A vehicle path safety boundary quantification method for actuator failure

Abstract

The application constructs a vehicle path safety boundary quantification method for actuator failure. Firstly, the fault self-adaptive input weight matrix is used to realize the active inhibition and fault tolerance of the control allocation layer for asymmetric fault risk. By constructing a robust error invariant set optimization framework based on linear matrix inequality, the complex nonlinear dynamic fault influence is successfully converted into a calculable robust performance gain, thereby quantifying the exact physical space error safety boundary, providing a hard constraint for the upper layer of autonomous driving path planning. The real-time feedback control law ensures that the actual running trajectory of the vehicle converges within the quantified safety boundary under the influence of external disturbance and actuator performance degradation. The method breaks the limitations of traditional fixed safety boundary and significantly improves the dynamic obstacle avoidance ability and running safety of intelligent vehicles in extreme failure conditions.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle control technology, and in particular to a method for quantifying vehicle path safety boundaries in the face of actuator failure. Background Technology

[0002] Advanced autonomous driving systems (L4 and above) place increasingly stringent demands on vehicle safety and fault tolerance. In autonomous driving mode, the system takes the lead in driving, which means that when key actuators such as the vehicle's steering system, braking system, or drive system fail partially or completely, the system must have the ability to autonomously handle the fault and operate safely.

[0003] In existing technological approaches, countermeasures for actuator failure primarily focus on hardware-level redundancy design or simple fault isolation. However, when actuators experience asymmetric failure (such as only one side of the brakes being effective or steering assist being weakened), the vehicle's dynamics exhibit high nonlinearity and coupling. Traditional methods have significant limitations: in situations where actuator failure causes abrupt changes in control gain, traditional fixed safety boundaries cannot reflect the true impact of the failure on the overall vehicle's performance characteristics. This results in safety margin estimates that are either too conservative, limiting vehicle maneuverability, or too aggressive, leading to lateral instability. In complex real-world conditions, accurately quantifying the trajectory deviation caused by both disturbances and failures is a core challenge for achieving safe path tracking. The upper-level path planning layer cannot obtain precise hard positional constraints, making it difficult to make highly safe obstacle avoidance decisions during the transient process following a failure. Summary of the Invention

[0004] The purpose of this invention is to determine the safe error execution boundary of the vehicle path tracking process for high-level autonomous vehicles by combining fault information, environmental information, and vehicle status, and to control the vehicle status within the safe error execution boundary through feedback control, so as to ensure that the overall vehicle operation does not exceed the execution capability boundary after failure, improve the vehicle path tracking accuracy, and prevent unexpected collisions.

[0005] To achieve the above objectives, a first aspect of the present invention proposes a vehicle path safety boundary quantification system for actuator failure, comprising a sensor array, an autonomous driving controller, a fault identification unit, a post-failure error safety boundary calculation module, and an actuator subsystem, wherein:

[0006] The sensor combination is used to collect the vehicle's dynamic state data in real time, and to fuse the perception of the environment data using vision and radar algorithms. The processed vehicle motion state and environmental perception information are sent to the autonomous driving controller in real time, and simultaneously sent to the error safety boundary calculation module after execution failure. The autonomous driving controller is used to generate an ideal autonomous driving trajectory and basic control commands based on environmental information under nominal operating conditions, and at the same time receive the error safety boundary value fed back by the error safety boundary calculation module after execution failure, and finally send the nominal trajectory data and basic control commands to the error safety boundary calculation module after execution failure. The fault identification unit is used to monitor the health status of the vehicle's underlying actuators in real time and quantify the degree of fault. By analyzing the actuator's response residual, current feedback or hydraulic status, it identifies the specific failure location of the steering system, drive system or braking system, and further calculates the remaining efficiency ratio and asymmetric fault characteristic parameters of the failed actuator, thereby generating a vehicle actuator fault status feedback signal containing the fault location and remaining efficiency coefficient, and sending it to the error safety boundary calculation module after the execution failure in real time. The error safety boundary calculation module after the execution failure is used to receive the open-loop trajectory and control commands from the autonomous driving module, the sensing information from the sensor combination, and the vehicle actuator fault status feedback from the fault identification unit, feed them back to the error safety boundary of the autonomous driving controller, and send the control commands of the actuator subsystem. The actuator subsystem includes a four-wheel independent braking brake system, a four-wheel independent drive system, a steer-by-wire system, and sensor combinations of the corresponding subsystems, used to execute the generated actuator subsystem execution commands.

