Vehicle control parameter calibration method, device and nonvolatile storage medium

Abstract

This invention discloses a vehicle control parameter calibration method, apparatus, and non-volatile storage medium. The method includes: acquiring first motion state information, first planning information, and a pre-calibrated first calibration relationship matching the vehicle model; determining first vehicle control parameters for a first moment based on the first motion state information and the first planning information within the first calibration relationship; after adjusting the motion state according to the first vehicle control parameters, acquiring second motion state information of the vehicle; updating the first calibration relationship based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle; iteratively updating the second calibration relationship to determine a target calibration relationship matching the vehicle. This invention solves the technical problem of mismatch between vehicle control parameters and vehicle operating states caused by the inability of traditional vehicle control parameter calibration methods to comprehensively calibrate vehicle control parameters.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving, and more specifically, to a method, apparatus, and non-volatile storage medium for calibrating vehicle control parameters. Background Technology

[0002] Vehicle control parameters, as the cornerstone of autonomous driving control, determine whether an autonomous vehicle can efficiently follow a planned path and the stability and comfort of vehicle control. This is because the vehicle control output provides the desired acceleration / deceleration information, while the actual control of the vehicle's movement is achieved through the throttle / brake opening information. This requires combining the vehicle's current speed to find the mapping relationship between the throttle / brake opening and the desired acceleration, ensuring the matching of physical quantities during vehicle control. This is achieved through data calibration testing to generate the vehicle's throttle / brake calibration table.

[0003] Traditional vehicle throttle / brake calibration methods place high demands on the calibration test scenario, vehicle condition, road surface conditions, and surrounding environment. Calibration is not only demanding and complex, but also fails to consider the impact of different road conditions on vehicle control parameters due to the limited test scenarios. It is only suitable for single-vehicle testing and calibration, neglecting the manpower, material, and time costs of large-scale vehicle testing and calibration. Furthermore, the limited number of test scenarios covering different load capacities fails to cover all load requirements, ultimately leading to a mismatch between vehicle control parameters and actual vehicle conditions, thus impacting vehicle safety. Therefore, how to efficiently, in real-time, and comprehensively calibrate vehicle control parameters has become a pressing problem to be solved.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This invention provides a vehicle control parameter calibration method, apparatus, and non-volatile storage medium to at least solve the technical problem of mismatch between vehicle control parameters and vehicle operating state caused by the inability of traditional vehicle control parameter calibration methods to fully calibrate vehicle control parameters.

[0006] According to one aspect of the present invention, a vehicle control parameter calibration method is provided, comprising: acquiring first motion state information of a vehicle, first planning information of a vehicle, and a pre-calibrated first calibration relationship matching a vehicle model, wherein the first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information, the vehicle control parameters are parameters for controlling vehicle power, the first motion state information characterizes the motion state of the vehicle at a first moment, and the first planning information characterizes the expected motion state of the vehicle at a second moment after the first moment; determining first vehicle control parameters of the vehicle at the first moment in the first calibration relationship based on the first motion state information and the first planning information; acquiring second motion state information of the vehicle after the vehicle adjusts its motion state according to the first vehicle control parameters, wherein the second motion state information characterizes the motion state of the vehicle at the second moment; updating the first calibration relationship based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle, wherein the second calibration relationship characterizes the correspondence between vehicle control parameters and motion state information in the current state; iteratively updating the second calibration relationship at multiple moments after the second moment using the method for obtaining the second calibration relationship to determine a target calibration relationship matching the vehicle.

[0007] According to another aspect of the present invention, a vehicle control parameter calibration method is also provided, comprising: acquiring a vehicle control parameter range and a vehicle speed range of a test vehicle; selecting a plurality of first sampling points within the vehicle control parameter range and a plurality of second sampling points within the vehicle speed range; inputting the plurality of first sampling points and the plurality of second sampling points into a target neural network model to determine a plurality of vehicle acceleration values, wherein the target neural network model is used to determine the vehicle acceleration values ​​through the vehicle control parameter values ​​and the vehicle speed values; determining a first calibration relationship based on the plurality of first sampling points, the plurality of second sampling points, and the plurality of vehicle acceleration values, wherein the first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles whose vehicle model matches the test vehicle, and the vehicle motion state information includes vehicle speed and vehicle acceleration.

[0008] According to another aspect of the present invention, a vehicle control parameter calibration device is also provided, comprising: a first acquisition module, configured to acquire first motion state information of a vehicle, first planning information of a vehicle, and a pre-calibrated first calibration relationship matching the vehicle model, wherein the first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information, the vehicle control parameters are parameters for controlling vehicle power, the first motion state information characterizes the motion state of the vehicle at a first moment, and the first planning information characterizes the expected motion state of the vehicle at a second moment after the first moment; a first determination module, configured to determine the first vehicle control parameters of the vehicle at the first moment based on the first motion state information and the first planning information in the first calibration relationship; a second acquisition module, configured to acquire second motion state information of the vehicle after the vehicle adjusts its motion state according to the first vehicle control parameters, wherein the second motion state information characterizes the motion state of the vehicle at the second moment; and a generation module, configured to update the first calibration relationship based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle, wherein the second calibration relationship characterizes the correspondence between the vehicle control parameters and motion state information in the current state.

[0009] According to another aspect of the present invention, a vehicle control parameter calibration device is also provided, comprising: a third acquisition module, configured to acquire a range of vehicle control parameters and a range of vehicle speed of a test vehicle; a sampling module, configured to select a plurality of first sampling points within the range of vehicle control parameters and a plurality of second sampling points within the range of vehicle speed; a second determination module, configured to input the plurality of first sampling points and the plurality of second sampling points into a target neural network model to determine a plurality of vehicle acceleration values, wherein the target neural network model is used to determine the vehicle acceleration values ​​through the vehicle control parameter values ​​and the vehicle speed values; and a third determination module, configured to determine a first calibration relationship based on the plurality of first sampling points, the plurality of second sampling points, and the plurality of vehicle acceleration values, wherein the first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles of the same model as the test vehicle, and the vehicle motion state information includes vehicle speed and vehicle acceleration.

[0010] According to another aspect of the present invention, a non-volatile storage medium is also provided, the non-volatile storage medium including a stored program, wherein, when the program is running, the device where the non-volatile storage medium is located is controlled to execute any of the above-described vehicle control parameter calibration methods.

[0011] According to another aspect of the present invention, a computer device is also provided, comprising: a memory and a processor, wherein the memory stores a computer program; and the processor is configured to execute the computer program stored in the memory, wherein the computer program, when running, causes the processor to perform the following method: acquiring first motion state information of a vehicle, first planning information of a vehicle, and a pre-calibrated first calibration relationship matching the vehicle model, wherein the first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information, the vehicle control parameters are parameters for controlling vehicle power, the first motion state information characterizes the motion state of the vehicle at a first moment, and the first planning information characterizes the expected motion state of the vehicle at a second moment after the first moment; Based on the first motion state information and the first planning information, the first vehicle control parameters of the vehicle at the first moment are determined in the first calibration relationship. After the vehicle adjusts its motion state according to the first vehicle control parameters, the second motion state information of the vehicle is obtained, wherein the second motion state information represents the motion state of the vehicle at the second moment. The first calibration relationship is updated according to the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle, wherein the second calibration relationship represents the correspondence between the vehicle control parameters and the motion state information in the current state. The second calibration relationship is iteratively updated at multiple moments after the second moment using the method of obtaining the second calibration relationship to determine the target calibration relationship matching the vehicle.

[0012] According to another aspect of the present invention, a computer device is also provided, comprising: a memory and a processor, the memory storing a computer program; the processor executing the computer program stored in the memory, wherein the computer program, when running, causes the processor to perform the following method: acquiring a range of vehicle control parameters and a range of vehicle speed of a test vehicle; selecting a plurality of first sampling points within the range of vehicle control parameters and a plurality of second sampling points within the range of vehicle speed; inputting the plurality of first sampling points and the plurality of second sampling points into a target neural network model to determine a plurality of vehicle acceleration values, wherein the target neural network model is used to determine vehicle acceleration values ​​through vehicle control parameter values ​​and vehicle speed values; determining a first calibration relationship based on the plurality of first sampling points, the plurality of second sampling points, and the plurality of vehicle acceleration values, wherein the first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles whose vehicle model matches the test vehicle, the vehicle motion state information including vehicle speed and vehicle acceleration.

