from the center of mass to the front and rear axles, respectively. The moment of inertia of the vehicle body during yaw. This represents the longitudinal velocity.

3. The autonomous driving control method based on flexible safety-critical control according to claim 2, characterized in that, The optimization problem that considers time-varying security constraints is specifically as follows: Design a reasonable This causes tracking error Asymptotic stability ensures accurate path tracking; At the same time, the resilient safety-critical control model must meet safety constraints and account for tracking errors. Limited to the prescribed scope.

4. The autonomous driving control method based on flexible safety-critical control according to claim 1, characterized in that, The aforementioned resilient safety-critical control model includes: Designed to provide a finite-time perturbation observer for estimating time-varying perturbations and their estimation error bounds; An elastic control barrier function with positive time-varying penalty parameters is established based on a finite-time perturbation observer, and the non-negativity of the penalty parameters is guaranteed by input-output linearization of the control barrier function of the auxiliary dynamic system. A unified constrained optimization framework is established with the goal of minimizing control costs. Safety constraints, stability constraints, and control input constraints are integrated into a quadratic optimization problem. By solving the quadratic optimization problem, an input control that simultaneously satisfies safety constraints, stability constraints, and minimizes control costs is obtained.

5. The autonomous driving control method based on flexible safety-critical control according to claim 4, characterized in that, The finite-time perturbation observer is specifically: Constructing the dynamic equations for the auxiliary state and estimating the time-varying perturbation: ， ， in, Indicates the auxiliary state of the observer; This indicates an estimate of the time-varying perturbation. This represents the estimation error, i.e., the system disturbance. d The difference between the estimated time-varying perturbation and the actual time-varying perturbation; This represents the super-twisting term, used to achieve finite-time convergence. ; Indicates the observer parameters, Let represent the hyperparameters of the higher-order sliding mode term, and α > 0, β > 0; For a dependent system state x positive function, and The relationship is , , L min for The lower bound; u The input is the control input; f and g are known nonlinear terms.

6. The autonomous driving control method based on flexible safety-critical control according to claim 5, characterized in that, The estimation error of the system disturbance The estimation converges to a bounded neighborhood within a finite time; the extent of this neighborhood is denoted as the estimation error bound. , means as follows: ， ， ， in, This represents the neighborhood where the estimation error eventually converges; , For calculation Two intermediate variables; Let be the relaxation constant; Represents system disturbance d The upper bound of the derivative.

7. The autonomous driving control method based on flexible safety-critical control according to claim 4, characterized in that, The elastic control barrier function for the time-varying penalty parameter is as follows: Define a continuously differentiable function To express safety regulations, use the following formula: ， ， Wherein, the relative order of the above equation is 2, if there exists a control input satisfy That is, to ensure the position Not exceeding the upper limit It is not lower than the lower limit. This ensures compliance with safety standards; Introducing a time-varying penalty parameter , The sequence of high-order elastic control barrier functions is constructed as follows: ， ， ， in, These are the basic safety constraint function, the first-order elastic safety constraint function, and the second-order elastic safety constraint function, respectively. They are respectively , The derivative; They represent about K-class functions.

8. The autonomous driving control method based on flexible safety-critical control according to claim 7, characterized in that, In order to have a positive time-varying penalty parameter , Stable to a sufficiently small positive value, defined The control of the Lyapunov function makes Stabilizes to a sufficiently small positive value; because It is not infinitesimal, so that The stability condition for stabilizing to a sufficiently small positive value is directly set in optimization problems.

9. The autonomous driving control method based on flexible safety-critical control according to claim 4, characterized in that, Establishing a unified constrained optimization framework includes: Define a continuously differentiable Lyapunov function to describe stability constraints. : ， Where the relative order of the stability constraint is 2, we can obtain: ， in, This represents the unknown portion of the system disturbance effect on the Lyapunov function; Considering the unknown parts caused by concentrated disturbances If any control input satisfy: ， The original higher-order control Lyapunov function condition is then guaranteed.

10. The autonomous driving control method based on flexible safety-critical control according to claim 9, wherein the quadratic optimization problem includes: Minimize the objective function: , Safety constraints: , Stability constraints: , Control input constraints: , ， Among them, the control input for secondary optimization Including system control input Virtual input of auxiliary systems Penalty parameters and relaxation factors of stability conditions , ; P This is a weight matrix used to control the cost weights of each optimization variable; Let be the coefficient vector of the first-order terms. for Stabilize to a sufficiently small target value, where T is the vector transpose; Indicates an adjustable parameter; For the safety constraint matrix; The stability constraint matrix; For the safety constraint threshold vector; For the stable constraint threshold vector; To control the input constraint matrix; To control the input constraint threshold vector.

Images