[0007] Optionally, the error safety boundary calculation module after execution fault further includes a fault adaptive input weighting unit, a robust error invariant set optimization unit, an error safety boundary quantization unit, an error feedback control unit, and an execution command allocation unit, wherein: The fault adaptive input weighting unit is used to reconstruct the fault adaptive input weight matrix allocated by the control according to the fault state. The robust error invariant set optimization unit is used to construct a convex optimization problem based on the received execution weight matrix and the vehicle nonlinear dynamics model. By solving linear matrix inequalities, it finds the optimal state-related error gain and the corresponding feedback gain matrix, thereby establishing the system's error performance gain and feedback control rate under the current fault weight. The optimized error performance measurement gain and feedback control rate are then transmitted to the error safety boundary quantization unit and the error feedback control unit, respectively. Furthermore, the unit constructs a lookup table relationship between vehicle state and measurement gain and feedback control gain through offline calculation, reducing real-time requirements. The error safety boundary quantization unit is used to receive the error performance measurement gain output by the robust error invariant set optimization unit, and combined with the preset upper bound of the external disturbance norm, calculate the maximum state deviation that the actual trajectory may produce relative to the nominal trajectory under the combined effect of the current fault level and external disturbance, thereby quantifying the exact error safety boundary and feeding it back to the autonomous driving controller. The error feedback control unit is used to receive the feedback gain matrix output by the robust error invariant set optimization unit, generate a robust correction control law that pulls the vehicle back to the nominal trajectory in real time, and send it to the execution command allocation unit.

[0008] The execution command allocation unit is used to receive error control input from the error feedback control unit and nominal control command from the automatic driving controller, generate the final front wheel steering angle command, four-wheel braking torque command and four-wheel drive torque command, and send them to the actuator subsystem.

[0009] To achieve the above objectives, a second aspect of the present invention proposes a method for quantifying vehicle path safety boundaries in the event of actuator failure, comprising: The system acquires dynamic state information such as the vehicle's current position, speed, and heading angle, as well as environmental disturbance information, through sensors; combines the actuator state information output by the fault identification unit to determine the remaining execution capability of each actuator; models the vehicle modeling error and random disturbance and estimates the upper limit of the disturbance error; constructs positive and negative fault information matrices for the actuators, and determines the permissible execution range of each actuator under fault conditions based on the fault information matrices. A basic weight is constructed based on the proportion of remaining execution capacity of each actuator, an asymmetric risk weight is constructed based on the positive and negative remaining execution capacity of the actuator, and a fault adaptive input weight matrix is ​​constructed based on the basic weight and the asymmetric risk weight. A vehicle path tracking dynamic model considering four-wheel differential control is established and represented as a nonlinear affine system. An error control system is constructed based on the state error between the actual vehicle state and the open-loop trajectory of autonomous driving and the control input error. Linear matrix inequality constraints are constructed and convex optimization problems are solved to obtain the robust gain, performance index matrix and control input feedback matrix of the error invariant set. The state error safety boundary is calculated based on the robust gain, and the error safety boundary of the vehicle path tracking system is determined according to the state error safety boundary. The control input is solved using the pseudospectral method, and the control input is constructed as the sum of the nominal control input and the modified control input; The vehicle actuators are controlled according to the control input, wherein positive values ​​of the four-wheel drive / braking torque are executed by the corresponding wheel drive system, negative values ​​are executed by the corresponding wheel braking system and drive system in coordination, and the front wheel steering angle command is executed by the steering system.

[0010] Optionally, the vehicle's current position, speed, heading angle, and other dynamic state information, as well as environmental disturbance information, are acquired through sensors; combined with the actuator state information output by the fault identification unit, the remaining execution capability of each actuator is determined; vehicle modeling errors and random disturbances are modeled and the upper limit of disturbance error is estimated; positive and negative fault information matrices of the actuators are constructed, and the permissible execution range of each actuator under fault conditions is determined based on the fault information matrices, including: Based on the vehicle's environmental information and dynamic state, the vehicle modeling error and random disturbances are modeled, and the upper limit of the disturbance error is estimated, expressed as:

[0011] in, Modeling errors and random disturbance matrices for vehicles. This is the upper limit of the disturbance error. Let be the infinite norm of the variable.

[0012] Based on the fault information, the following model is performed:

[0013]

[0014] in, , This is a matrix of positive and negative fault information for vehicle actuators. These are the positive residual actuator capacity coefficients for the steering system and the four-wheel drive / braking system, respectively, representing the ratio of residual actuator capacity to actuator capacity under fault-free conditions; The negative residual actuator capacity coefficient of the execution system; Let be a function that generates a matrix based on its diagonal elements. Then the permissible execution range of the vehicle actuator is:

[0015] in, For the execution command of the i-th executor, These represent the front wheel steering angle command and the left front, right front, left rear, and right rear wheel drive braking commands, respectively. The maximum and minimum values ​​for the i-th execution subsystem under healthy conditions.

