Control parameter determination method and device, equipment, medium and program product

Abstract

The invention provides a control parameter determination method and device, equipment, a medium and a program product. The method comprises the following steps: determining a first association relationship based on a current prediction step length of the automatic driving system; wherein the first association relationship comprises an association relationship between a state error and a control parameter; and determining a current control parameter based on the current state error and the first association relationship. According to the invention, the determination efficiency of the current control parameter of the vehicle can be improved.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and in particular to a method, apparatus, equipment, medium, and program product for determining control parameters. Background Technology

[0002] Trajectory tracking control is one of the fundamental technologies for achieving autonomous driving functions in autonomous vehicles (AVs). It controls the AV to continuously track a reference trajectory by controlling the AV's longitudinal speed and steering angle. In the field of path tracking control, Model Predictive Control (MPC) determines the optimal longitudinal and lateral inputs through online calculations, enabling the AV to follow the reference trajectory with the best performance in each sampling time.

[0003] However, the online calculation process of MPC requires continuous solution of differential equations during the iterative operation of the prediction model and rolling optimization. This process consumes a lot of data processing resources of the AV, thereby prolonging the determination time of control parameters, including longitudinal and lateral inputs, reducing the real-time performance of the AV's autonomous driving control, and consequently reducing the safety of the AV during autonomous driving. Summary of the Invention

[0004] This application provides a method, apparatus, device, medium, and program product for determining control parameters, which can improve the efficiency of determining vehicle control parameters and thus improve the safety of the vehicle's autonomous driving process.

[0005] The technical solution of this application embodiment is implemented as follows:

[0006] This application provides a method for determining control parameters, the method comprising:

[0007] A first correlation is determined based on the current prediction step size of the autonomous driving system; wherein, the first correlation includes the correlation between state error and control parameters;

[0008] Based on the current state error and the first correlation, the current control parameters are determined.

[0009] In some embodiments, before determining the first association based on the current prediction step size of the autonomous driving system, the method further includes:

[0010] Determine the current tracking status;

[0011] The current tracking state is processed by a fuzzy control model to obtain the current prediction step size; wherein the fuzzy control model is used to characterize the relationship between at least two parameters associated with the autonomous driving process.

[0012] In some embodiments, the fuzzy control model includes a second correlation relationship; the second correlation relationship includes a correlation relationship between the prediction step size and the tracking state; the step of processing the current tracking state through the fuzzy control model to obtain the current prediction step size includes:

[0013] The current prediction step size is determined based on the degree of matching between the current tracking state and the tracking state in the second association relationship.

[0014] In some embodiments, the fuzzy control model includes a third correlation relationship; the third correlation relationship includes a correlation relationship between tracking error and tracking state; determining the current tracking state includes:

[0015] Determine the current tracking error; wherein, the current tracking error includes the longitudinal tracking error and the lateral tracking error corresponding to the current time of the autonomous driving system;

[0016] The current tracking state is determined based on the current tracking error and the third correlation.

[0017] In some embodiments, before determining the current control parameters based on the current state error and the first correlation, the method further includes:

[0018] Based on the current prediction step size and the set of relationships, the first association relationship is determined; wherein, the set of relationships includes a set of association relationships between the prediction step size set and the set of autonomous driving parameters; the set of autonomous driving parameters includes a set of state errors and control parameters.

[0019] In some embodiments, the method further includes:

[0020] Determine a trajectory tracking optimization function; wherein the trajectory tracking optimization function is at least associated with a kinematic prediction model; and the trajectory tracking optimization function is at least associated with the increments of state error and control parameters;

[0021] The explicit functional model between the state error and the control parameters is determined based on the trajectory tracking optimization function.

[0022] The set of relations is determined using the display function model.

[0023] In some embodiments, determining the explicit functional model between the state error and the control parameters based on the trajectory tracking optimization function includes:

[0024] The trajectory tracking optimization function is transformed to obtain an intermediate optimization function that includes at least the control parameters and the state error;

[0025] Determine a control sequence; wherein the control sequence includes multiple distinct control parameters;

[0026] Discretize the state error to obtain the discrete state error result;

[0027] Based on the control sequence and the state error discretization result, the intermediate optimization function is processed to obtain the explicit function model.

[0028] This application embodiment also provides a control parameter determining device, the control parameter determining device comprising:

[0029] The determination module is used to determine a first correlation based on the current prediction step size of the autonomous driving system; wherein, the first correlation includes the correlation between state error and control parameters;

[0030] The determining module is further configured to determine the current control parameters based on the current state error and the first correlation relationship.

[0031] In some embodiments, the determining module is further configured to determine the current tracking state; process the current tracking state using a fuzzy control model to obtain the current prediction step size; wherein the fuzzy control model is used to characterize the correlation between at least two parameters associated with the autonomous driving process.

[0032] In some embodiments, the fuzzy control model includes a second correlation relationship; the second correlation relationship includes the correlation relationship between the prediction step size and the tracking state;

[0033] The determining module is further configured to determine the current prediction step size based on the degree of matching between the current tracking state and the tracking state in the second association relationship.

[0034] In some embodiments, the fuzzy control model includes a third correlation relationship; the third correlation relationship includes the correlation relationship between tracking error and tracking state;

[0035] The determining module is further configured to determine the current tracking error; and based on the current tracking error and the third correlation, determine the current tracking state; wherein the current tracking error includes the longitudinal tracking error and the lateral tracking error corresponding to the current time of the autonomous driving system.

[0036] In some embodiments, the determining module is further configured to determine the first association relationship based on the current prediction step size and the relationship set; wherein the relationship set includes a set of association relationships between the prediction step size set and the autonomous driving parameter set; and the autonomous driving parameter set includes a set of state errors and control parameters.

[0037] In some embodiments, the determining module is further configured to determine a trajectory tracking optimization function; wherein the trajectory tracking optimization function is associated with at least a kinematic prediction model; and the trajectory tracking optimization function is associated with at least the increments of state error and control parameters;

[0038] The determining module is further configured to determine an explicit function model between the state error and the control parameters based on the trajectory tracking optimization function; and to determine the set of relationships through the explicit function model.

[0039] In some embodiments, the determining module is further configured to perform a transformation process on the trajectory tracking optimization function to obtain an intermediate optimization function that includes at least the control parameters and the state error; and determine a control sequence; wherein the control sequence includes multiple control parameters that are different from each other;

[0040] The determining module is also used to discretize the state error to obtain a discrete state error result; based on the control sequence and the discrete state error result, the intermediate optimization function is processed to obtain the explicit function model.

[0041] This application embodiment also provides a control parameter determining device, the control parameter determining device comprising:

[0042] Memory is used to store executable instructions for a computer;

[0043] When the processor executes computer-executable instructions stored in the memory, it is able to implement the control parameter determination method as described above.

[0044] This application also provides a computer-readable storage medium storing computer-executable instructions or computer programs, which, when executed by a processor, can implement the control parameter determination method as described above.

[0045] This application also provides a computer program product, including computer-executable instructions or a computer program, which, when executed by a processor, can implement the control parameter determination method as described in any of the preceding claims.