[0013] In this embodiment of the invention, by acquiring the vehicle's first motion state information, the vehicle's first planning information, and a pre-calibrated first calibration relationship matching the vehicle's model, wherein the first calibration relationship represents the correspondence between vehicle control parameters and vehicle motion state information, the vehicle control parameters are parameters for controlling vehicle power, the first motion state information represents the vehicle's motion state at a first moment, and the first planning information represents the vehicle's expected motion state at a second moment after the first moment; based on the first motion state information and the first planning information, the vehicle's first vehicle control parameters at the first moment are determined in the first calibration relationship; after the vehicle adjusts its motion state according to the first vehicle control parameters, the vehicle's second motion state information is acquired, wherein the second motion state information represents the vehicle's expected motion state at a second moment after the first moment; The motion state is as follows: Based on the second motion state information and the first planning information, the first calibration relationship is updated to generate a second calibration relationship applicable to the current state of the vehicle. The second calibration relationship represents the correspondence between the vehicle control parameters and the motion state information in the current state. At multiple time points after the second time point, the second calibration relationship is iteratively updated using the method of obtaining the second calibration relationship to determine the target calibration relationship matching the vehicle. This achieves the purpose of comprehensively calibrating the vehicle control parameters, thereby realizing the technical effect of obtaining a vehicle control parameter calibration relationship that matches the vehicle's operating state. This solves the technical problem of mismatch between vehicle control parameters and vehicle operating state caused by the inability of traditional vehicle control parameter calibration methods to comprehensively calibrate vehicle control parameters. Attached Figure Description

[0014] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0015] Figure 1 A hardware block diagram of a computer terminal for implementing a vehicle control parameter calibration method is shown.

[0016] Figure 2 This is a flowchart illustrating a vehicle control parameter calibration method 1 provided according to an embodiment of the present invention;

[0017] Figure 3 This is a flowchart illustrating a second method for calibrating vehicle control parameters according to an embodiment of the present invention.

[0018] Figure 4 This is a schematic diagram of a vehicle control parameter calibration system provided according to an optional embodiment of the present invention;

[0019] Figure 5 This is a structural block diagram of a vehicle control parameter calibration device according to an optional embodiment of the present invention;

[0020] Figure 6 This is a structural block diagram of a vehicle control parameter calibration device II provided according to an optional embodiment of the present invention. Detailed Implementation

[0021] 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.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] According to an embodiment of the present invention, a method for calibrating vehicle control parameters is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal for implementing a vehicle control parameter calibration method is shown. Figure 1As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0025] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be implemented wholly or partially as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be wholly or partially integrated into any other element in the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as processor control (e.g., selection of a variable resistor termination path connected to an interface).

[0026] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the vehicle control parameter calibration method in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby implementing the vehicle control parameter calibration method of the aforementioned application program. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0027] The display may be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0028] Figure 2 This is a flowchart illustrating the vehicle control parameter calibration method provided in an embodiment of the present invention, as shown below. Figure 2 As shown, the method includes the following steps:

[0029] Step S202: Obtain the vehicle's first motion state information, the vehicle's first planning information, and the pre-calibrated first calibration relationship that matches the vehicle model. The first calibration relationship represents the correspondence between the vehicle control parameters and the vehicle's motion state information. The vehicle control parameters are the parameters that control the vehicle's power. The first motion state information represents the vehicle's motion state at the first moment, and the first planning information represents the vehicle's expected motion state at the second moment after the first moment.

[0030] In this step, the first calibration relationship can be a table describing the correspondence between vehicle control parameters and vehicle motion state information. Vehicle control parameters can be the throttle / brake opening, and vehicle motion state information can be the vehicle's speed and acceleration data. In the vehicle's autonomous driving mode, the system acquires the vehicle's throttle or brake control output value, wheel angle information, position information, current speed, and acceleration information within a single control cycle; this constitutes the first motion state information. The acceleration data is a matched acceleration value considering control latency. Simultaneously, the system receives the planned trajectory information from the autonomous vehicle planning module; this constitutes the first planning information.

[0031] Step S204: Based on the first motion state information and the first planning information, determine the first vehicle control parameters of the vehicle at the first moment in the first calibration relationship.

[0032] In this step, based on the vehicle's first motion state information at the current moment (i.e., the first moment) and the vehicle's planning information at the next moment (i.e., the second moment), the autonomous driving system can determine the values ​​of the vehicle's control parameters at the current moment according to the first calibration relationship.

[0033] Step S206: After the vehicle adjusts its motion state according to the first vehicle control parameters, the second motion state information of the vehicle is obtained, wherein the second motion state information represents the motion state of the vehicle at the second moment.

[0034] In this step, the vehicle operates according to the vehicle control parameters obtained in the previous step, namely the first vehicle control parameters, and at the second moment, it acquires the vehicle's motion state information at the second moment, namely the second motion state information.

[0035] Step S208: Update the first calibration relationship based on the second motion state information and the first planning information, and generate a second calibration relationship applicable to the current state of the vehicle. The second calibration relationship represents the correspondence between the vehicle control parameters and the motion state information in the current state.

[0036] In this step, during the autonomous driving operation of the vehicle, control delay is considered, and data on the vehicle's longitudinal control output, vehicle position, and vehicle status are collected. This data, combined with autonomous driving trajectory planning data, is used to calculate the control error. Then, the vehicle calibration table data is periodically updated using the online vehicle calibration update formula to adapt to vehicle control under different road conditions and loads.

[0037] Step S210: At multiple time points after the second time point, the second calibration relationship is iteratively updated using the method of obtaining the second calibration relationship to determine the target calibration relationship that matches the vehicle.

[0038] In this step, one cycle is a planning cycle. During autonomous driving, all cycles can use the method provided by this invention to update the previous calibration relationship. Updating the first calibration relationship to the second calibration relationship is one planning cycle, updating the second calibration relationship is the second planning cycle, and so on, for a total of n planning cycles, updating the calibration relationship n times: Obtain the nth motion state information of the vehicle at the current sampling time, compare it with the (n-1)th planning information at the previous sampling time, determine the error value of the (n-1)th planning information, then determine the update coefficients of the nth calibration relationship under different vehicle control parameter ranges based on the nth motion state information, and then generate the (n+1)th calibration relationship suitable for the current state of the vehicle based on the error value and update coefficients of the (n-1)th planning information. This can determine the target calibration relationship that matches the vehicle's own parameters and performance.

[0039] Through the above steps, the goal of comprehensively calibrating the vehicle control parameters is achieved, thereby realizing the technical effect of obtaining a vehicle control parameter calibration relationship that matches the vehicle's operating state. This solves the technical problem of mismatch between vehicle control parameters and vehicle operating state caused by the inability of traditional vehicle control parameter calibration methods to comprehensively calibrate vehicle control parameters.

[0040] Optionally, the data received in the current cycle can be processed, and the calibration relationship can be updated based on the data. Data processing may include determining whether the wheel angle data received in the current cycle is greater than a certain threshold. If so, the vehicle calibration table correction process for that cycle is skipped. This is because the throttle / brake calibration table does not consider the steering wheel angle condition. When calibrating to obtain this table, the vehicle's power control must be applied almost entirely to the vehicle's direction of travel to ensure the accuracy of subsequent vehicle control.

[0041] As an optional embodiment, updating the first calibration relationship based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle can be achieved through the following steps: determining the error value between the second motion state information and the first planning information; determining the correction values ​​for the motion state information and vehicle control parameters in the first calibration relationship based on the error value, the second motion state information, and the first calibration relationship; updating the first calibration relationship based on the correction values ​​to generate the second calibration relationship.

[0042] Optionally, if the calculated vehicle calibration table calibration switch flag modify_table If true, it means that the calibration table data needs to be corrected within the current control cycle. Therefore, the actual speed v of the vehicle at time t is used. t Vehicle accelerator or brake control output cmd t And construct improved two-dimensional Gaussian distribution models for throttle correction and brake correction respectively from the speed and throttle or brake control output sample data in the first calibration relationship, and then (v t cmd t The center of the two-dimensional Gaussian distribution is used as the reference point. Then, based on the received expected acceleration information 'a' at the planning time and the time-delay-calibrated vehicle acceleration information 'a'... 实际 Calculate the acceleration error Δa. Then, by traversing the grid data of the current throttle and brake calibration tables, and combining the gain coefficients of the improved two-dimensional Gaussian distribution model at each grid point, correct the calibration acceleration data at the grid points. Where Δa = a 期望 -a 实际 .

[0043] As an optional embodiment, before determining the correction value of the motion state information in the first calibration relationship based on the error value, the second motion state information, and the first calibration relationship, the following steps may be included: obtaining a predetermined error range; if the error value is within the error range, determining the correction value of the motion state information in the first calibration relationship based on the error value, the second motion state information, and the first calibration relationship; if the error value includes position error and velocity error, and one of the position error and velocity error is within the error range, while the other error is outside the error range and less than the minimum value of the error range, determining the correction value of the motion state information in the first calibration relationship based on the velocity error, the second motion state information, and the first calibration relationship.