[0016] Optionally, a basic weight is constructed based on the proportion of remaining execution capacity of each actuator, and an asymmetric risk weight is constructed based on the positive and negative remaining execution capacity of the actuators. A fault adaptive input weight matrix is ​​then constructed based on the basic weight and the asymmetric risk weight, including: The basic weights are designed based on the proportion of the actuator's remaining execution capacity, and are expressed as follows:

[0017]

[0018] in, Let be the coefficient of the remaining execution capability range of the i-th actuator. The weight coefficient for the remaining execution capacity of the i-th executor; , These are adjustable parameters; Considering the risks arising from the uneven distribution of positive and negative faults, an asymmetric risk weight is designed, expressed as:

[0019] in, Let be the asymmetric risk weight coefficient for the i-th actuator. , These are adjustable parameters.

[0020] The fault adaptive input weight matrix after execution failure is calculated using the following formula, and is expressed as follows:

[0021] in, The fault adaptive input weight matrix is ​​used to handle failures.

[0022] Optionally, a vehicle path tracking dynamics model considering four-wheel differential control is established and represented as a nonlinear affine system; an error control system is constructed based on the state error and control input error between the actual vehicle state and the open-loop trajectory of autonomous driving; linear matrix inequality constraints are constructed and a convex optimization problem is solved to obtain the robust gain, performance index matrix, and control input feedback matrix of the error invariant set, including: Modeling is performed for vehicle path tracking dynamics that consider differential motion:

[0023]

[0024] In the formula, This represents the vehicle's longitudinal position in the geodetic coordinate system. This represents the horizontal position in the geodetic coordinate system. For the longitudinal speed of the vehicle, The longitudinal acceleration of the vehicle; The yaw angle of the vehicle. Let yaw rate be the vehicle's angular velocity. The yaw acceleration of the vehicle; The lateral speed of the vehicle. This refers to the vehicle's lateral acceleration. Let Z be the vehicle's moment of inertia along the z-axis. For the overall vehicle weight; This refers to the front wheel steering angle. ( (representing the front and rear axles respectively) represents the lateral forces on the front and rear axles. For the longitudinal forces on the front and rear axles, For four-wheel drive / braking force command ( (These represent the front left, front right, rear left, and rear right wheels, respectively). For four-wheel drive braking torque, and , The radius of the wheel; and These are the distances from the front axle to the center of gravity and the rear axle to the center of gravity, respectively. The wheelbase of the vehicle; This refers to the actual yaw moment generated by the four-wheel differential. For modeling errors or unknown disturbances; system state Control input System modeling error or random disturbance ; The above system can then be expressed as a nonlinear affine system:

[0025] in Let be the system dynamic state function. Given the system performance index function, the corresponding error control system is:

[0026] in, , , , and Designed according to performance requirements, here ; The error between the system state and the open-loop trajectory state of the autonomous driving system; The error between the system control input and the open-loop trajectory state input of the autonomous driving system; This represents the error in system performance indicators; Then construct the following linear matrix inequalities:

[0027]

[0028] in, ; ; Optimize variables for performance metrics. To control the input auxiliary variables, To control the input feedback matrix; , To optimize auxiliary variables, Robust gain for the error-invariant set; and All are identity matrices; Define the above linear matrix inequalities as LMI(1) and LMI(2), then solve the following convex optimization problem:

[0029] in, The permissible set of vehicle states; By solving, the robust gain of the error invariant set is obtained. Performance index matrix With control input feedback matrix .

[0030] Optionally, calculating the state error safety boundary based on the robust gain, and determining the error safety boundary of the vehicle path tracking system according to the state error safety boundary, includes: Define the safety boundary of state error Its robust safety boundary is quantized as follows:

[0031] The error safety boundary of the vehicle path tracking system is defined as follows:

[0032] in, This is the error safety boundary for the vehicle path tracking system.

[0033] Optionally, the control input is solved using a pseudospectral method, and the control input is constructed as the sum of the nominal control input and the modified control input, including: Solving using the pseudospectral method Constructed as nominal control input With correction control input sum:

[0034] in, These are path parameters, representing the path from the actual state to the target trajectory point; The trajectory that consumes the least energy to travel from the actual state to the target trajectory point.

[0035] To achieve the above objectives, a third aspect of this application provides an electronic device, including a processor and a memory; wherein the processor runs a program corresponding to the executable program code stored in the memory to implement the method described in the first aspect.

[0036] To achieve the above objectives, a fourth aspect of this application provides a non-transitory computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method described in the first aspect.