[0046] The embodiments of this application have the following beneficial effects:

[0047] The control parameter determination method provided in this application determines a first correlation based on the current prediction step size of the autonomous driving system. The first correlation includes the correlation between the state error and the control parameters, thereby shortening the time required to determine the first correlation. Furthermore, by determining the first correlation based on the current prediction step size, the matching degree between the first correlation and the autonomous driving system can be improved, thus increasing the accuracy of the first correlation. Simultaneously, the current control parameters are determined based on the current state error and the first correlation, where the first correlation includes the correlation between the state error and the control parameters. Thus, through the above operations, not only can the accuracy of the current control parameters be improved, but the efficiency of determining the current control parameters of the vehicle can also be improved, thereby meeting the actual control parameter determination requirements of the vehicle. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the structure for implementing the MPC algorithm;

[0049] Figure 2 A flowchart illustrating the control parameter determination method provided in this application embodiment;

[0050] Figure 3 A schematic diagram illustrating the principle of a vehicle's kinematic prediction model;

[0051] Figure 4 This is a second flowchart illustrating the control parameter determination method provided in the embodiments of this application;

[0052] Figure 5A A schematic diagram of the vehicle structure simulated for the control parameter determination method provided in the embodiments of this application;

[0053] Figure 5B A schematic diagram of vehicle parameters simulated for the control parameter determination method provided in the embodiments of this application;

[0054] Figure 5C This is a schematic diagram of the data transmission structure of the vehicle simulation model provided in the embodiments of this application;

[0055] Figure 6A A schematic diagram illustrating the tracking effect of the control parameter determination method provided in the embodiments of this application;

[0056] Figure 6B A schematic diagram illustrating the tracking error effects of the three tracking methods provided in the embodiments of this application;

[0057] Figure 6C A comparative illustration of the speed control effect provided in the embodiments of this application;

[0058] Figure 6DA schematic diagram showing the effect comparison between the prediction step size and the state space of the control law provided in the embodiments of this application;

[0059] Figure 6E A comparative diagram illustrating the computational efficiency provided in the embodiments of this application;

[0060] Figure 7 This is a schematic diagram of the control parameter determination device provided in the embodiments of this application;

[0061] Figure 8 This is a schematic diagram of the control parameter determination device provided in the embodiments of this application.

[0062] It should be noted that the terms "first" and "second" mentioned above are only used to distinguish between different options and do not represent the degree of superiority or inferiority of the options or their priority in the implementation process. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0064] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0065] In the following description, the terms "first, second, third" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0066] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0067] Unless otherwise defined, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in the embodiments of this application is for the purpose of describing the embodiments of this application only and is not intended to limit this application.

[0068] In the implementation of this application, the collection and processing of relevant data should strictly comply with the requirements of relevant national laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of laws and regulations and the authorization of the personal information subject.

[0069] Before providing a further detailed description of the embodiments of this application, the nouns and terms involved in the embodiments of this application will be explained, and the nouns and terms involved in the embodiments of this application shall be interpreted as follows.

[0070] 1) AV (Autonomous Vehicle): Integrates perception systems, artificial intelligence, and advanced control strategies. Currently, AV has become an emerging and popular research field, and it has also become the core of the next-generation mobility framework. In future autonomous driving scenarios, AV may reshape mobility by improving vehicle safety, convenience, energy efficiency, and passenger comfort.

[0071] 2) MPC: This algorithm is an optimal control strategy that calculates the optimal longitudinal and lateral inputs online, enabling the vehicle to follow the reference trajectory with the best performance in each sampling time.

[0072] 3) Explicit Model Predictive Control (EMPC): By pre-compiling the MPC algorithm offline and treating the state variables as parameters in the original quadratic programming problem, the feedback law between the state variables and the control input can be obtained by solving the multi-parameter quadratic programming problem (QP), thus making the originally implicit state feedback explicit.

[0073] 4) Fuzzy control: A control method that utilizes the basic ideas and theories of fuzzy mathematics.

[0074] Trajectory tracking control is one of the fundamental technologies of autonomous driving, which involves controlling the longitudinal speed and steering angle of the AV to control the AV to follow a given reference trajectory.

[0075] Related technologies offer various control strategies for trajectory tracking control; these strategies include traditional control algorithms, learning-based control algorithms, sliding mode controllers, and Lyapunov-based controllers. Among all these strategies, the MPC algorithm is an optimal control strategy, capable of calculating the optimal longitudinal and lateral inputs of the vehicle online, enabling the vehicle to follow the reference trajectory with best performance in each sampling time. In practical applications, the MPC algorithm needs to minimize a real-time cost function that includes the future state, control action, expected reference state, and AV control input. It tracks the difference between the actual and reference values ​​and the deviation of the control input during the control process, and attempts to optimize these two deviations to zero under specified constraints.

[0076] In practical applications, MPC is used in linear MPC controllers. Some types of MPC controllers are configured in a cascaded control manner; these controllers are primarily used for vehicle kinematics and dynamics models. Practical research shows that the control strategies of the aforementioned MPC controllers can significantly improve the effectiveness of control strategies at different speed settings.

[0077] Another category of MPC controllers includes those using vehicle dynamics models with varying fidelity. Simulation results demonstrate that these MPC controllers exhibit good tracking performance. Related technologies also provide a nonlinear MPC controller capable of performing trajectory tracking in obstacle avoidance scenarios.

[0078] Figure 1 A schematic diagram of the structure of the MPC algorithm implementation, as shown below. Figure 1 As shown, this structure typically includes: a determination unit 101, a rolling optimization unit 102, a control unit 103, a prediction unit 104, and a feedback correction unit 105; wherein:

[0079] The determining unit 101 is used to receive the system-set trajectory and the observed trajectory Y. o And based on the above two trajectories, determine the reference trajectory Y. r ; Rolling optimization unit 102 is used to optimize the reference trajectory Y output by determination unit 101. r The error output by the feedback correction unit 105 is used to perform online rolling optimization calculations to determine control parameters, including the direction angle and speed, and these control parameters are sent to the control unit 103. The control unit 103 is used to control the vehicle's autonomous driving state based on the control parameters. The prediction unit 104 is used to predict the observed trajectory Y. o And control parameters determine the predicted trajectory Y p The feedback correction unit 105 is used to correct the predicted trajectory Y. p With the observed trajectory Y o The error between them is sent to the rolling optimization unit 102.

[0080] However, the iterative and rolling optimization process of the MPC algorithm requires continuous online solution of differential equations, which leads to a significant consumption of online computing resources, thereby increasing the time required to determine control parameters. Therefore, the MPC algorithm faces considerable challenges when dealing with control systems requiring high speed. For example, in situations where online computing resources for AV systems are scarce, the excessively long continuous online computation time of MPC will directly impact vehicle driving safety.

[0081] To overcome the aforementioned shortcomings of the MPC model, the EMPC model is proposed. This model employs offline pre-computation of the MPC algorithm and treats state variables as parameters in a QP problem. By solving a multi-parameter QP problem, it obtains the feedback law between the state variables and the control input, thus making the originally implicit state feedback explicit. Furthermore, by extending EMPC to the cases of 1-norm and infinite-norm objective functions, and explicitly defining the LP-based MPC structure as the corresponding piecewise radiation function and state control law, the EMPC algorithm fully demonstrates its computational advantages in the offline solution process. It transforms online optimization into offline search, significantly reducing the computation time for control parameters and improving computational efficiency.

[0082] However, in practical applications, due to the complex and ever-changing actual operating conditions of AV systems, as the prediction step size increases, the storage space required to store the control laws determined by EMPC also increases, which puts an impact on the vehicle's limited storage space.