[0044] Optionally, when updating the first calibration relationship within this cycle, data processing can be performed on the first operating status information and the second motion status information received within this cycle. When it is determined that the data received in this cycle meets the expected conditions, the vehicle calibration table correction switch flag is controlled. modify_tableIf true, the first calibration relationship is updated based on the first and second motion state information within this cycle; if the data received in this cycle does not meet the expected conditions, the vehicle calibration table correction switch flag is activated. modify_table If the value is false, the first calibration relationship will not be updated in this cycle, and the calibration relationship will be updated in the next cycle. The condition for determining whether to update the calibration relationship in this cycle can be whether the error value is within a predetermined error range. If the error value is within the predetermined error range, the vehicle calibration table calibration switch flag is activated. modify_table Set to true; when there are multiple error values, if all error values ​​are within the error range, set the vehicle calibration table calibration switch flag. modify_table If the value is true, and all error values ​​are outside the error range, then the vehicle calibration table calibration switch flag is set. modify_table If the value is false, and some error values ​​are within the error range while others are outside the error range, it is necessary to determine whether the error values ​​outside the error range are less than the minimum value of the error range. If the error values ​​outside the error range are less than the minimum value of the error range, the vehicle calibration table calibration switch flag should also be checked. modify_table If true, then the vehicle calibration table calibration switch flag is enabled; otherwise, it is disabled. modify_table It is false.

[0045] Specifically, the vehicle control position error e can be obtained by comparing and calculating the filtered real-time vehicle position and speed information with the desired position and speed of the planned trajectory. s and speed error e v Set position error e s Permissible range [es] min ,es max and speed error e v Allowable range [ev] min ,ev max ], where es min es is the minimum allowable range of position error. max es represents the maximum allowable range of position error. min es is the minimum allowable range of speed error. max This represents the maximum value within the allowable range of speed error. When e s and e v Within the allowable error range, the calibration switch flag of the calibration table... modify_table If the flag is true, calibration table data correction will be performed. Otherwise, if the correction switch is false, no correction will be performed in this control cycle, and the process will wait for the next control cycle. modify_table With e s and e v The mapping relationship is as follows:

[0046]

[0047] If e s <es min or e v <ev min This means that the corresponding position error or velocity error is considered to have met the control accuracy requirements, if e s >es max or e v >ev max The corresponding control error is considered problematic, meaning that the collected real-time position or velocity information is an anomaly and should be filtered out, and the calibration relationship should not be updated based on it.

[0048] As an optional embodiment, this can be achieved through the following steps: determining the correction values ​​for the motion state information and vehicle control parameters in the first calibration relationship based on the error value, the second motion state information, and the first calibration relationship, including: generating a two-dimensional Gaussian distribution with the numerical values ​​of the physical quantities representing the second motion state information as the distribution center based on the first calibration relationship; determining the update coefficients for the motion state information and vehicle control parameters in the first calibration relationship based on the two-dimensional Gaussian distribution; and determining the correction values ​​based on the error value and the update coefficients.

[0049] Optionally, since in the calibration relationship table, a positive value for the vehicle control parameter represents the throttle opening value, and a negative value represents the brake opening value, under the same vehicle operating state, a positive vehicle control parameter indicates throttle control, and a negative vehicle control parameter indicates brake control, the update coefficients corresponding to the vehicle control parameters when the vehicle control parameters are positive or negative are calculated separately. This means setting the template size and data interval for the throttle correction Gaussian distribution model and the brake correction Gaussian distribution model respectively. The vehicle throttle or brake opening data and the corresponding current vehicle speed data in the first calibration relationship are mapped to the variable ranges of the two Gaussian distribution models, and the data mean μ is calculated. thr μ brk and standard deviation σ thr σ brk , where μ thr To correct the mean of the Gaussian distribution model for throttle adjustment, μ brk To correct the mean of the Gaussian distribution model for braking, σ thr To correct the standard deviation of the Gaussian distribution model for throttle adjustment, σ brk To correct the standard deviation of the Gaussian distribution model for braking. For example, the throttle opening range is [thr min ,thr max ], where thr min thr is the minimum throttle opening. maxThe maximum throttle opening is given by thr. gap The speed variation range is [v min ,v max ], where v min For the minimum velocity, v max The maximum value of the velocity is given by the velocity sampling interval v. gap The throttle-corrected Gaussian distribution template variable X has a range of [-M, M] and an interval of X. gap The variable Y has a range of [-N, N] and an interval of Y. gap The mapping formulas from throttle opening to variable X and from speed to variable Y are as follows:

[0050]

[0051] Among them, thr i and v i These are sampled values ​​for throttle opening and speed, respectively, X. i and Y i These are the mapping values ​​for throttle opening and speed, respectively, with [] representing the rounding symbol. The data mapping for brake correction follows the same principle.

[0052] Secondly, the vehicle's current actual speed and the vehicle's throttle or brake control output data (v) collected at time t will be used. t cmd t Mapped to corresponding template variable data in, It is a template variable corresponding to the vehicle's current actual speed. It is the template variable corresponding to the vehicle throttle or brake control output data, and the vehicle control parameter output cmd t The formula is as follows:

[0053]

[0054] Where throttle is the throttle opening value of the vehicle at time t, and brake is the brake opening value or brake pressure value of the vehicle at time t. Therefore, when cmd t When the value is ≥0, it is mapped to data within the range of the throttle-corrected Gaussian distribution model; otherwise, it is mapped to data within the range of the brake-corrected Gaussian distribution model.

[0055] Secondly, an improved two-dimensional Gaussian probability density model is used as the formula for calculating the gain coefficient of the current throttle / brake calibration table, thereby constructing an update function for data correction. The update formula is as follows:

[0056] T new [v i [cmd] j ] = T[v i [cmd] j]+f(v i cmd j )*△a,

[0057] Where T[v i [cmd] j The calibration table at time t is in (v t cmd t The original value on the grid, T new [v i [cmd] j [This represents the corrected new value.] f(v) i cmd j Based on the improved throttle and brake Gaussian distribution template determined at time t, the calibration table grid points (v) are traversed. i cmd j The gain coefficient at time (v) i cmd j The values ​​in the calibration table are actual values ​​and need to be converted into mapping values ​​corresponding to the Gaussian template. Again Then the gain coefficient f(v) i cmd j ) is equivalent to The formula for the gain coefficient is as follows:

[0058]

[0059] Where μ is the sample mean matrix of the variables involved in X, and is the center of the improved two-dimensional Gaussian distribution mentioned above. Here, it is taken as the mapping value of the vehicle state data collected at time t. As the central value, that is When cmd t When ≥0, ∑ represents the throttle correction update coefficient; otherwise, it represents the brake correction update coefficient. ∑ is the covariance matrix of the involved variables calculated based on the first calibration relationship; for two-dimensional variables x1 and x2, it is:

[0060]

[0061] Matrix element σ(x) m ,x k The calculation formula is as follows:

[0062]

[0063] Where x mi Represents a random variable x m The i-th observation sample, x ki Represents a random variable x kLet n represent the sample size, where n is the i-th observation sample. It's easy to see that the formula reaches its maximum value of 1 at the mean center, at which point the acceleration error Δa at time t is entirely used for (v... t cmd t Data correction on the grid. By traversing all data grid points for throttle or brake in this way, the data update of the throttle or brake calibration table can be achieved within this control cycle.

[0064] Figure 3 This is a flowchart illustrating the vehicle control parameter calibration method provided in an embodiment of the present invention, as shown below. Figure 3 As shown, the method includes the following steps:

[0065] Step S302: Obtain the vehicle control parameter range and vehicle speed range of the test vehicle.

[0066] In this step, the range of changes in the vehicle's control parameters and the range of changes in the vehicle's speed can be obtained.

[0067] Step S304: Select multiple first sampling points within the range of vehicle control parameters and multiple second sampling points within the range of vehicle speed.

[0068] In this step, data can be sampled at equal intervals within the range of vehicle control parameter variation to obtain multiple first sampling points, and data can be sampled at equal intervals within the range of vehicle speed variation to obtain multiple second sampling points.

[0069] Step S306: Input multiple first sampling points and multiple second sampling points into the target neural network model to determine multiple vehicle acceleration values. The target neural network model is used to determine the vehicle acceleration values ​​through vehicle control parameter values ​​and vehicle speed values.

[0070] In this step, based on the sampling points obtained within the range of variation of vehicle control parameters and vehicle speed, multiple first sampling points and multiple second sampling points are input into the target neural network model. The target neural network model will then output the corresponding vehicle acceleration value. It should be noted that the input to the target neural network model is a set of data: a first sampling point representing a vehicle control parameter value and a second sampling point representing a vehicle speed value. The output is an acceleration value corresponding to the first and second sampling points.

[0071] Step S308: Determine a first calibration relationship based on multiple first sampling points, multiple second sampling points, and multiple vehicle acceleration values. The first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles that match the vehicle model and the test vehicle. The vehicle motion state information includes vehicle speed and vehicle acceleration.