[0037] The embodiments of this invention have the following beneficial effects: By constructing a vehicle path safety boundary quantification method and system oriented towards actuator failure, this invention effectively solves the problems of unknown safety boundaries and path tracking instability faced by high-level autonomous vehicles when critical actuators experience asymmetric failures. Through the above method, this invention first utilizes a fault-adaptive input weight matrix to achieve active suppression and fault tolerance of asymmetric failure risks at the control allocation level; secondly, by constructing a robust error invariant set optimization framework based on linear matrix inequalities, it successfully transforms the complex nonlinear dynamic failure effects into calculable robust performance gains, thereby quantifying a precise physical space error safety boundary and providing a hard constraint for upper-level autonomous driving path planning; finally, the solved real-time feedback control law ensures that the vehicle's actual operating trajectory always converges within the quantified safety boundary under the dual influence of external disturbances and actuator performance degradation. This method breaks through the limitations of traditional fixed safety boundaries and significantly improves the dynamic obstacle avoidance capability and operational safety of intelligent vehicles under extreme failure conditions. Attached Figure Description

[0038] The above-described and additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, in which: Figure 1 A framework diagram of a vehicle path safety boundary quantization system for actuator failure is provided in an embodiment of the present invention. Figure 2 A flowchart for quantizing vehicle path safety boundaries in response to actuator failure, provided as an embodiment of the present invention. Detailed Implementation

[0039] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0040] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0041] The following describes, with reference to the accompanying drawings, a vehicle path safety boundary quantification system and method for actuator failure based on an embodiment of the present invention.

[0042] Example 1 This embodiment provides a vehicle path safety boundary quantification system for actuator failure. For example... Figure 1 As shown, the system includes a sensor array, an autonomous driving controller, a fault identification unit, a fault-based error safety boundary calculation module, and an actuator subsystem. The specific connections and processing logic of each module are as follows: In this embodiment of the invention, the sensor suite supporting autonomous driving is used to perceive the vehicle's motion state and surrounding environment information in all directions. Specifically, it includes high-precision hardware devices such as lidar, cameras, an integrated inertial navigation system (IMU), and wheel speed sensors. This module is responsible for real-time acquisition of dynamic state data such as the vehicle's yaw rate, lateral acceleration, heading angle, and wheel speed. It also utilizes visual and radar algorithms to fuse perception of environmental data such as lane curvature, road boundary geometry, and obstacle positions. Finally, the processed vehicle motion state and environmental perception information are sent to the autonomous driving controller in real time. Simultaneously, this information is also sent to the post-fault error safety boundary calculation module, providing data support for subsequent dynamic modeling and robust boundary calculation.

[0043] In this embodiment of the invention, the autonomous driving controller is used to generate an ideal autonomous driving trajectory and basic control commands based on environmental information under nominal operating conditions. Simultaneously, it receives error safety boundary values ​​fed back by the error safety boundary calculation module after an execution failure, and finally sends the nominal trajectory data and basic control commands to the error safety boundary calculation module after an execution failure.

[0044] In this embodiment of the invention, the fault identification unit monitors the health status of the vehicle's underlying actuators in real time and quantifies the degree of fault. This module accurately identifies the specific failure location of the steering system, drive system, or braking system by analyzing the actuator's response residuals, current feedback, or hydraulic status. It further calculates the remaining efficiency ratio and asymmetric fault characteristic parameters of the failed actuator, thereby generating a vehicle actuator fault status feedback signal containing the fault location and remaining efficiency coefficient, and sending it in real time to the error safety boundary calculation module after the execution failure.

[0045] In this embodiment of the invention, the error safety boundary calculation module after the execution failure is used to receive the open-loop trajectory and control commands from the autonomous driving module, the sensing information from the sensor combination, and the vehicle actuator fault status feedback from the fault identification unit, feed them back to the error safety boundary of the autonomous driving controller, and send the control commands of the actuator subsystem. In this embodiment of the invention, the actuator subsystem includes a four-wheel independent braking brake system, a four-wheel independent drive system, a steer-by-wire system, and sensor combinations of the corresponding subsystems, used to execute the generated actuator subsystem execution commands.

[0046] Furthermore, in one embodiment of the present invention, the post-fault error safety boundary calculation module further includes a fault adaptive input weighting unit, a robust error invariant set optimization unit, an error safety boundary quantization unit, an error feedback control unit, and an execution command allocation unit, wherein: The fault adaptive input weighting unit is used to reconstruct the fault adaptive input weight matrix allocated by the control based on the fault state. This unit receives feedback signals from the fault identification unit and calculates the weights using an internal risk penalty algorithm. Specifically, for actuators with low residual efficiency, the penalty weight in the optimization objective function is increased. At the same time, additional risk weights are added for asymmetric fault characteristics that may exacerbate vehicle lateral instability (such as unilateral drive failure). This generates a dynamic execution weight matrix, which is then passed to the robust error invariant set optimization unit to guide subsequent control law calculations to automatically avoid dependence on failed actuators.

[0047] The robust error invariant set optimization unit constructs a convex optimization problem based on the received execution weight matrix and the vehicle nonlinear dynamics model. By solving linear matrix inequalities, it finds the optimal state-related error gain and the corresponding feedback gain matrix, thereby establishing the system's error performance gain and feedback control rate under the current fault weight. The optimized error performance metric gain and feedback control rate are then transmitted to the error safety boundary quantization unit and the error feedback control unit, respectively.