[0083] Based on the above technical problems, embodiments of this application provide a method, apparatus, device, medium, and program product for determining control parameters.

[0084] It should be noted that the control parameter determination method provided in this application embodiment can be implemented through the vehicle's autonomous driving system.

[0085] Figure 2 This is a flowchart illustrating the control parameter determination method provided in the embodiments of this application, as shown below. Figure 2 As shown, the process may include the following steps:

[0086] S201. Determine the first association based on the current prediction step size of the autonomous driving system.

[0087] The first correlation includes the correlation between state error and control parameters.

[0088] In one implementation, the autonomous driving system may be configured in a vehicle; exemplarily, the vehicle may be a car equipped with an autonomous driving device or an autonomous driving system; exemplarily, the vehicle may be an AV.

[0089] In one implementation, autonomous driving can be triggered at the current moment or it can have been running for some time.

[0090] In one implementation, the current prediction step size may include the time length corresponding to any prediction operation of the vehicle's autonomous driving system or during autonomous driving, or the average time length corresponding to any prediction operation within a preset time period.

[0091] In one implementation, the current prediction step length can be the path length associated with the time length corresponding to a prediction operation during the vehicle's autonomous driving process, at the current moment or within the current time period.

[0092] It should be noted that before determining the first association, the current prediction step size can be determined using any of the following methods:

[0093] The current prediction step size is determined in response to the user's setting of the corresponding autonomous driving mode for the autonomous driving system.

[0094] The current prediction step size is determined based on the path parameters of the reference trajectory corresponding to the vehicle during autonomous driving. For example, the path parameters of the reference trajectory may include the length, gradient and / or gradient changes of the reference trajectory. For instance, the current prediction step size corresponding to the reference trajectory of the first gradient may be smaller than the current prediction step size corresponding to the reference trajectory of the second gradient, wherein the first gradient may be greater than the second gradient.

[0095] In one implementation, control parameters can be used to adjust the vehicle's autonomous driving system; for example, the control parameters can adjust the vehicle's autonomous driving system in at least one dimension of driving direction, driving speed, and driving time.

[0096] In one implementation, the state error may include the error between the driving state embodied by the autonomous driving system and the reference state; for example, the driving state embodied by the autonomous driving system may include the driving speed and / or driving direction of the vehicle during the autonomous driving process; the reference state may include the reference driving speed and / or driving direction of the vehicle corresponding to the reference trajectory.

[0097] In one implementation, the first association can be predetermined and stored in the vehicle's storage space.

[0098] In one implementation, the first correlation can be represented in the form of a function or a table, in which the state error and the control parameters can be represented in a one-to-one correspondence.

[0099] In one implementation, the first association relationship can be determined in the following way:

[0100] The EMPC algorithm pre-calculates based on vehicle parameters and autonomous driving status to determine the relationship between the vehicle's state error and control parameters at the current prediction step size. For example, vehicle parameters may include the vehicle's weight, load range, tire size, drive mode, and path parameters of the predicted path. The drive mode may include front-wheel drive, rear-wheel drive, or four-wheel drive.

[0101] S202. Based on the current state error and the first correlation, determine the current control parameters.

[0102] In one implementation, the current state error may include the error between the driving state corresponding to the vehicle's autonomous driving system and the reference state at a certain moment or time period; such as the error between the driving state corresponding to the vehicle's autonomous driving system and the reference state at the current moment or time period.

[0103] In one implementation, the current state error of the vehicle can be determined before determining the current control parameters by:

[0104] The driving status of the vehicle in autonomous driving is detected or collected in real time to obtain the real-time driving status, and the degree of difference between the real-time driving status and the reference status is quantified as the current error status.

[0105] In one implementation, the current control parameters can be used to adjust the autonomous driving system of the vehicle at the current moment or during the current time period; for example, the current control parameters can be used to adjust at least one of the direction, speed and time of the autonomous driving system at a certain moment or during a certain time period; such as to adjust at least one of the direction, speed and time of the autonomous driving system at the current moment or during the current time period.

[0106] In one implementation, the current control parameters can be determined in the following way:

[0107] Based on the matching relationship between the state error in the first association relationship and the current state error, the control parameter corresponding to the state error that matches the current state error is determined from the first association relationship, and this control parameter is determined as the current control parameter.

[0108] In one implementation, the vehicle can be controlled to continue performing autonomous driving based on the current control parameters.

[0109] In one implementation, controlling the vehicle to continue performing autonomous driving can be achieved in the following way:

[0110] Within a specified time period, the vehicle's driving direction and / or speed are controlled based on the current control parameters to ensure the vehicle continues to perform autonomous driving; wherein, the specified time period may include the period between the time when the current control parameters are determined and the time when the next current control parameters are determined.

[0111] As can be seen from the above, the control parameter determination method provided in this application determines a first correlation relationship based on the current prediction step size of the autonomous driving system. The first correlation relationship includes the correlation between the state error and the control parameters, thereby shortening the time required to determine the first correlation relationship. Furthermore, by determining the first correlation relationship based on the current prediction step size, the matching degree between the first correlation relationship and the autonomous driving system can be improved, thus increasing the accuracy of the first correlation relationship. At the same time, the current control parameters are determined based on the current state error and the first correlation relationship, which includes the correlation between the state error and the control parameters. Thus, through the above operations, not only can the accuracy of the current control parameters be improved, but the efficiency of determining the current control parameters of the vehicle can also be improved, thereby meeting the actual control parameter determination requirements of the vehicle.

[0112] Based on the foregoing embodiments, in the control parameter determination method provided in this application, before determining the first correlation based on the current prediction step size of the autonomous driving system, the following steps can be performed:

[0113] S-A1: Determine the current tracking status.

[0114] In one implementation, the current tracking state may include the state or effect of the vehicle tracking the reference trajectory during autonomous driving at a certain moment or time period; such as the state or effect of the vehicle tracking the reference trajectory during autonomous driving at the current moment or time period.

[0115] In one implementation, the current tracking state can be represented as a fuzzy interval; for example, different tracking states may correspond to different fuzzy intervals.

[0116] In one implementation, the current tracking state can be determined in the following way:

[0117] The current tracking state is determined based on the vehicle's tracking efficiency during autonomous driving; for example, the current tracking state may change as the tracking efficiency changes, such as as the tracking efficiency decreases as the vehicle's prediction step length increases.

[0118] S-A2: The current tracking state is processed by a fuzzy control model to obtain the current prediction step size.

[0119] Among them, the fuzzy control model is used to characterize the relationship between at least two parameters associated with the autonomous driving process.

[0120] In one implementation, the degree of influence of at least two parameters on the control efficiency, stability, and safety of the vehicle's autonomous driving process can be greater than or equal to a threshold.

[0121] In one implementation, at least two parameters are associated with at least the vehicle's prediction step size, or at least two parameters may include the vehicle's prediction step size; for example, the prediction step size included in the vehicle's fuzzy control model may be multiple.

[0122] In one implementation, the fuzzy control model may include a series of processes and steps that pre-process at least two parameters associated with the autonomous driving process of the vehicle through fuzzy control theory, including fuzzification, logical judgment, and defuzzification.

[0123] In one implementation, the fuzzy control model may include the sum of models or mapping relationships that convert between at least two types of parameters; for example, the above-mentioned models or mapping relationships may be represented in the form of functions or tables.