[0072] In this step, a first calibration relationship can be determined based on multiple first sampling points, multiple second sampling points, and multiple vehicle acceleration values ​​output by the target neural network model. It should be noted that the first calibration relationship can only accurately characterize the correspondence between the test vehicle's control parameters and vehicle motion state information. However, when a more common testing scenario is chosen for the test vehicle, the resulting first calibration relationship can be widely applied to vehicles matching the test vehicle model for vehicle parameter control. When the first calibration relationship is sent to other vehicles of the same model as the test vehicle, these vehicles can adjust and update the first calibration relationship based on their own driving data, making the updated calibration relationship more accurately characterize the correspondence between their own vehicle control parameters and vehicle motion state information.

[0073] In this embodiment, vehicle control parameters, vehicle speed, and acceleration information are collected in manual driving mode. After data preprocessing, a neural network deep learning model is used to train the processed data. Then, the model is used to predict the data to form discrete data of vehicle control parameters and vehicle motion state information. After data integration, an initial calibration table of vehicle control parameters, i.e., the first calibration relationship, is finally formed.

[0074] As an optional embodiment, the target neural network model is a model trained through the following steps: obtaining multiple sets of correspondences for the test vehicle, wherein each set of correspondences represents the correspondence between vehicle control parameter values, vehicle speed values, and vehicle acceleration values; training multiple different types of original neural network models according to the multiple sets of correspondences to obtain multiple trained neural network models; and selecting the target neural network model from the multiple trained neural network models.

[0075] As an optional embodiment, obtaining multiple sets of correspondences for the test vehicle can be achieved through the following steps: obtaining multiple sets of test data for the test vehicle, and multiple wheel angle values ​​that correspond one-to-one with the multiple sets of test data, wherein each set of test data includes the correspondence between vehicle test control parameter values, vehicle test speed values, and vehicle test acceleration values; determining that the angle values ​​within a predetermined wheel angle range among the multiple wheel angle values ​​are multiple target angle values; and determining that the multiple sets of test data corresponding to the multiple target angle values ​​are multiple sets of correspondences.

[0076] Optionally, real-world calibration tests can be performed on the test vehicle to obtain multiple sets of test data. Each set of test data includes the correspondence between vehicle test control parameter values, vehicle test speed values, and vehicle test acceleration values. Multiple wheel angle values ​​corresponding to each set of test data are also obtained. The test data corresponding to the wheel angle values ​​within the wheel angle range are selected as multiple correspondences. The original neural network model is then trained based on these correspondences to obtain the target neural network model. This is because the throttle / brake calibration table does not consider steering wheel angle conditions. When calibrating this table, the vehicle's power control must be applied approximately entirely to the vehicle's direction of travel to ensure the accuracy of subsequent vehicle control.

[0077] Specifically, firstly, the device of the present invention is installed on the vehicle to be calibrated, and the device's data is calibrated so that the positioning signal output by the device can represent the vehicle's positioning status information; secondly, manual driving and data collection are performed. On a straight road, the vehicle is manually driven to accelerate from a standstill to a certain state in a straight line, and then the vehicle is controlled to decelerate to a stop, repeating this action multiple times. During this process, the throttle / brake opening is controlled to increase or decrease slowly, or to maintain a certain throttle / brake opening for a period of time. Real-time feedback information from the vehicle, including wheel angle information, throttle or brake opening information, current vehicle speed information, and acceleration feedback information, is received. The vehicle speed information and acceleration information can also be obtained through the combined positioning module of the present invention. The acceleration information needs to take into account the vehicle's actuation delay, that is, the time delay from the issuance of the control command to the vehicle generating a continuous and stable torque output, so as to ensure the time matching of the control parameter output, vehicle speed information, and acceleration information; secondly, the received throttle or brake opening information, current vehicle speed information, and acceleration information are preprocessed.

[0078] The preprocessing process includes: process data division, which divides the data into two groups based on whether the acceleration or braking action was performed at a certain moment, namely acceleration process data and deceleration process data; angle filtering, which filters out vehicle state data with excessively large angles at corresponding moments based on the wheel angle information fed back at different moments, namely throttle or brake opening data, vehicle speed data, and acceleration data; sliding time window data filtering, which uses a certain time span as a window and performs mean filtering on the data within the window using a sliding window method. The filtered data includes throttle or brake opening information, vehicle speed information, and acceleration information from both groups of data; and outlier filtering, which calculates the mean and standard deviation of the collected throttle or brake opening data, vehicle speed data, and acceleration data, and uses the 3σ criterion to filter out outliers outside the range (μ-σ,μ+σ), where μ is the data mean and σ is the data standard deviation.

[0079] Secondly, multiple neural network regression models with different algorithmic support can be established. The received data can be substituted into each regression model to train the acceleration and deceleration process data respectively. Then, through horizontal evaluation and vertical screening, the regression model with the best performance can be obtained. The optimal regression model can be used to predict the data to obtain the acceleration calibration model and the deceleration calibration model.

[0080] Finally, the throttle or brake opening data, vehicle speed data, and acceleration data in the two preprocessed data sets can be divided into training set and test set according to a certain ratio. The training set is used as the input and output of the model for model training, and the test set is used to verify the accuracy of the model prediction.

[0081] As an optional embodiment, this can be achieved through the following steps: selecting the target neural network model from multiple trained neural network models, including: calculating the mean squared error and coefficient of determination for each of the multiple neural network models; and determining the neural network model with the smallest mean squared error and the largest coefficient of determination as the target neural network model among the multiple neural network models.

[0082] Optionally, during model training, the throttle or brake opening data and vehicle speed data of the training set are used as model inputs, and acceleration data is used as model outputs. The network training parameters of each neural network regression model are set, such as the number of hidden layer neurons, the maximum number of model iterations, the expected accuracy, and the learning rate. Then, the two sets of model data are trained separately.

[0083] The two sets of test set input data are substituted into their respective trained models to obtain the model's predicted output values. These predicted output values ​​are then compared with the test set output data. The mean squared error (MSE) and coefficient of determination (R²) are used to evaluate the generalization ability of the neural network model. The MSE and R² are... 2 The calculation formula is as follows:

[0084]

[0085]

[0086] in Let y be the predicted value of the i-th sample. i (i = 1, 2, ..., n) represents the true value of the i-th sample, and n is the number of samples in the test set. For training the model, the smaller the mean squared error, the better the model is trained. The coefficient of determination ranges from [0, 1]. The closer the value is to 1, the better the model performance; the closer it is to 0, the worse the performance.

[0087] The optimal neural network regression model is selected through a comprehensive evaluation of both horizontal and vertical models. Vertically, the same model is trained repeatedly using optimization methods to adjust its training parameters and quantify its performance. Horizontally, the performance of different algorithm models is compared under the same number of iterations. Finally, through a comprehensive evaluation of different algorithm models, the regression model with the best performance is obtained, i.e., the mean squared error and coefficient of determination both meet the threshold conditions, and the model exhibits the best overall performance.

[0088] The two sets of high-performing neural network models selected above were used as acceleration and deceleration calibration models. The ranges for vehicle throttle or brake opening data and vehicle speed were defined. Data was then sampled at equal intervals within these ranges. The sampled throttle or brake opening data and vehicle speed data were used as inputs for model prediction to obtain the corresponding acceleration and deceleration data. This resulted in discretized initial calibration tables for vehicle throttle acceleration and braking deceleration.

[0089] The aforementioned discrete throttle and brake initial calibration tables are merged into a single table to facilitate subsequent vehicle control parameter queries and calibration table updates. The merging method involves mapping the throttle opening values ​​in the throttle calibration table to positive values ​​(the magnitude remains unchanged, but the sign is positive), while the magnitude and sign of other data in the table remain unchanged. Similarly, the brake opening values ​​in the brake calibration table are mapped to negative values ​​(the magnitude remains unchanged, but the sign is negative), while the magnitude and sign of other data in the table also remain unchanged. This merges the data from the two calibration tables into a single, non-overlapping throttle / brake initial calibration table.

[0090] As an optional embodiment, multiple first sampling points and multiple second sampling points are input into a target neural network model to determine multiple vehicle acceleration values, including: inputting multiple first sampling points and multiple second sampling points into a first neural network model, the first neural network model outputting a first acceleration value, wherein the target neural network model includes a first neural network model, which is trained based on throttle opening data and vehicle motion state information measured during the acceleration of the test vehicle, and the vehicle control parameters include throttle opening data; inputting multiple first sampling points and multiple second sampling points into a second neural network model, the second neural network model outputting a second acceleration value, wherein the target neural network model includes a second neural network model, which is trained based on brake opening data and vehicle motion state information measured during the deceleration of the test vehicle, and the vehicle control parameters include brake opening data.