[0048] The error safety boundary quantization unit receives the error performance metric gain output by the robust error invariant set optimization unit, and combines it with the preset upper bound of the external disturbance norm. Based on the calculation of the maximum state deviation that the actual trajectory may produce relative to the nominal trajectory under the combined effect of the current fault level and external disturbance, the precise error safety boundary is quantified and fed back to the autonomous driving controller.

[0049] The error feedback control unit receives the feedback gain matrix output by the robust error invariant set optimization unit and generates a robust correction control law in real time to pull the vehicle back to its nominal trajectory. To solve the computationally complex real-time problem, the minimum energy path in the state space is quickly solved and integral calculation is performed along the path to obtain an error control input that can offset disturbances and model uncertainties, which is then sent to the execution command allocation unit.

[0050] The execution command allocation unit receives error control input from the error feedback control unit and nominal control instructions from the automatic driving controller, generates the final front wheel steering angle command, four-wheel braking torque command and four-wheel drive torque command, and sends them to the actuator subsystem.

[0051] Example 2 This invention relates to a method for quantifying vehicle path safety boundaries in the face of actuator failure, such as... Figure 2 As shown. This method is implemented using the vehicle path safety boundary quantization system shown in Example 1. (Refer to...) Figure 2 The method includes the following steps: S1. The system acquires dynamic state information such as the vehicle's current position, speed, and heading angle, as well as environmental disturbance information, through sensors. Combined with the actuator state information output by the fault identification unit, the remaining execution capability of each actuator is determined. The system models vehicle modeling errors and random disturbances and estimates the upper limit of disturbance error. The system constructs positive and negative fault information matrices for the actuators and determines the permissible execution range of each actuator under fault conditions based on the fault information matrices.

[0052] In this embodiment of the invention, the current state of the vehicle is obtained through a combination of sensors. (Including position, speed, heading angle, etc.) and environmental interference information. At the same time, combined with the actuator status fed back by the fault identification unit, the remaining actuator execution capacity under the condition of vehicle execution failure is calculated.

[0053] Specifically, considering the vehicle's environmental information and dynamic state, the vehicle modeling error and random disturbances are modeled, and the upper limit of the disturbance error is estimated, expressed as:

[0054] in, Modeling errors and random disturbance matrices for vehicles. This is the upper limit of the disturbance error. Let be the infinite norm of the variable.

[0055] Based on the fault information, the following model is performed:

[0056]

[0057] in, , This is a matrix of positive and negative fault information for vehicle actuators. These are the positive residual actuator capacity coefficients for the steering system and the four-wheel drive / braking system, respectively, representing the ratio of residual actuator capacity to actuator capacity under fault-free conditions; The negative residual actuator capacity coefficient of the execution system; Let be a function that generates a matrix based on its diagonal elements. Then the permissible execution range of the vehicle actuator is:

[0058] in, For the execution command of the i-th executor, These represent the front wheel steering angle command and the left front, right front, left rear, and right rear wheel drive braking commands, respectively. The maximum and minimum values ​​for the i-th execution subsystem under healthy conditions.

[0059] S2, construct basic weights based on the proportion of remaining execution capacity of each actuator, construct asymmetric risk weights based on the positive and negative remaining execution capacity of the actuators, and construct a fault adaptive input weight matrix based on the basic weights and asymmetric risk weights.

[0060] It should be noted that, in order to suppress the use of failed actuators in control allocation and reduce the risk of instability caused by asymmetric faults, this embodiment of the invention designs a state-related weight matrix that depends on the fault state. .

[0061] Specifically, firstly, the basic weights are designed based on the proportion of the actuator's remaining execution capacity, expressed as:

[0062]

[0063] in, Let be the coefficient of the remaining execution capability range of the i-th actuator. The weight coefficient for the remaining execution capacity of the i-th executor; , These are adjustable parameters.

[0064] Furthermore, considering the risks arising from the uneven distribution of positive and negative faults, an asymmetric risk weight is designed, expressed as:

[0065] in, Let be the asymmetric risk weight coefficient for the i-th actuator. , These are adjustable parameters.

[0066] Finally, the embodiment of the present invention calculates the fault adaptive input weight matrix after execution failure according to the following formula, expressed as:

[0067] in, The fault adaptive input weight matrix is ​​used to handle failures.

[0068] S3. Establish a vehicle path tracking dynamic model considering four-wheel differential control and represent it as a nonlinear affine system; construct an error control system based on the state error and control input error between the actual vehicle state and the open-loop trajectory of autonomous driving; construct linear matrix inequality constraints and solve the convex optimization problem to obtain the robust gain, performance index matrix and control input feedback matrix of the error invariant set.