[0124] In one implementation, the current prediction step size can be obtained in the following way:

[0125] First, the current tracking state is divided into intervals to determine the current fuzzy state interval corresponding to the current tracking state. Then, based on the correspondence between the fuzzy state interval and the fuzzy prediction step size in the fuzzy control model, the current fuzzy prediction step size interval corresponding to the current fuzzy state interval is determined. Next, the current fuzzy prediction step size interval is defuzzified, and the result of the defuzzification process is determined as the current prediction step size.

[0126] As can be seen from the above, the control parameter determination method provided in this application uses a fuzzy control model to characterize the correlation between at least two parameters associated with the autonomous driving process of a vehicle. Thus, through the fuzzy control concept upon which the fuzzy control model is based, the correlation between at least two parameters associated with the autonomous driving process of a vehicle can be accurately and comprehensively reflected. Furthermore, after determining the current tracking state, the current tracking state is processed by the fuzzy control model to obtain the current predicted step size, which can improve the efficiency of obtaining the current predicted step size. At the same time, when the parameters among the at least two parameters include multiple predicted step sizes, the above operation can enhance the richness and comprehensiveness of the current predicted step size, thereby meeting the need to determine the current predicted step size under various conditions.

[0127] Based on the foregoing embodiments, in the control parameter determination method provided in this application, the fuzzy control model includes a second correlation relationship; the second correlation relationship includes the correlation relationship between the prediction step size and the tracking state.

[0128] In one implementation, the second association relationship may include a one-to-one or many-to-one correspondence between multiple prediction step sizes and multiple tracking states of the autonomous driving system; for example, the second association relationship may be represented by a table or function; for example, Table 1 is a data summary diagram of the second association relationship provided in the embodiments of this application.

[0129]

[0130] Table 1

[0131] As shown in Table 1, the prediction step size can include three values: 5, 10, and 15. When the tracking state is represented by fuzzy symbols S, M, and L, the prediction step size can be the corresponding option in Table 1, depending on the actual autonomous driving requirements.

[0132] For example, when the fuzzy representation of the tracking state is S, M and L respectively, it can be used to characterize the tracking effect when the tracking state is a small, medium and large interval, respectively; among them, when the tracking state is L, it can characterize large interval tracking, and the tracking step size can be appropriately reduced to reduce the tracking error; and the second correlation in Table 1 is only an example and can be adjusted in actual settings.

[0133] It should be noted that in practical applications, the fuzzy representation of the tracking state can be limited to three types, and the prediction step size is not limited to the results shown in Table 1.

[0134] Accordingly, the current prediction step size is obtained by processing the current tracking state through a fuzzy control model, which can be achieved in the following way:

[0135] The current prediction step size is determined based on the degree of matching between the current tracking state and the tracking state in the second correlation.

[0136] In one implementation, the prediction step size corresponding to the tracking state that matches the current tracking state in the second association relationship can be determined as the current prediction step size.

[0137] As can be seen from the above, the control parameter determination method provided in this application includes a second correlation between the predicted step size and the tracking state in the fuzzy control model. Furthermore, based on the degree of matching between the current tracking state and the tracking state in the second correlation, the current predicted step size can be determined. Thus, through the second correlation, the interaction and mutual influence between the predicted step size and the tracking state during the autonomous driving process of the vehicle can be fully reflected; and through the above operations, the efficiency of determining the current predicted step size can be improved.

[0138] Based on the foregoing embodiments, in the control parameter determination method provided in this application, the fuzzy control model includes a third correlation relationship; the third correlation relationship includes the correlation relationship between tracking error and tracking state.

[0139] In one implementation, the tracking error can characterize the degree of difference between the autonomous driving system and the reference trajectory in at least one dimension; wherein the at least one dimension may include the lateral and longitudinal dimensions of the vehicle in the horizontal direction, and accordingly, the tracking error may include at least the lateral tracking error and the longitudinal tracking error.

[0140] In one implementation, the third correlation can be represented by a function or a table, and the third correlation can include a one-to-one correspondence between tracking error and tracking state; for example, the third correlation can be represented by Table 2. As shown in Table 2, the error fuzzy sets corresponding to the lateral tracking error and the longitudinal tracking error can be the same, that is, they respectively include PS, PM, PL, NS, NM and NL, while the value range of the tracking state can include S, M and L; for example, the error range of the tracking error corresponding to PS to NL can show a sequentially increasing trend.

[0141] Error fuzzy set PS PM PL NS NM NL PS S S M S S M PM S M M S M M PL M M L M M L NS S S M S S M NM S M M S M M NL M M L M M L

[0142] Table 2

[0143] As shown in Table 2, when the lateral tracking error and the longitudinal tracking error change, their corresponding fuzzy sets can also change accordingly. At this time, the tracking state determined according to the third correlation relationship can also change accordingly.

[0144] Accordingly, the current tracking status can be determined in the following ways:

[0145] Determine the current tracking error; based on the current tracking error and the third correlation, determine the current tracking status.

[0146] The current tracking error includes the longitudinal tracking error and the lateral tracking error corresponding to the current time of the autonomous driving system.

[0147] In one implementation, the current time may include the current moment or the current time period.

[0148] In one implementation, the current tracking error can be determined in the following way:

[0149] The system collects the actual driving trajectory of the vehicle in real time during the autonomous driving process. Then, based on the errors between the actual driving trajectory and the reference trajectory in the lateral and longitudinal dimensions, the longitudinal tracking error and the lateral tracking error are determined respectively based on the errors in the above two dimensions.

[0150] In one implementation, the current tracking state can be determined in the following way:

[0151] Based on the longitudinal and lateral tracking errors corresponding to the current time, and the matching relationships between the longitudinal and lateral tracking errors in the third correlation, the tracking state that matches the longitudinal and lateral tracking errors corresponding to the current time is determined from the third correlation, and this tracking state is determined as the current tracking state.

[0152] It should be noted that the fuzzy control model provided in this application embodiment may include both a second correlation relationship and a third correlation relationship; and when applying the fuzzy control model, the current tracking state can be determined first through the third correlation relationship, and then the current prediction step size can be determined through the second correlation relationship.

[0153] As can be seen from the above, in the control parameter determination method provided in this application embodiment, the fuzzy control model includes a third correlation between tracking error and tracking state. Thus, through the third correlation, the fuzzy control model can comprehensively demonstrate the mutual influence between tracking state and tracking error. Furthermore, after determining the current tracking error, which includes the longitudinal tracking error and the lateral tracking error corresponding to the current time, the current tracking state is determined based on the current tracking error and the third correlation. This can improve the efficiency of determining the current tracking state and also improve the accuracy of the current tracking state.

[0154] Based on the foregoing embodiments, in the control parameter determination method provided in this application, before determining the current control parameter based on the current state error and the first correlation relationship, the following operations may also be performed:

[0155] Based on the current prediction step size and the set of relationships, the first association relationship is determined.

[0156] The relationship set includes the set of relationships between the prediction step size set and the autonomous driving parameters; the autonomous driving parameters include the set of state errors and control parameters.

[0157] In one implementation, the relation set may include the correspondence between the t-th prediction step size and the t-th group of fourth associations. That is, the relation set may include a set of at least two prediction steps and their corresponding fourth associations. The fourth association may include the association between the state error and the control parameters, and t is an integer greater than or equal to 1.