[0091] As an optional embodiment, determining a first calibration relationship based on multiple first sampling points, multiple second sampling points, and multiple vehicle acceleration values ​​includes: generating a throttle calibration relationship for the test vehicle based on the multiple first sampling points, multiple second sampling points, and a first acceleration value, wherein the first acceleration value is a non-negative value among the multiple vehicle acceleration values, the throttle calibration relationship characterizes the correspondence between throttle opening data and vehicle motion state information, and the values ​​of the multiple second sampling points characterizing the throttle opening data in the throttle calibration relationship are non-negative; generating a brake calibration relationship for the test vehicle based on the multiple first sampling points, multiple second sampling points, and a second acceleration value, wherein the second acceleration value is a negative value among the multiple vehicle acceleration values, the brake calibration relationship characterizes the correspondence between brake opening data and vehicle motion state information, and the values ​​of the multiple second sampling points characterizing the brake opening data in the brake calibration relationship are negative; and merging the throttle calibration relationship and the brake calibration relationship to obtain the first calibration relationship.

[0092] Optionally, during the generation of the first calibration relationship, after obtaining multiple sets of correspondences through field testing, the processes representing vehicle acceleration and deceleration in the multiple sets of correspondences can be processed separately. Since the vehicle is controlled by the brake and accelerator respectively during acceleration and deceleration, two neural network models can be used to predict the acceleration data during the vehicle acceleration and deceleration processes respectively. Based on the output of the neural network models, throttle calibration relationship and brake calibration relationship are generated respectively. Then, the values ​​of the vehicle control parameters in the brake calibration relationship are adjusted to negative values, and the throttle calibration relationship and brake calibration relationship can be combined into the first calibration relationship.

[0093] Specifically, when training the neural network model, the target neural network model can include a first neural network model and a second neural network model. The first neural network model is trained using data from the acceleration process of the test vehicle, and the second neural network model is trained using data from the deceleration process of the test vehicle. Therefore, after obtaining multiple first sampling points and multiple second sampling points, these points can be input into the first neural network model. The first acceleration value output by the first neural network model is the possible acceleration value of the test vehicle when the vehicle speed is the speed value represented by the first sampling point, and the throttle is under the control parameters represented by the second sampling point. The first acceleration value is always non-negative. Similarly, the second acceleration value output by the second neural network model is the possible acceleration value of the test vehicle when the vehicle speed is the speed value represented by the first sampling point, and the brake is under the control parameters represented by the second sampling point. The second acceleration value is always negative. Therefore, based on the multiple first sampling points, multiple second sampling points, and the first acceleration value, the throttle calibration relationship can be obtained, and based on the multiple first sampling points, multiple second sampling points, and the second acceleration value, the brake calibration relationship can be obtained. Since the values ​​of vehicle control parameters in the throttle calibration relationship and the brake calibration relationship are the same, the throttle calibration relationship and the brake calibration relationship cannot be directly merged. It is necessary to map the vehicle control parameters, i.e., the throttle opening value, in the throttle calibration relationship to positive values ​​and the vehicle control parameters, i.e., the brake opening value, in the brake calibration relationship to negative values. This will distinguish the throttle calibration relationship and the brake calibration relationship. Then, the modified throttle calibration relationship and the brake calibration relationship can be merged to obtain the first calibration relationship.

[0094] As a specific embodiment, the method provided by the present invention includes the following steps:

[0095] Step 1: Collect throttle / brake opening information, vehicle speed information, and acceleration information in manual driving mode. After data preprocessing, use a neural network deep learning model to learn and train the processed data. Then, use the model to predict the data and form discretized data of vehicle throttle / brake opening, speed, and acceleration. Finally, after data integration, form the initial calibration table of vehicle throttle / brake.

[0096] Step one specifically includes the following:

[0097] Step 1.1: Install and calibrate the device of the present invention. Install the device of the present invention on the vehicle to be calibrated and perform data calibration of the device to ensure that the output position, speed and acceleration information of the device are consistent with the actual state of the vehicle.

[0098] Step 1.2: On a straight road, manually drive the vehicle for a period of time, approximately 10 to 20 minutes, controlling the vehicle to accelerate from a standstill to a certain state in a straight line, and then controlling the vehicle to decelerate to a stop, repeating this action multiple times. During this period, slowly increase or decrease the throttle / brake opening, or keep a certain throttle / brake opening constant for a period of time to ensure smooth control. During the above vehicle acceleration and deceleration process, receive feedback information from the vehicle, including wheel angle information, throttle or brake opening information, current vehicle speed information, and acceleration feedback information. The position, speed, and acceleration information can be selected from vehicle status feedback information, depending on the data accuracy and update frequency requirements, or obtained through the combined positioning module of the device of this invention. The acceleration information needs to consider the vehicle's action delay. Assuming the vehicle's action delay is Δt, and the vehicle's speed at time t is collected as v. t The acceleration is a t Then, based on the currently collected data and historical data, data interpolation is performed to obtain the actual speed and acceleration of the vehicle at time t-Δt as v. t-△t ,a t .

[0099] Step 1.3: Preprocess the received throttle or brake opening information, current vehicle speed information, and acceleration information.

[0100] Step 1.3.1: Process data division. Based on the vehicle control output at a certain moment, the collected data is divided into throttle acceleration process data and braking deceleration process data. For example, if the vehicle control output at time t is cmd... t If cmd t If the value is ≥0, all collected data will be classified as acceleration data; otherwise, they will be classified as deceleration data.

[0101] Step 1.3.2: Angle filtering, based on the wheel angle information δ fed back at time t. t If the wheel angle is greater than a certain threshold δ min_offset , i.e. δ t >δ min_offset If the longitudinal offset of the vehicle is too large at this moment, the data collected at this moment will be filtered out and not used as sample data for the initial longitudinal calibration.

[0102] Step 1.3.3: Sliding Time Window Data Filtering. Using a certain time span as the window, a sliding window method is employed to perform mean filtering on the data within the window. The filtered data includes throttle or brake opening information, vehicle speed information, and acceleration information from both sets of data. Assume that a segment of data x = {x1, x2, ..., x...} has been collected and sorted by time after grouping and corner filtering. n The data length is n, and the time window span is t. window If data x k and data x mThe corresponding times and satisfy and And if m,k∈{1,2,…,n}, then we consider {x} m ,x m+1 ,…,x k-1 ,x k This data segment represents the data within the time window. The following formula is used to perform mean filtering on the data:

[0103]

[0104] Step 1.3.4: Outlier Filtering. For the collected throttle or brake opening data, vehicle speed data, and acceleration data, calculate their respective mean μ and standard deviation σ. Using the 3σ criterion, filter outliers outside the range [μ-σ, μ+σ], i.e., when |x i When -μ|>σ, filter out the abnormal data x. i .

[0105] Step 1.4: Establish multiple neural network regression models with different algorithmic support, such as back propagation (BP), radial basis function (RBF) neural networks with multivariate interpolation, support vector machine for regression (SVR), extreme learning machine (ELM), etc. Substitute the received data into each regression model to train the acceleration and deceleration processes respectively. Then, through horizontal evaluation and vertical selection, obtain the regression model with the best performance. Use the optimal regression model to perform data prediction to obtain the acceleration calibration model and deceleration calibration model.

[0106] Step 1.4.1: Divide the throttle or brake opening data, vehicle speed data and acceleration data of the two preprocessed data sets into training set and test set in a 9:1 ratio. The training set is used as the input and output of the model for model training, and the test set is used to verify the accuracy of the model prediction.

[0107] Step 1.4.2: During model training, use the throttle or brake opening data and vehicle speed data from the training set as model inputs and acceleration data as model outputs. Set the network training parameters for each neural network regression model, such as the number of hidden layer neurons, the maximum number of model iterations, the expected accuracy, and the learning rate. Then, train the two sets of model data separately.

[0108] Step 1.4.3: Substitute the input data from the two test sets into their respective trained models to obtain the predicted output values. Compare these predicted output values ​​with the output data from the test sets, and use the mean squared error (MSE) and coefficient of determination (R²) to evaluate the generalization ability of the neural network model. The mean squared error (E) and coefficient of determination (R²) are used to evaluate this generalization ability. 2 The calculation formula is as follows:

[0109]

[0110]

[0111] in Let y be the predicted value of the i-th sample. i (i = 1, 2, ..., n) represents the true value of the i-th sample, and n is the number of samples in the test set.

[0112] Step 1.4.4: Select the optimal neural network regression model through comprehensive evaluation of the model both horizontally and vertically. Vertically, for the same model, adjust the model training parameters using certain optimization methods, perform repeated training a certain number of times, and screen out the regression models that meet the threshold condition, i.e., E≤E. tolerance And R≥R tolerance , where R tolerance E is the threshold for the coefficient of determination R. tolerance The threshold value for the mean square error E is E, E,E tolerance ≥0,R,R tolerance ∈[0,1]; horizontally, the performance of different algorithm models is compared under the same number of iterations. Finally, through a comprehensive evaluation of different algorithm models both horizontally and vertically, the regression model with the best performance is obtained.