[0069] In this embodiment of the invention, modeling is performed for vehicle path tracking dynamics that take into account differential motion:

[0070]

[0071] In the formula, This represents the vehicle's longitudinal position in the geodetic coordinate system. This represents the horizontal position in the geodetic coordinate system. For the longitudinal speed of the vehicle, The longitudinal acceleration of the vehicle; The yaw angle of the vehicle. Let yaw rate be the vehicle's angular velocity. The yaw acceleration of the vehicle; The lateral speed of the vehicle. This refers to the vehicle's lateral acceleration. Let Z be the vehicle's moment of inertia along the z-axis. For the overall vehicle weight; This refers to the front wheel steering angle. ( (representing the front and rear axles respectively) represents the lateral forces on the front and rear axles. For the longitudinal forces on the front and rear axles, For four-wheel drive / braking force command ( (These represent the front left, front right, rear left, and rear right wheels, respectively). For four-wheel drive braking torque, and , The radius of the wheel; and These are the distances from the front axle to the center of gravity and the rear axle to the center of gravity, respectively. The wheelbase of the vehicle; This refers to the actual yaw moment generated by the four-wheel differential. For modeling errors or unknown disturbances; system state Control input System modeling error or random disturbance ; The above system can then be expressed as a nonlinear affine system:

[0072] in Let be the system dynamic state function. Given the system performance index function, the corresponding error control system is:

[0073] in, , , , and Designed according to performance requirements, here ; The error between the system state and the open-loop trajectory state of the autonomous driving system; The error between the system control input and the open-loop trajectory state input of the autonomous driving system; This represents the error in system performance indicators; Then construct the following linear matrix inequalities:

[0074]

[0075] in, ; ; Optimize variables for performance metrics. To control the input auxiliary variables, To control the input feedback matrix; , To optimize auxiliary variables, Robust gain for the error-invariant set; and All are identity matrices; Define the above linear matrix inequalities as LMI(1) and LMI(2), then solve the following convex optimization problem:

[0076] in, The permissible set of vehicle states; By solving, the robust gain of the error invariant set is obtained. Performance index matrix With control input feedback matrix .

[0077] S4. Calculate the state error safety boundary based on the robust gain, and determine the error safety boundary of the vehicle path tracking system according to the state error safety boundary.

[0078] In this embodiment of the invention, the robust gain is applied to the minimum error invariant set obtained in step S3. Calculate the error safety boundary in the physical space. First, define the state error safety boundary. Its robust safety boundary is quantized as follows:

[0079] The error safety boundary of the vehicle path tracking system is defined as follows:

[0080] in, This is the error safety boundary for the vehicle path tracking system.

[0081] S5. The pseudospectral method is used to solve for the control input, and the control input is constructed as the sum of the nominal control input and the modified control input.

[0082] In this embodiment of the invention, the control input is calculated based on path error and feedback control law. To generate control commands that pull the vehicle back to its nominal trajectory in real time, this embodiment employs a pseudospectral method for solving the problem. Constructed as nominal control input With correction control input The sum is expressed as:

[0083] in, These are path parameters, representing the path from the actual state to the target trajectory point; The trajectory that consumes the least energy to travel from the actual state to the target trajectory point.

[0084] S6, control the vehicle actuators according to the control input, wherein positive values ​​of the four-wheel drive / braking torque are executed by the corresponding wheel drive system, negative values ​​are executed by the corresponding wheel braking system and drive system in coordination, and the front wheel steering angle command is executed by the steering system.

[0085] Ultimately, the actuator executes the corresponding control command, adjusting the four-wheel drive / braking torque. Positive values ​​are executed by the corresponding wheel drive system, while negative values ​​are coordinated and controlled by the corresponding wheel braking and drive systems. (Front wheel steering angle command) This is performed by the steering system.

[0086] Example 3 To implement the methods of the above embodiments, the present invention also provides an electronic device, which includes a memory and a processor; wherein the processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the various steps of the methods described above.

[0087] Example 4 To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described in the foregoing embodiments.

[0088] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0089] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0090] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A vehicle path safety boundary quantification system for actuator failure, characterized in that, It includes a sensor array, an autonomous driving controller, a fault identification unit, a post-fault error safety boundary calculation module, and an actuator subsystem, wherein: The sensor combination is used to collect the vehicle's dynamic state data in real time, and to fuse the perception of the environment data using vision and radar algorithms. The processed vehicle motion state and environmental perception information are sent to the autonomous driving controller in real time, and simultaneously sent to the error safety boundary calculation module after execution failure. The autonomous driving controller is used to generate an ideal autonomous driving trajectory and basic control commands based on environmental information under nominal operating conditions, and at the same time receive the error safety boundary value fed back by the error safety boundary calculation module after execution failure, and finally send the nominal trajectory data and basic control commands to the error safety boundary calculation module after execution failure. The fault identification unit is used to monitor the health status of the vehicle's underlying actuators in real time and quantify the degree of fault. By analyzing the actuator's response residual, current feedback or hydraulic status, it identifies the specific failure location of the steering system, drive system or braking system, and further calculates the remaining efficiency ratio and asymmetric fault characteristic parameters of the failed actuator, thereby generating a vehicle actuator fault status feedback signal containing the fault location and remaining efficiency coefficient, and sending it to the error safety boundary calculation module after the execution failure in real time. The error safety boundary calculation module after the execution failure is used to receive the open-loop trajectory and control commands from the autonomous driving module, the sensing information from the sensor combination, and the vehicle actuator fault status feedback from the fault identification unit, feed them back to the error safety boundary of the autonomous driving controller, and send the control commands of the actuator subsystem. The actuator subsystem includes a four-wheel independent braking brake system, a four-wheel independent drive system, a steer-by-wire system, and sensor combinations of the corresponding subsystems, used to execute the generated actuator subsystem execution commands.