[0158] In one implementation, the set of relations can be determined in the following way:

[0159] The third and second correlation relationships in the fuzzy control model are predetermined to determine multiple prediction step sizes for the vehicle during the autonomous driving process. Then, the fourth correlation relationship under the k-th prediction step size condition is calculated using the EMPC algorithm. The set of the fourth correlation relationships under the first prediction step size condition to the fourth correlation relationship under the K-th prediction step size condition is then determined as the relation set. Here, k is an integer greater than or equal to 1, and K is an integer greater than k.

[0160] In one implementation, the first association relationship can be determined in the following way:

[0161] Based on the matching relationship between the prediction step size in the relation set and the current prediction step size, the fourth association relationship corresponding to the prediction step size that matches the current prediction step size in the relation set is determined as the first association relationship.

[0162] As can be seen from the above, the control parameter determination method provided in this application embodiment includes a relationship set comprising the association between a prediction step size set and a vehicle's autonomous driving parameter set. Based on the current prediction step size and the relationship set, a first association relationship is determined, and the autonomous driving parameter set includes a set of state errors and control parameters. Thus, determining the first association relationship in this manner before determining the current control parameters improves the efficiency of determining the first association relationship. Furthermore, since the relationship set includes a set of associations between the prediction step size set and the vehicle's autonomous driving parameter set, and the autonomous driving parameter set includes a set of vehicle state errors and control parameters, the richness of the data in the relationship set, including the prediction step size set and autonomous driving parameters, can be increased.

[0163] Based on the foregoing embodiments, the control parameter determination method provided in this application can also perform the following steps:

[0164] S-E1, determine the trajectory tracking optimization function.

[0165] Specifically, the trajectory tracking optimization function is associated with at least the kinematic prediction model; the trajectory tracking optimization function is associated with at least the increments of the state error and control parameters.

[0166] In one implementation, the vehicle's kinematic prediction model can be obtained by... Figure 3 The kinematic prediction model shown was determined through analysis and derivation. Figure 3 A schematic diagram illustrating the principle of a vehicle's kinematic prediction model, such as... Figure 3 As shown, both reference vehicle 301 and vehicle 302 can be AV; and in the global coordinate system XOY, the rear axle of vehicle 302 is located at position (x, y) with a heading angle of . The vehicle travels at speed v, with a wheel track of L and a front wheel steering angle of δ. Reference trajectory 303 is defined relative to reference vehicle 301; the rear axle of reference vehicle 301 is located at position (x...). r y r Its heading angle is

[0167] The vehicle speed can be decomposed according to the direction of the global coordinate system using equations (1) and (2) as follows:

[0168]

[0169] At the same time, by combining the heading angle and the front wheel steering angle, the calculation process shown in equation (3) can be performed:

[0170]

[0171] Therefore, by combining equations (1) to (3), we can obtain the kinematic model of the vehicle as shown in equation (4):

[0172]

[0173] At the same time, the kinematic model of the reference vehicle, as shown in equation (5), can also be obtained:

[0174]

[0175] To simplify the representation of the data, the state variables and control variables of the vehicle and the reference vehicle can be simplified as shown in equation (6):

[0176]

[0177] Therefore, based on equations (4) to (6), the kinematic models of the vehicle and the reference vehicle, as shown in equations (7) to (8), can be determined respectively:

[0178]

[0179] Next, at each sampling time of the reference trajectory, the kinematic model (7) is linearized using Taylor expansion, and higher-order terms are ignored, resulting in equation (9):

[0180]

[0181] By performing the difference operation on equations (8) to (9), we can obtain equation (10):

[0182]

[0183] If we perform variable substitution according to equations (11) to (12):

[0184]

[0185] Then we can obtain the kinematic error model shown in equation (13):

[0186]

[0187] Based on this, by discretizing the model shown in equation (13) using the forward Euler method, we can obtain equation (14):

[0188]

[0189] Where T is the time step during simulation, and k represents the k-th time. By rearranging equation (14), we can obtain equation (15):

[0190]

[0191] The variables in equation (15) can be represented by equation (16):

[0192]

[0193] In the field of control parameter determination, predictive models are used to predict the future state of a vehicle based on the current state error and the increment of control parameters. The discretized error model shown in equations (15) to (16) is actually about the relationship between the future state at time k+1 and the current state at time k; based on this, the vehicle state at multiple time steps can be predicted by iteratively calculating in (15).

[0194] To iterate over equation (15), we can first define the prediction time domain N. p and control time domain N c And set N p Greater than N c Next, perform N on equation (15). p This iteration predicts the future N. p The state error of the step is shown in equation (17):

[0195]

[0196] Through the above iterative processing, a new state-space prediction model can be derived to construct the prediction model; wherein, the increments of the prediction state error and the control quantity can be stacked as shown in Equation (18):

[0197]

[0198] Thus, we can obtain the stacked vehicle kinematics prediction model based on the current state error and control parameters, as shown in equation (19):

[0199]

[0200] The parameters in equation (19) can be represented as shown in equation (20):

[0201]

[0202] In order for the vehicle to track the reference trajectory with minimal deviation, the effects of state error and control input increment should be considered when constructing the cost optimization function at each sampling time. Furthermore, the goal of the controller used to implement the automatic control function is to minimize the control error as much as possible while tracking the reference trajectory defined by the reference position and yaw angle. More specifically, the cost optimization function of the model shown in Equation (19) should follow the principle of convergence of state error and control input error. Therefore, the cost function design for the optimization problem of trajectory tracking control is shown in Equation (21).

[0203]

[0204] Where Q and R are respectively related to Y(k) and The corresponding weights are used to characterize the importance of their respective parameters. Substituting equation (19) into equation (21) yields the trajectory tracking optimization function shown in equation (22):

[0205]

[0206] The relevant parameters in equation (22) are shown in equation (23):

[0207]

[0208] S-E2, an explicit function model for determining the relationship between state error and control parameters based on trajectory tracking optimization function.

[0209] In one implementation, the explicit function model may include a function with state error as the independent variable and control parameters as the dependent variable; for example, the explicit function model may present the process of control parameters changing with the state error in a linear or nonlinear manner.

[0210] In one implementation, the explicit function model can be determined in the following way:

[0211] The trajectory tracking optimization function is transformed and solved using EMCP, and the result of the transformation is determined as the explicit function model.

[0212] S-E3: Determine the set of relations through explicit function models.

[0213] In one implementation, the above-described solution process can determine the display function model under a prediction step size. Thus, by changing the prediction step size, display function models under different prediction step sizes can be obtained, and the set of relationships between state error and control parameters reflected by the set of the above-described display function models is determined as a relation set.

[0214] As can be seen from the above, the control parameter determination method provided in this application determines a trajectory tracking optimization function and, based on the trajectory tracking optimization function, determines an explicit function model relating the vehicle's state error and control parameters. Furthermore, the trajectory tracking optimization function is at least associated with a kinematic prediction model and at least with the increments of the state error and control parameters. Thus, through the above operations, not only can the comprehensiveness and diversity of parameters in the trajectory tracking optimization function be improved, but the consistency between the trajectory tracking optimization function and the motion state during autonomous driving can also be enhanced. Moreover, the explicit function model simplifies the expression of the relationship between the vehicle's state error and control parameters during autonomous driving. Based on this, determining the relation set through the explicit function model improves the stability of the relation set determination and the accuracy of the data in the relation set.