[0113] Step 1.4.5: Using the two sets of high-performing neural network models selected above as acceleration and deceleration calibration models, the range of vehicle throttle or brake opening data variation and the range of vehicle speed variation are set. Then, the data is sampled at equal intervals within the set range. The sampled vehicle throttle or brake opening data and vehicle speed data are used as inputs for model prediction to obtain the corresponding acceleration and deceleration acceleration data. Finally, a discretized initial calibration table for vehicle throttle acceleration and initial calibration table for brake deceleration are formed.

[0114] Step 1.4.6: Merge the discretized throttle initial calibration table and brake initial calibration table into a single table to facilitate subsequent vehicle control parameter queries and calibration table updates. The merging method involves mapping the throttle opening value in the throttle calibration table to a positive value (the numerical value remains unchanged, but the sign is positive), while the values ​​and signs of other data in the table remain unchanged. Similarly, mapping the brake opening value in the brake calibration table to a negative value (the numerical value remains unchanged, but the sign is negative), while the values ​​and signs of other data in the table also remain unchanged. This merges the two calibration tables into a single, non-overlapping throttle / brake initial calibration table.

[0115] Step Two: Based on the existing initial calibration table for vehicle throttle / brake, operate the vehicle in autonomous driving mode. During autonomous driving operation, considering control delay, collect data on vehicle longitudinal control output, vehicle position, and vehicle status, and calculate control error by combining this data with autonomous driving trajectory planning. Then, using the online vehicle calibration update formula, periodically update the vehicle calibration table data to adapt to vehicle control under different road conditions and loads.

[0116] Step two specifically includes the following:

[0117] Step 2.1: In the vehicle's autonomous driving mode, acquire the vehicle's throttle or brake control output values, wheel angle information, position information, current vehicle speed, and acceleration information within a single control cycle. The acceleration data is the matched acceleration value after considering control latency. Simultaneously, receive the planned trajectory information from the autonomous vehicle planning module.

[0118] Step 2.2: Perform corner filtering on the data received in Step 2.1 according to the method described in Step 1.3.2, that is, determine whether the wheel corner data received in the current cycle is greater than a certain threshold. If so, skip the vehicle calibration table correction process for that cycle.

[0119] Step 2.3: The vehicle control position error e is obtained by comparing the filtered real-time vehicle position and speed information with the desired position and speed of the planned trajectory. s and speed error e v Set position error e s Permissible range [es] min ,es max and speed error e v Allowable range [ev] min ,ev max ]. When e s and e v Within the allowable error range, the calibration switch flag of the calibration table... modify_table If the flag is true, calibration table data correction will be performed. Otherwise, if the correction switch is false, no correction will be performed during this control cycle, and the system will wait for further processing. modify_table With e s and e v The mapping relationship is as follows:

[0120]

[0121] Step 2.4: If the vehicle calibration table flag calculated in Step 2.3 is correct... modify_tableIf true, it means that the calibration table data needs to be corrected within the current control cycle. Therefore, the actual speed v of the vehicle at time t is used. t Vehicle accelerator or brake control output cmd t And the improved two-dimensional Gaussian distribution models for throttle correction and brake correction are constructed from the sample data processed in step 1.3, respectively, and (v t cmd t The center of the two-dimensional Gaussian distribution is used as the reference point. Then, based on the received expected acceleration information a at the planning time... 期望 and time-delay calibrated vehicle acceleration information a 实际 Calculate the acceleration error Δa. Then, by traversing the grid data of the current throttle calibration table and brake calibration table, and combining the gain coefficient of the improved two-dimensional Gaussian distribution model at each grid point, the calibration acceleration data at the grid point is corrected.

[0122] Step 2.4.1: Based on the data variation range and sampling interval of throttle opening, brake opening, and current vehicle speed set in Step 1.4.4, set the template size and data interval for the throttle-corrected Gaussian distribution model and the brake-corrected Gaussian distribution model respectively, keeping the number of sampling points consistent. Map the vehicle throttle or brake opening data processed in Step 1.3 and their corresponding current vehicle speed data to the variable range of the two Gaussian distribution models respectively, and calculate the data mean μ. thr μ brk and standard deviation σ thr σ brk For example, the range of throttle opening variation is [thr min ,thr max The sampling interval is thr gap The speed variation range is [v min ,v max The sampling interval is v. gap The throttle-corrected Gaussian distribution template variable X has a range of [-M, M] and an interval of X. gap The variable Y has a range of [-N, N] and an interval of Y. gap The mapping formulas from throttle opening to variable X and from speed to variable Y are as follows:

[0123]

[0124] Among them, thr i and v i These are sampled values ​​for throttle opening and speed, respectively, X. i and Y i These are the mapping values ​​for throttle opening and speed, respectively, with [] representing the rounding symbol. The data mapping for brake correction follows the same principle.

[0125] Step 2.4.2: Collect the vehicle's current actual speed and the vehicle's throttle or brake control output data (v) at time t. t cmd t Mapped to corresponding template data Vehicle control parameter output cmd t The formula is as follows:

[0126]

[0127] Where throttle is the throttle opening value of the vehicle at time t, and brake is the brake opening value or brake pressure value of the vehicle at time t. Therefore, when cmd t When the value is greater than 0, it is mapped to data within the range of the throttle-corrected Gaussian distribution model; otherwise, it is mapped to data within the range of the brake-corrected Gaussian distribution model.

[0128] Step 2.4.3: Utilize the improved two-dimensional Gaussian probability density model as the gain coefficient calculation formula for the current throttle / brake calibration table, thereby constructing the data correction update function. The update formula is as follows:

[0129] T new [v i [cmd] j ] = T[v i [cmd] j ]+f(v i cmd j )*△a,

[0130] Where T[v i [cmd] j The calibration table at time t is in (v t cmd t The original value on the grid, T new [v i [cmd] j [This represents the corrected new value.] f(v) i cmd j Based on the improved throttle and brake Gaussian distribution template determined at time t, the calibration table grid points (v) are traversed. i cmd j The gain coefficient at time (v) i cmd j The values ​​in the calibration table are actual values ​​and need to be converted into mapping values ​​corresponding to the Gaussian template. Again Then the gain coefficient f(v) i cmd j ) is equivalent to The formula for the gain coefficient is as follows:

[0131]

[0132] Where μ is the sample mean matrix of the variables involved in X, and is the center of the improved two-dimensional Gaussian distribution mentioned above. Here, it is taken as the mapping value of the vehicle state data collected at time t in step 2.4.2. As the central value, that is When cmd t When >0, ∑ represents the throttle correction update coefficients; otherwise, it represents the brake correction update coefficients. ∑ is the covariance matrix of the involved variables. For two-dimensional variables x1 and x2, the covariance matrix is:

[0133]

[0134] Matrix element σ(x) m ,x k The calculation formula is as follows:

[0135]

[0136] Where x mi Represents a random variable x m The i-th observation sample, x ki Similarly, n represents the sample size. It is easy to see that formula (8) reaches its maximum value of 1 at the mean center, at which point the acceleration error Δa at time t is entirely used for (v t cmd t Data correction on the grid. By traversing all data grid points for throttle or brake in this way, the data update of the throttle or brake calibration table can be achieved within this control cycle.

[0137] Figure 4 This is a schematic diagram of a vehicle control parameter calibration system provided according to an optional embodiment of the present invention, such as... Figure 4 As shown, the present invention also provides an online adaptive calibration system for vehicle control parameters, comprising:

[0138] The data receiving module is used to acquire in real time vehicle position, speed, acceleration, wheel angle, throttle opening, brake opening or brake pressure values, and expected position, expected speed, and expected acceleration information related to vehicle throttle / brake calibration. This information is then used for subsequent offline data learning and real-time online data updates.

[0139] If the combined positioning module cannot obtain vehicle status data through the data receiving module, or if the received vehicle position, speed and acceleration information, wheel angle information, throttle opening information and brake opening information cannot meet the real-time requirements for online data updates, then the GPS and IMU combined positioning module can obtain vehicle position information, speed information and acceleration information with higher accuracy and faster update frequency related to the vehicle status.

[0140] The offline learning module uses vehicle speed and acceleration data and throttle / brake control outputs collected under manual driving mode. After data preprocessing, a deep learning neural network is used to train a portion of the collected data samples, and then the performance of the trained model is evaluated using the remaining data samples. This iterative process yields a high-performance throttle / brake training model, which is finally used to generate an initial throttle / brake calibration table.

[0141] The online update module, in autonomous driving mode, collects vehicle status data and control outputs in real time within a single control cycle, and calculates the control error by combining this with the planned trajectory information within that control cycle. Then, it uses an improved two-dimensional Gaussian probability density model to update the current throttle / brake calibration table data in real time.