2. The system according to claim 1, characterized in that, The error safety boundary calculation module after execution fault also includes a fault adaptive input weighting unit, a robust error invariant set optimization unit, an error safety boundary quantization unit, an error feedback control unit, and an execution command allocation unit, wherein: The fault adaptive input weighting unit is used to reconstruct the fault adaptive input weight matrix allocated by the control according to the fault state. The robust error invariant set optimization unit is used to construct a convex optimization problem based on the received execution weight matrix and the vehicle nonlinear dynamics model. By solving linear matrix inequalities, it finds the optimal state-related error gain and the corresponding feedback gain matrix, thereby establishing the system's error performance gain and feedback control rate under the current fault weight. The optimized error performance measurement gain and feedback control rate are then transmitted to the error safety boundary quantization unit and the error feedback control unit, respectively. Furthermore, the unit constructs a lookup table relationship between vehicle state and measurement gain and feedback control gain through offline calculation, reducing real-time requirements. The error safety boundary quantization unit is used to receive the error performance measurement gain output by the robust error invariant set optimization unit, and combined with the preset upper bound of the external disturbance norm, calculate the maximum state deviation that the actual trajectory may produce relative to the nominal trajectory under the combined effect of the current fault level and external disturbance, thereby quantifying the exact error safety boundary and feeding it back to the autonomous driving controller. The error feedback control unit is used to receive the feedback gain matrix output by the robust error invariant set optimization unit, generate a robust correction control law that pulls the vehicle back to the nominal trajectory in real time, and send it to the execution command allocation unit. The execution command allocation unit is used to receive error control input from the error feedback control unit and nominal control command from the automatic driving controller, generate the final front wheel steering angle command, four-wheel braking torque command and four-wheel drive torque command, and send them to the actuator subsystem.

3. A method for quantifying vehicle path safety boundaries in the face of actuator failure, characterized in that, Implemented by the vehicle path safety boundary quantification system according to any one of claims 1-2, comprising: The system acquires dynamic state information such as the vehicle's current position, speed, and heading angle, as well as environmental disturbance information, through sensors; combines the actuator state information output by the fault identification unit to determine the remaining execution capability of each actuator; models the vehicle modeling error and random disturbance and estimates the upper limit of the disturbance error; constructs positive and negative fault information matrices for the actuators, and determines the permissible execution range of each actuator under fault conditions based on the fault information matrices. A basic weight is constructed based on the proportion of remaining execution capacity of each actuator, an asymmetric risk weight is constructed based on the positive and negative remaining execution capacity of the actuator, and a fault adaptive input weight matrix is ​​constructed based on the basic weight and the asymmetric risk weight. A vehicle path tracking dynamic model considering four-wheel differential control is established and represented as a nonlinear affine system. An error control system is constructed based on the state error between the actual vehicle state and the open-loop trajectory of autonomous driving and the control input error. Linear matrix inequality constraints are constructed and convex optimization problems are solved to obtain the robust gain, performance index matrix and control input feedback matrix of the error invariant set. The state error safety boundary is calculated based on the robust gain, and the error safety boundary of the vehicle path tracking system is determined according to the state error safety boundary. The control input is solved using the pseudospectral method, and the control input is constructed as the sum of the nominal control input and the modified control input; The vehicle actuators are controlled according to the control input, wherein positive values ​​of the four-wheel drive / braking torque are executed by the corresponding wheel drive system, negative values ​​are executed by the corresponding wheel braking system and drive system in coordination, and the front wheel steering angle command is executed by the steering system.

4. The method according to claim 3, characterized in that, The vehicle's current position, speed, heading angle, and other dynamic state information, as well as environmental disturbance information, are obtained through sensors. Based on the actuator status information output by the fault identification unit, the remaining execution capacity of each actuator is determined; vehicle modeling errors and random disturbances are modeled and the upper limit of disturbance error is estimated; Construct positive and negative fault information matrices for the actuators, and determine the permissible execution range of each actuator under fault conditions based on the fault information matrices, including: Based on the vehicle's environmental information and dynamic state, the vehicle modeling error and random disturbances are modeled, and the upper limit of the disturbance error is estimated, expressed as: in, Modeling errors and random disturbance matrices for vehicles. This is the upper limit of the disturbance error. Let be the infinite norm of the variable. Based on the fault information, the following model is performed: in, , This is a matrix of positive and negative fault information for vehicle actuators. These are the positive residual actuator capacity coefficients for the steering system and the four-wheel drive / braking system, respectively, representing the ratio of residual actuator capacity to actuator capacity under fault-free conditions; The negative residual actuator capacity coefficient of the execution system; Let be a function that generates a matrix based on its diagonal elements. Then the permissible execution range of the vehicle actuator is: in, For the execution command of the i-th executor, These represent the front wheel steering angle command and the left front, right front, left rear, and right rear wheel drive braking commands, respectively. The maximum and minimum values ​​for the i-th execution subsystem under healthy conditions.