[0215] Based on the foregoing embodiments, the automated driving method provided in this application, which determines the display function model between state error and control parameters based on the trajectory tracking optimization function, can be achieved through the following steps:

[0216] S-G1 transforms the trajectory tracking optimization function to obtain an intermediate optimization function that includes at least control parameters and state errors.

[0217] In practical applications, the MPC algorithm solves its optimization function at each sampling time to achieve precise and smooth control of the vehicle's autonomous driving process. However, as mentioned earlier, the above-mentioned solution process of MPC requires repeated online optimization. Therefore, it is suitable for slower or simpler dynamic control processes.

[0218] EMPC can transform the online process of repeatedly solving QP problems into a multi-parameter QP problem, and obtain and store the explicit control law between state input and control output through different critical regions. In practical applications, it can directly search for the explicit control law based on the current vehicle state through point positioning, which can directly reduce the computation time of online optimization problem solving. This application embodiment will draw on the above ideas of EMPC to transform the trajectory tracking optimization function, and the process can be as follows:

[0219] Assume Q≥0, R≥0 and It can be determined that, at this point, the trajectory tracking optimization function represented by equations (21) to (22) can be transformed into the intermediate optimization function shown in equation (24):

[0220]

[0221] question.

[0222] S-G2, Determine the control sequence.

[0223] The control sequence includes multiple distinct control parameters.

[0224] For the MPC algorithm, its optimal solution at each time step Regarding state error The result is obtained through calculation. For the EMPC algorithm, a unique optimal solution needs to be selected as the optimal control sequence for the system; the optimal control sequence can be shown in equation (25):

[0225]

[0226] In Equation (25), each element corresponds to the optimal control error of M-1 state error partitions, which are multiple control parameters contained in the control sequence; M is an integer greater than 2.

[0227] S-G3 discretizes the state error to obtain the discrete state error result.

[0228] Accordingly, the same operation as the MPC algorithm can be used to perform stacked discretization on equation (17) to obtain the state error discretization result shown in equation (26):

[0229]

[0230] S-G4, based on the control sequence and the discrete results of state error, processes the intermediate optimization function to obtain an explicit function model.

[0231] For example, equation (26) can be substituted into equation (24) to obtain equation (27):

[0232]

[0233] H, F, G, W and E can all be calculated from the two weight matrices Q and R, while H and F are the weight matrices of equation (27), and G, W and E are the weight matrices of the constraints in equation (27).

[0234] To simplify the expression, a new variable z can be defined using equation (28):

[0235]

[0236] in, For a non-repeating sequence of system state variables; in this case, equation (27) can be transformed into equation (29):

[0237]

[0238] In the above case, based on the normal solution of the QP problem, the KKT conditions can be expanded in detail, as shown in equation (30):

[0239]

[0240] Where λ is a Lagrange multiplier, and by transforming equation (30) with the relaxation complementarity condition, we can obtain equation (31):

[0241]

[0242] Grandes multipliers:

[0243]

[0244] The explicit relationship between the optimal control law and the state error is shown below:

[0245]

[0246] Among them, CR i Let represent the i-th feasible region that satisfies the constraints.

[0247] For example, equation (33) can be a generalized expression of an explicit function model.

[0248] In practical applications, EMPC only uses the first control error of each control sequence to control the vehicle and constructs the control law between the non-repeating state error and the control error. Therefore, after the above process, all state error partitions and their corresponding control laws are stored. Thus, online optimization can be transformed into a lookup process using offline control laws through equation (33). For example, when the EMPC algorithm is called, the lookup rule shown in equation (34) is executed:

[0249]

[0250] For example, equation (34) can be a refined expression of the explicit function model.

[0251] As can be seen from the above, the control parameter determination method provided in this application, by transforming the trajectory tracking optimization function, obtains an intermediate optimization function that includes at least control parameters and state error, thereby enabling the transformation of the trajectory tracking optimization function from one parameter to at least two parameters; furthermore, by determining a control sequence including multiple distinct control parameters, discretizing the state error to obtain a state error discretization result, and then post-processing the intermediate optimization function based on the control sequence and the state error discretization result to obtain an explicit function model, the processing efficiency of the intermediate optimization function can be improved, thereby improving the determination efficiency of the explicit function model.

[0252] Figure 4 This is a second flowchart illustrating the control parameter determination method provided in the embodiments of this application, as shown below. Figure 4 As shown, the process may include the following steps:

[0253] S401, Reconstruct the sequence of control and state variables.

[0254] For example, the reconstruction of the sequence of control variables and state variables can be achieved through equations (25) to (26).

[0255] S402. Establish a multi-parameter optimization function.

[0256] For example, a multi-parameter optimization function can be established using equation (24).

[0257] For example, the multi-parameter optimization function may include the intermediate optimization function in the foregoing embodiments.

[0258] S403, Solve and store all control laws.

[0259] For example, all control laws can be solved and stored using equations (33) to (34).

[0260] For example, all control laws can be stored in the form of second and third associations.

[0261] S404. Fuzzy control: Select an appropriate control law.

[0262] For example, the current state error output by the relevant sub-step in S405 can be processed by the fuzzy control model to determine the current prediction step size, and the current prediction step size can be sent to the sub-step in the feedback control 405 for offline calculation so that it can determine the final current control parameters.

[0263] S405, Feedback Control

[0264] For example, S404 may include the following sub-steps:

[0265] S4051, Preprocessing.

[0266] For example, preprocessing can achieve the same function as the determination unit mentioned above.

[0267] S4052, Feedback Correction.

[0268] For example, feedback correction can achieve the function of the feedback correction unit mentioned above.

[0269] S4053, offline processing.

[0270] For example, offline processing can determine the specific current control parameters by looking up a table online for the control law calculated according to equation (34), thereby realizing offline processing.

[0271] S4054, Predictive processing.

[0272] For example, the prediction processing can implement the functionality of the prediction unit described above.

[0273] S4055, Controlling Driving.

[0274] For example, controlling driving can achieve the function of the control unit mentioned above.

[0275] As can be seen from the above process, the control parameter determination method provided in this application embodiment uses a fuzzy control algorithm to adjust the current prediction step size in real time, and then determines the current driving parameters in a targeted manner based on the current prediction step size, thereby improving the efficiency of AV path tracking control and enhancing the flexibility and real-time performance of AV path tracking control.

[0276] Based on the foregoing embodiments and the derivation process of the relevant formulas, it can be seen that the control parameter determination method provided in this application is a combination of fuzzy control and EMPC. Therefore, the control parameter determination method provided in this application can be abbreviated as FuzzyEMPC.

[0277] Figure 5A A schematic diagram of the vehicle structure simulated for the control parameter determination method provided in the embodiments of this application, such as... Figure 5AAs shown, vehicle 302 can be an AV; and during the simulation, parameters such as the rigid spring mass, vehicle class (B-Class), hatchback spring mass, and aerodynamic parameters can be set for vehicle 302. The simulation data display can also be set to a 3D vehicle model. The powertrain of vehicle 302 is front-wheel drive, the braking system is a four-wheel drive system, and the steering system is four-wheel steering. The front suspension is generic / independent. The springs, dampers, and suspension compatibility can be set to Front Comp. The tires can be specified as all four tires alike and all 185 / 65R15. The rear suspension... Suspension can be configured for both generic and independent kinematics.