[0142] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0143] Through the above description of the embodiments, those skilled in the art can clearly understand that the vehicle control parameter calibration method according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0144] According to embodiments of the present invention, an apparatus for implementing the above-described vehicle control parameter calibration method is also provided. Figure 5 This is a structural block diagram of a vehicle control parameter calibration device according to an embodiment of the present invention, as shown below. Figure 5 As shown, the device includes: a first acquisition module 52, a first determination module 54, a second acquisition module 56, and a generation module 58. The device will be described below.

[0145] The first acquisition module 52 is used to acquire the vehicle's first motion state information, the vehicle's first planning information, and a pre-calibrated first calibration relationship that matches the vehicle's model. The first calibration relationship represents the correspondence between the vehicle control parameters and the vehicle's motion state information. The vehicle control parameters are parameters for controlling the vehicle's power. The first motion state information represents the vehicle's motion state at the first moment, and the first planning information represents the vehicle's expected motion state at the second moment after the first moment.

[0146] The first determining module 54, connected to the first acquiring module 52, is used to determine the first vehicle control parameters of the vehicle at the first moment in the first calibration relationship based on the first motion state information and the first planning information.

[0147] The second acquisition module 56, connected to the first determination module 54, is used to acquire the second motion state information of the vehicle after the vehicle adjusts its motion state according to the first vehicle control parameters, wherein the second motion state information represents the motion state of the vehicle at the second moment.

[0148] The generation module 58, connected to the second acquisition module 56, is used to update the first calibration relationship based on the second motion state information and the first planning information, and generate a second calibration relationship applicable to the current state of the vehicle. The second calibration relationship represents the correspondence between the vehicle control parameters and the motion state information in the current state.

[0149] The iteration module 60, connected to the generation module 58, is used to iteratively update the second calibration relationship at multiple time points after the second time point by using the method of obtaining the second calibration relationship, so as to determine the target calibration relationship that matches the vehicle.

[0150] It should be noted that the first acquisition module 52, the first determination module 54, the second acquisition module 56, the generation module 58, and the iteration module 60 mentioned above correspond to steps S202 to S210 in the embodiments. Multiple modules implement the same instances and application scenarios as their corresponding steps, but are not limited to the content disclosed in the above embodiments. It should also be noted that the above modules, as part of the device, can run on the computer terminal 10 provided in the embodiments.

[0151] According to embodiments of the present invention, an apparatus for implementing the above-described vehicle control parameter calibration method is also provided. Figure 6 This is a structural block diagram of a vehicle control parameter calibration device two provided according to an embodiment of the present invention, as shown below. Figure 6 As shown, the device includes: a third acquisition module 62, a sampling module 64, a second determination module 66, and a third determination module 68. The device will be described below.

[0152] The third acquisition module 62 is used to acquire the range of vehicle control parameters and the range of vehicle speed of the test vehicle.

[0153] The sampling module 64, connected to the third acquisition module 62, is used to select multiple first sampling points within the range of vehicle control parameters and multiple second sampling points within the range of vehicle speed.

[0154] The second determining module 66, connected to the sampling module 64, is used to input multiple first sampling points and multiple second sampling points into the target neural network model to determine multiple vehicle acceleration values. The target neural network model is used to determine the vehicle acceleration values ​​through vehicle control parameter values ​​and vehicle speed values.

[0155] The third determining module 68, connected to the second determining module 66, is used to determine a first calibration relationship based on multiple first sampling points, multiple second sampling points and multiple vehicle acceleration values. The first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles with the same vehicle model as the test vehicle. The vehicle motion state information includes vehicle speed and vehicle acceleration.

[0156] It should be noted that the third acquisition module 62, sampling module 64, second determination module 66, and third determination module 68 mentioned above correspond to steps S302 to S308 in the embodiments. Multiple modules and their corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in the above embodiments. It should also be noted that the above modules, as part of the device, can run on the computer terminal 10 provided in the embodiments.

[0157] Embodiments of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0158] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the vehicle control parameter calibration method and device in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned vehicle control parameter calibration method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to a computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0159] The processor can invoke information and application programs stored in the memory via a transmission device to perform the following steps: acquiring the vehicle's first motion state information, the vehicle's first planning information, and a pre-calibrated first calibration relationship matching the vehicle's model, wherein the first calibration relationship represents the correspondence between vehicle control parameters and vehicle motion state information, the vehicle control parameters being parameters for controlling vehicle power, the first motion state information representing the vehicle's motion state at a first moment, and the first planning information representing the vehicle's expected motion state at a second moment after the first moment; determining the vehicle's first vehicle control parameters at the first moment in the first calibration relationship based on the first motion state information and the first planning information; acquiring the vehicle's second motion state information after the vehicle adjusts its motion state according to the first vehicle control parameters, wherein the second motion state information represents the vehicle's motion state at the second moment; updating the first calibration relationship based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the vehicle's current state, wherein the second calibration relationship represents the correspondence between the vehicle control parameters and motion state information in the current state.

[0160] Optionally, the processor may also execute program code for the following steps: obtaining the vehicle control parameter range and vehicle speed range of the test vehicle; selecting multiple first sampling points within the vehicle control parameter range and multiple second sampling points within the vehicle speed range; inputting the multiple first sampling points and multiple second sampling points into a target neural network model to determine multiple vehicle acceleration values, wherein the target neural network model is used to determine the vehicle acceleration values ​​through the vehicle control parameter values ​​and vehicle speed values; determining a first calibration relationship based on the multiple first sampling points, multiple second sampling points, and multiple vehicle acceleration values, wherein the first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles with the same vehicle model as the test vehicle, and the vehicle motion state information includes vehicle speed and vehicle acceleration.

[0161] This invention provides a vehicle parameter calibration method. It involves acquiring first motion state information of the vehicle, first planning information of the vehicle, and a pre-calibrated first calibration relationship matching the vehicle model. The first calibration relationship represents the correspondence between vehicle control parameters and the vehicle's motion state information. The vehicle control parameters are parameters for controlling the vehicle's power. The first motion state information represents the vehicle's motion state at a first moment, and the first planning information represents the vehicle's expected motion state at a second moment after the first moment. Based on the first motion state information and the first planning information, first vehicle control parameters for the vehicle at the first moment are determined in the first calibration relationship. The method then adjusts the vehicle's motion state according to the first vehicle control parameters. Then, the second motion state information of the vehicle is obtained, wherein the second motion state information represents the motion state of the vehicle at the second moment; the first calibration relationship is updated according to the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle, wherein the second calibration relationship represents the correspondence between the vehicle control parameters and the motion state information in the current state, thereby achieving the purpose of comprehensively calibrating the vehicle control parameters, thus realizing the technical effect of obtaining a vehicle control parameter calibration relationship that matches the vehicle operating state, and thus solving the technical problem of mismatch between vehicle control parameters and vehicle operating state caused by the inability of traditional vehicle control parameter calibration methods to comprehensively calibrate vehicle control parameters.

[0162] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0163] Embodiments of the present invention also provide a non-volatile storage medium. Optionally, in this embodiment, the aforementioned non-volatile storage medium can be used to store the program code executed by the vehicle control parameter calibration method provided in the above embodiments.

[0164] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0165] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: acquiring first motion state information of the vehicle, first planning information of the vehicle, and a pre-calibrated first calibration relationship matching the vehicle model, wherein the first calibration relationship represents the correspondence between vehicle control parameters and vehicle motion state information, vehicle control parameters are parameters for controlling vehicle power, the first motion state information represents the motion state of the vehicle at a first moment, and the first planning information represents the expected motion state of the vehicle at a second moment after the first moment; determining the first vehicle control parameters of the vehicle at the first moment in the first calibration relationship based on the first motion state information and the first planning information; after the vehicle adjusts its motion state according to the first vehicle control parameters, acquiring second motion state information of the vehicle, wherein the second motion state information represents the motion state of the vehicle at the second moment; updating the first calibration relationship based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle, wherein the second calibration relationship represents the correspondence between the vehicle control parameters and motion state information in the current state.

[0166] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining the vehicle control parameter range and vehicle speed range of the test vehicle; selecting multiple first sampling points within the vehicle control parameter range and multiple second sampling points within the vehicle speed range; inputting the multiple first sampling points and multiple second sampling points into a target neural network model to determine multiple vehicle acceleration values, wherein the target neural network model is used to determine the vehicle acceleration values ​​through the vehicle control parameter values ​​and vehicle speed values; determining a first calibration relationship based on the multiple first sampling points, multiple second sampling points, and multiple vehicle acceleration values, wherein the first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles with the same vehicle model as the test vehicle, and the vehicle motion state information includes vehicle speed and vehicle acceleration.