5. The method according to claim 4, characterized in that, A basic weight is constructed based on the proportion of remaining execution capacity of each actuator. An asymmetric risk weight is constructed based on the positive and negative remaining execution capacity of the actuators. A fault adaptive input weight matrix is ​​then constructed based on the basic weight and the asymmetric risk weight, including: The basic weights are designed based on the proportion of the actuator's remaining execution capacity, and are expressed as follows: in, Let be the coefficient of the remaining execution capability range of the i-th actuator. The weight coefficient for the remaining execution capacity of the i-th executor; , These are adjustable parameters; Considering the risks arising from the uneven distribution of positive and negative faults, an asymmetric risk weight is designed, expressed as: in, Let be the asymmetric risk weight coefficient for the i-th actuator. , These are adjustable parameters. The fault adaptive input weight matrix after execution failure is calculated using the following formula, and is expressed as follows: in, The fault adaptive input weight matrix is ​​used to handle failures.

6. The method according to claim 5, characterized in that, A vehicle path tracking dynamics model considering four-wheel differential control is established and represented as a nonlinear affine system; an error control system is constructed based on the state error and control input error between the actual vehicle state and the open-loop trajectory of autonomous driving. By constructing linear matrix inequality constraints and solving the convex optimization problem, we obtain the robust gain, performance index matrix, and control input feedback matrix of the error invariant set, including: Modeling is performed for vehicle path tracking dynamics that consider differential motion: In the formula, This represents the vehicle's longitudinal position in the geodetic coordinate system. This represents the horizontal position in the geodetic coordinate system. For the longitudinal speed of the vehicle, The longitudinal acceleration of the vehicle; The yaw angle of the vehicle. Let yaw rate be the vehicle's angular velocity. The yaw acceleration of the vehicle; The lateral speed of the vehicle. This refers to the vehicle's lateral acceleration. Let Z be the vehicle's moment of inertia along the z-axis. For the overall vehicle weight; This refers to the front wheel steering angle. ( (representing the front and rear axles respectively) represents the lateral forces on the front and rear axles. For the longitudinal forces on the front and rear axles, For four-wheel drive / braking force command ( (These represent the front left, front right, rear left, and rear right wheels, respectively). For four-wheel drive braking torque, and , The radius of the wheel; and These are the distances from the front axle to the center of gravity and the rear axle to the center of gravity, respectively. The wheelbase of the vehicle; This refers to the actual yaw moment generated by the four-wheel differential. For modeling errors or unknown disturbances; system state Control input System modeling error or random disturbance ; The above system can then be expressed as a nonlinear affine system: in Let be the system dynamic state function. Given the system performance index function, the corresponding error control system is: in, , , , and Designed according to performance requirements, here ; The error between the system state and the open-loop trajectory state of the autonomous driving system; The error between the system control input and the open-loop trajectory state input of the autonomous driving system; This represents the error in system performance indicators; Then construct the following linear matrix inequalities: in, ; ; Optimize variables for performance metrics. To control the input auxiliary variables, To control the input feedback matrix; , To optimize auxiliary variables, Robust gain for the error-invariant set; and All are identity matrices; Define the above linear matrix inequalities as LMI(1) and LMI(2), then solve the following convex optimization problem: in, The permissible set of vehicle states; By solving, the robust gain of the error invariant set is obtained. Performance index matrix With control input feedback matrix .

7. The method according to claim 6, characterized in that, The calculation of the state error safety boundary based on the robust gain, and the determination of the error safety boundary of the vehicle path tracking system based on the state error safety boundary, include: Define the safety boundary of state error Its robust safety boundary is quantized as follows: The error safety boundary of the vehicle path tracking system is defined as follows: in, This is the error safety boundary for the vehicle path tracking system.

8. The method according to claim 7, characterized in that, The control input is solved using the pseudospectral method, and the control input is constructed as the sum of the nominal control input and the modified control input, including: Solving using the pseudospectral method Constructed as nominal control input With correction control input sum: in, These are path parameters, representing the path from the actual state to the target trajectory point; The trajectory that consumes the least energy to travel from the actual state to the target trajectory point.

9. An electronic device, characterized in that, Including processor and memory; The processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the method as described in any one of claims 3-8.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 3-8.

Images