[0278] Figure 5B A schematic diagram of vehicle parameters simulated for the control parameter determination method provided in the embodiments of this application, as shown below. Figure 5B As shown, the height between the tire center and the roof of vehicle 302 can be 1535, the vehicle width can be 1695, the height of the midpoints of both left and right wheels is 310, the horizontal distance between the center of gravity and the midpoint of the front wheel in the unloaded state is 1040, and the height of the cargo box floor is 350; all units can be millimeters (mm). Other simulation configuration data for vehicle 302 are shown in Table 3:

[0279] Sprung mass 1110kg Roll inertia 440.6 kg-m2 Pitch inertia 1343.1 kg-m2 Yaw inertia 1343.1 kg-m2 Center point location (Rx, Ry, Rz) (0.630，1.100，1.100)(m)

[0280] Table 3

[0281] The relationship between inertia and radius of gyration can be calculated using equation (35):

[0282] I=G*r*r(35)

[0283]

[0284] Table 4

[0285] Where G represents the mass of the vehicle, and r represents the wheel radius of the vehicle.

[0286] Table 4 summarizes the results of the key simulation parameters.

[0287] Figure 5C This is a schematic diagram of the data transmission structure of the vehicle simulation model provided in the embodiments of this application, as shown below. Figure 5C As shown in the figure, the control parameter determination method provided in this application embodiment can be implemented through the S_Function in the simulation model 501 shown in the figure. Wherein: Vehicle Code is used to represent vehicle 302; X0, Y0, Yaw, VX, and Steer_SW are used to represent the vehicle's real-time position in the x-axis direction, position in the y-axis direction, heading angle, real-time speed, and front wheel steering angle in the geodetic coordinate system, respectively; XY Graph is used to generate the tracking path; V, Delta, x_error, y_error, and Lat_error correspond to the speed error vv, respectively. r Front wheel steering angle error δ-δ r Lateral position error xx r Longitudinal position error yy r Position error and heading angle error Among them, the control error consisting of speed error and front wheel steering angle error, and the state error consisting of lateral position error, longitudinal position error and heading angle error can be represented by equation (6); and the in&out unit is used to determine the position error. Measure the tracking performance; for example, different tracking performances may correspond to different current prediction step sizes.

[0288] For example, u*3.6, U*180 / pi and 0, which are associated with Vehicle Code, are used to convert the speed unit from m / s to k / h, the front wheel angle unit from rad to degrees, and to set the default rear wheel angle of a front-wheel drive vehicle to 0, respectively.

[0289] Figure 6A This is a schematic diagram illustrating the tracking effect of the control parameter determination method provided in the embodiments of this application, such as... Figure 6A As shown, both the horizontal and vertical coordinates of the coordinate system represent distance, and the unit is meters (m).

[0290] In the figure, the reference trajectory is defined as a circle with the same shape as the roundabout, and the reference trajectory can be expressed by the parametric equation shown in equation (36) so that an error model can be constructed by subtraction.

[0291]

[0292] As can be seen from the figure, the tracking performance of EMPC, MPC, and the FuzzyEMPC provided in this embodiment all show the same trend as the tracking performance of the reference trajectory, and the position error remains below 0.5 meters. However, the distance between the trajectory corresponding to the tracking performance of EMPC and the reference trajectory is greater than the distance between the trajectory corresponding to MPC and the reference trajectory corresponding to FuzzyEMPC. In other words, during the execution of the control parameter determination method provided in this embodiment, the vehicle's tracking performance on the reference trajectory is consistent with that of MPC, and is superior to the tracking performance of EMPC.

[0293] Figure 6B The following is a schematic diagram illustrating the tracking error effects of the three tracking methods provided in the embodiments of this application: Figure 6B As shown, the horizontal and vertical axes of the three coordinate systems in the figure represent the same physical meaning. The horizontal axis represents time, in seconds (s), and the vertical axis represents distance, in meters (m). Figure 6B In the figure, coordinate systems (a) to (c) respectively illustrate the lateral error (X Error), longitudinal error (Y Error), and position error (Position Error) in the control parameter determination process of MPC, EMPC, and FuzzyEMPC. As can be seen from the figure, the trends of change in the lateral error, longitudinal error, and position error dimensions of the control parameter determination process for the three methods are basically consistent. However, overall, the errors of FuzzyEMPC in these three dimensions stably follow the error curves corresponding to MPC. Table 5 shows... Figure 6B The corresponding error data summary results. The controller type can include control devices determined by control parameters set using MPC, EMPC, and FuzzyEMPC methods respectively.

[0294] Controller Type Lateral error (m) Longitudinal error (m) Position error (m) MPC 0.2232 0.1939 0.2957 EMPC 0.2675 0.4704 0.4531 FuzzyEMPC 0.3423 0.2968 0.3491

[0295] Table 5

[0296] from Figure 6B As can be seen from Table 5, the longitudinal error and positional distance error of the FuzzyEMPC provided in this application embodiment are smaller than those of EMPC, while the lateral error is slightly larger than that of EMPC.

[0297] Figure 6C This is a schematic diagram illustrating the comparative effects of speed control provided in an embodiment of this application. Figure 6C In the coordinate system shown, the horizontal axis represents time, in seconds (s), and the vertical axis represents speed, in meters per second (m / s). From Figure 6CIt can be seen that the speed control curve of FuzzyEMPC provided in this application embodiment is basically consistent with the speed control curve of MPC, while the difference between the speed control curve of EMPC and the speed control curve of MPC is more obvious.

[0298] Table 6 summarizes the speed control data.

[0299] Controller Type Effective speed (m / s) Velocity error (m / s) MPC 4.9338 0.0672 EMPC 4.9959 0.0041 FuzzyEMPC 4.9455 0.0555

[0300] Table 6

[0301] As can be seen from Table 6, both EMPC and FuzzyEMPC control methods are superior to MPC in terms of speed error, but the speed error of FuzzyEMPC is relatively larger than that of EMPC.

[0302] Figure 6D This diagram illustrates a comparison of the effects of the prediction step size and the state space of the control law provided in an embodiment of this application. Figure 6D In the included figures (d) to (f), different color intervals correspond to the state spaces of different control laws. The prediction step sizes for figures (d) to (f) can be 5, 10, and 15, respectively. Figure 6D It can be seen that as the prediction step size increases, the richness of the state space of the control law is significantly improved, which means that the control accuracy of the vehicle's autonomous driving process is also improved. However, in order to achieve high-precision autonomous driving control, the corresponding offline table lookup time will also be extended.

[0303] Figure 6E This is a comparative diagram illustrating the computational efficiency provided in the embodiments of this application. For example... Figure 6E In the coordinate systems shown in Figures (g) to (i), the horizontal axis represents the simulation time, and the vertical axis represents the time for one optimization calculation. Table 7 shows the results. Figure 6E The corresponding computational efficiency data summary results.

[0304] Controller Type Time (s) MPC 0.16621 EMPC 0.00088 FuzzyEMPC 0.00297

[0305] Table 7

[0306] from Figure 6E As shown in Table 7, the average control time of FuzzyEMPC is 0.00297 seconds, which is significantly less than that of MPC (0.16621 seconds). Compared with MPC, FuzzyEMPC's calculation speed is about 55 times faster. However, due to the addition of fuzzy control and the process of determining the current prediction step size, FuzzyEMPC is relatively longer than EMPC's 0.0088 seconds.