[0167] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0168] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0169] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0170] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0171] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0172] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0173] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for calibrating vehicle control parameters, characterized in that, include: The system acquires the vehicle's first motion state information, the vehicle's first planning information, and a pre-calibrated first calibration relationship that matches the vehicle's model. The first calibration relationship represents the correspondence between vehicle control parameters and vehicle motion state information. This first calibration relationship is pre-calibrated based on the test vehicle's control parameter range and speed range. The test vehicle matches the vehicle's model. The vehicle control parameters are parameters for controlling vehicle power. The first motion state information represents the vehicle's motion state at a first moment, and the first planning information represents the vehicle's expected motion state at a second moment after the first moment. Based on the first motion state information and the first planning information, the first vehicle control parameters of the vehicle at the first moment are determined in the first calibration relationship; After the vehicle adjusts its motion state according to the first vehicle control parameters, the second motion state information of the vehicle is obtained, wherein the second motion state information represents the motion state of the vehicle at the second moment. The first calibration relationship is updated based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle, wherein the second calibration relationship represents the correspondence between the vehicle control parameters and the motion state information in the current state; At multiple time points following the second time point, the second calibration relationship is iteratively updated using the method for obtaining the second calibration relationship, thereby determining the target calibration relationship that matches the vehicle.

2. The method according to claim 1, characterized in that, The step of updating the first calibration relationship based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle includes: Determine the error value between the second motion state information and the first planning information; Based on the error value, the second motion state information, and the first calibration relationship, determine the correction value of the motion state information in the first calibration relationship; The first calibration relationship is updated based on the correction value, and the second calibration relationship is generated.

3. The method according to claim 2, characterized in that, The step of determining the correction value of the motion state information in the first calibration relationship based on the error value, the second motion state information, and the first calibration relationship includes: Based on the first calibration relationship, a two-dimensional Gaussian distribution is generated, with the numerical values ​​of the physical quantities representing the second motion state information serving as the distribution center. Based on the two-dimensional Gaussian distribution, determine the update coefficients of the motion state information of the first calibration relationship; The correction value is determined based on the error value and the update coefficient.

4. The method according to claim 2, characterized in that, Before determining the correction value of the motion state information in the first calibration relationship based on the error value, the second motion state information, and the first calibration relationship, the method further includes: Obtain a predetermined error range; If the error value is within the error range, a correction value for the motion state information in the first calibration relationship is determined based on the error value, the second motion state information, and the first calibration relationship. When the error value includes position error and velocity error, and one of the position error and the velocity error is within the error range, while the other error is outside the error range and less than the minimum value of the error range, the correction value of the motion state information in the first calibration relationship is determined based on the velocity error, the second motion state information, and the first calibration relationship.

5. A method for calibrating vehicle control parameters, characterized in that, include: Obtain the range of vehicle control parameters and vehicle speed of the test vehicle; Multiple first sampling points are selected within the range of vehicle control parameters, and multiple second sampling points are selected within the range of vehicle speed, wherein the multiple first sampling points and the multiple second sampling points correspond one-to-one; The plurality of first sampling points and the plurality of second sampling points are input into the target neural network model to determine a plurality of vehicle acceleration values, wherein the target neural network model is used to determine the vehicle acceleration values ​​through vehicle control parameter values ​​and vehicle speed values; A first calibration relationship is determined based on the plurality of first sampling points, the plurality of second sampling points, and the plurality of vehicle acceleration values. The first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles whose vehicle models match the test vehicle. The vehicle motion state information includes vehicle speed and vehicle acceleration. The first calibration relationship is sent to a vehicle whose vehicle model matches the test vehicle, wherein the vehicle obtains a second calibration relationship applicable to the actual operating state of the vehicle by updating the first calibration relationship in real time during vehicle operation.

6. The method according to claim 5, characterized in that, The target neural network model is a model trained through the following steps: Multiple sets of correspondences are obtained for the test vehicle, wherein each set of correspondences represents the correspondence between vehicle control parameter values, vehicle speed values, and vehicle acceleration values; Based on the multiple sets of correspondences, multiple different types of original neural network models are trained to obtain multiple trained neural network models. The target neural network model is selected from the plurality of trained neural network models.

7. The method according to claim 6, characterized in that, The step of selecting the target neural network model from the plurality of trained neural network models includes: Calculate the mean squared error and coefficient of determination for each of the multiple neural network models; Among the multiple neural network models, the neural network model with the smallest mean square error and the largest coefficient of determination is selected as the target neural network model.

8. The method according to claim 6, characterized in that, The acquisition of multiple sets of correspondences for the test vehicle includes: Acquire multiple sets of test data for the test vehicle, and multiple wheel angle values ​​that correspond one-to-one with the multiple sets of test data. Each set of test data includes the correspondence between vehicle test control parameter values, vehicle test speed values, and vehicle test acceleration values. Among the plurality of wheel rotation angle values, the angle values ​​that fall within a predetermined wheel rotation angle range are determined to be multiple target angle values; The multiple sets of test data corresponding to the multiple target angle values ​​are determined as the multiple sets of correspondences.

9. The method according to claim 5, characterized in that, The step of determining the first calibration relationship based on the plurality of first sampling points, the plurality of second sampling points, and the plurality of vehicle acceleration values ​​includes: Based on the plurality of first sampling points, the plurality of second sampling points, and the first acceleration value, a throttle calibration relationship for the test vehicle is generated, wherein the first acceleration value is a non-negative value among the plurality of vehicle acceleration values, the throttle calibration relationship characterizes the correspondence between throttle opening data and vehicle motion state information, and the values ​​of the plurality of second sampling points characterizing the throttle opening data in the throttle calibration relationship are non-negative values; Based on the plurality of first sampling points, the plurality of second sampling points, and the second acceleration value, a braking calibration relationship for the test vehicle is generated, wherein the second acceleration value is a negative value among the plurality of vehicle acceleration values, and the braking calibration relationship characterizes the correspondence between brake opening data and vehicle motion state information, wherein the values ​​of the plurality of second sampling points characterizing the brake opening data in the braking calibration relationship are negative values; The throttle calibration relationship and the brake calibration relationship are combined to obtain the first calibration relationship.

10. The method according to claim 9, characterized in that, The step of inputting the plurality of first sampling points and the plurality of second sampling points into the target neural network model to determine multiple vehicle acceleration values ​​includes: The plurality of first sampling points and the plurality of second sampling points are input into a first neural network model, and the first neural network model outputs the first acceleration value. The target neural network model includes the first neural network model, which is trained based on the throttle opening data and vehicle motion state information measured during the acceleration of the test vehicle. The vehicle control parameters include the throttle opening data. The plurality of first sampling points and the plurality of second sampling points are input into a second neural network model, and the second neural network model outputs the second acceleration value. The target neural network model includes the second neural network model, which is trained based on the brake opening data and vehicle motion state information measured during the deceleration of the test vehicle. The vehicle control parameters include the brake opening data.

11. A computer device, characterized in that, include: Memory and processor The memory stores computer programs; The processor is configured to execute a computer program stored in the memory, wherein the computer program, when executed, causes the processor to perform the following methods: The system acquires the vehicle's first motion state information, the vehicle's first planning information, and a pre-calibrated first calibration relationship that matches the vehicle's model. The first calibration relationship represents the correspondence between vehicle control parameters and vehicle motion state information. The vehicle control parameters are parameters for controlling vehicle power. The first motion state information represents the vehicle's motion state at a first moment, and the first planning information represents the vehicle's expected motion state at a second moment after the first moment. Based on the first motion state information and the first planning information, the first vehicle control parameters of the vehicle at the first moment are determined in the first calibration relationship; After the vehicle adjusts its motion state according to the first vehicle control parameters, the second motion state information of the vehicle is obtained, wherein the second motion state information represents the motion state of the vehicle at the second moment. The first calibration relationship is updated based on the second motion state information and the first planning information to generate a second calibration relationship applicable to the current state of the vehicle, wherein the second calibration relationship represents the correspondence between the vehicle control parameters and the motion state information in the current state; At multiple time points following the second time point, the second calibration relationship is iteratively updated using the method for obtaining the second calibration relationship, thereby determining the target calibration relationship that matches the vehicle.

12. A computer device, characterized in that, include: Memory and processor The memory stores computer programs; The processor is configured to execute a computer program stored in the memory, wherein the computer program, when executed, causes the processor to perform the following methods: Obtain the range of vehicle control parameters and vehicle speed of the test vehicle; Multiple first sampling points are selected within the range of vehicle control parameters, and multiple second sampling points are selected within the range of vehicle speed. The plurality of first sampling points and the plurality of second sampling points are input into the target neural network model to determine a plurality of vehicle acceleration values, wherein the target neural network model is used to determine the vehicle acceleration values ​​through vehicle control parameter values ​​and vehicle speed values; A first calibration relationship is determined based on the plurality of first sampling points, the plurality of second sampling points, and the plurality of vehicle acceleration values. The first calibration relationship characterizes the correspondence between vehicle control parameters and vehicle motion state information in vehicles with the same vehicle model as the test vehicle. The vehicle motion state information includes vehicle speed and vehicle acceleration.

13. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the non-volatile storage medium to perform the vehicle control parameter calibration method according to any one of claims 1 to 10.

Images