[0307] In summary, compared with MPC, in terms of tracking performance, the FuzzyEMPC provided in this application embodiment can achieve accurate trajectory tracking control of AV under the same simulation conditions; in terms of computational efficiency, FuzzyEMPC can effectively shorten the computation time by converting online optimization to offline calculation, selecting an appropriate control law based on the tracking effect at the current moment, and directly determining the current control parameters by using the pre-determined mapping relationship between the control law and the state error, thereby meeting the control requirements of complex autonomous driving scenarios with high real-time performance and high accuracy.

[0308] Based on the foregoing embodiments, this application also provides a control parameter determination device. Figure 7 This is a schematic diagram of the control parameter determination device provided in the embodiments of this application, as shown below. Figure 7 As shown, the control parameter determining device 7 may include:

[0309] The determination module 701 is used to determine a first correlation based on the current prediction step size of the autonomous driving system; wherein, the first correlation includes the correlation between the state error and the control parameters;

[0310] The determination module 701 is also used to determine the current control parameters based on the current state error and the first correlation relationship.

[0311] In some embodiments, the determining module 701 is used to determine the current tracking state; process the current tracking state through a fuzzy control model to obtain the current prediction step size; wherein the fuzzy control model is used to characterize the relationship between at least two parameters associated with the autonomous driving process.

[0312] In some embodiments, the fuzzy control model includes a second correlation; the second correlation includes the correlation between the prediction step size and the tracking state;

[0313] The determination module 701 is also used to determine the current prediction step size based on the degree of matching between the current tracking state and the tracking state in the second association relationship.

[0314] In some embodiments, the fuzzy control model includes a third correlation relationship; the third correlation relationship includes the correlation relationship between tracking error and tracking state;

[0315] The determination module 701 is also used to determine the current tracking error; and to determine the current tracking state based on the current tracking error and the third correlation relationship; wherein the current tracking error includes the longitudinal tracking error and the lateral tracking error corresponding to the current time of the autonomous driving system.

[0316] In some embodiments, the determining module 701 is further configured to determine a first association relationship based on the current prediction step size and the relationship set; wherein the relationship set includes a set of association relationships between the prediction step size set and the autonomous driving parameter set; and the autonomous driving parameter set includes a set of state errors and control parameters.

[0317] In some embodiments, the determining module 701 is further configured to determine a trajectory tracking optimization function; wherein the trajectory tracking optimization function is associated with at least the kinematic prediction model; and the trajectory tracking optimization function is associated with at least the increments of the state error and the control parameters;

[0318] The determination module 701 is also used to determine the explicit functional model between the state error and the control parameters based on the trajectory tracking optimization function; and to determine the set of relationships through the explicit functional model.

[0319] In some embodiments, the determining module 701 is further configured to perform a transformation process on the trajectory tracking optimization function to obtain an intermediate optimization function that includes at least control parameters and state error; and determine a control sequence; wherein the control sequence includes multiple control parameters that are different from each other;

[0320] The determination module 701 is also used to discretize the state error to obtain the state error discretization result; based on the control sequence and the state error discretization result, the intermediate optimization function is processed to obtain the explicit function model.

[0321] Based on the foregoing embodiments, this application also provides a control parameter determination device. Figure 8 This is a schematic diagram of the control parameter determining device provided in an embodiment of this application. Figure 8 As shown, the control parameter determining device 8 includes a processor 801 and a memory 802; wherein:

[0322] Memory 802 is used to store computer-executable instructions;

[0323] The processor 801, when executing computer-executable instructions stored in memory, is capable of implementing any of the control parameter determination methods described above.

[0324] This application also provides a computer-readable storage medium storing computer-executable instructions or a computer program, which, when executed by a processor, can implement the control parameter determination method as described above.

[0325] This application also provides a computer program product, including computer-executable instructions or a computer program, which, when executed by a processor, can implement the control parameter determination method as described in any of the preceding claims.

[0326] In some embodiments, the computer-readable storage medium may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM), etc.; or it may be a device that includes one or any combination of the above-mentioned memories.

[0327] In some embodiments, computer-executable instructions may take the form of programs, software, software modules, scripts, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as stand-alone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

[0328] As an example, computer-executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple co-located files (e.g., files that store one or more modules, subroutines, or code sections).

Claims

1. A method for determining control parameters, characterized in that, The method includes: A first correlation is determined based on the current prediction step size of the autonomous driving system; wherein, the first correlation includes the correlation between state error and control parameters; Based on the current state error and the first correlation, the current control parameters are determined.

2. The method according to claim 1, characterized in that, Before determining the first association relationship based on the current prediction step size of the autonomous driving system, the method further includes: Determine the current tracking status; The current tracking state is processed by a fuzzy control model to obtain the current prediction step size; The fuzzy control model is used to characterize the relationship between at least two parameters associated with the autonomous driving process.

3. The method according to claim 2, characterized in that, The fuzzy control model includes a second correlation relationship; the second correlation relationship includes the correlation between the prediction step size and the tracking state; the step of processing the current tracking state through the fuzzy control model to obtain the current prediction step size includes: The current prediction step size is determined based on the degree of matching between the current tracking state and the tracking state in the second association relationship.

4. The method according to claim 2, characterized in that, The fuzzy control model includes a third correlation relationship; the third correlation relationship includes the correlation relationship between tracking error and tracking state; Determining the current tracking status includes: Determine the current tracking error; wherein, the current tracking error includes the longitudinal tracking error and the lateral tracking error corresponding to the current time of the autonomous driving system; The current tracking state is determined based on the current tracking error and the third correlation.

5. The method according to claim 1, characterized in that, Before determining the current control parameters based on the current state error and the first correlation, the method further includes: Based on the current prediction step size and the set of relationships, the first association relationship is determined; The relationship set includes a set of associations between the prediction step size set and the autonomous driving parameter set; the autonomous driving parameter set includes a set of state errors and control parameters.

6. The method according to claim 5, characterized in that, The method further includes: Determine a trajectory tracking optimization function; wherein the trajectory tracking optimization function is at least associated with a kinematic prediction model; and the trajectory tracking optimization function is at least associated with the increments of state error and control parameters; The explicit functional model between the state error and the control parameters is determined based on the trajectory tracking optimization function. The set of relations is determined using the display function model.

7. The method according to claim 6, characterized in that, The explicit functional model for determining the relationship between the state error and the control parameters based on the trajectory tracking optimization function includes: The trajectory tracking optimization function is transformed to obtain an intermediate optimization function that includes at least the control parameters and the state error; Determine a control sequence; wherein the control sequence includes multiple distinct control parameters; Discretize the state error to obtain the discrete state error result; Based on the control sequence and the state error discretization result, the intermediate optimization function is processed to obtain the explicit function model.

8. A control parameter determining device, characterized in that, The control parameter determination device includes: The determination module is used to determine a first correlation based on the current prediction step size of the autonomous driving system; wherein, the first correlation includes the correlation between state error and control parameters; The determining module is further configured to determine the current control parameters based on the current state error and the first correlation relationship.

9. A control parameter determining device, characterized in that, The control parameter determining device includes: Memory is used to store executable instructions for a computer; When the processor executes computer-executable instructions stored in the memory, it is capable of implementing the control parameter determination method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, they can implement the control parameter determination method according to any one of claims 1 to 7.

11. A computer program product comprising computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, they can implement the control parameter determination method according to any one of claims 1 to 7.

Images