WO2024082213A1

WO2024082213A1 - Method and apparatus for constructing vehicle dynamics model, device, and storage medium

Info

Publication number: WO2024082213A1
Application number: PCT/CN2022/126417
Authority: WO
Inventors: 罗杰; 刘栋豪; 梁艺潇; 张永生; 何朗
Original assignee: 华为技术有限公司
Priority date: 2022-10-20
Filing date: 2022-10-20
Publication date: 2024-04-25

Abstract

The present application is applied to the technical field of automobiles, and disclosed thereby are a method and apparatus for constructing a vehicle dynamics model, a device, and a storage medium. The vehicle dynamics model constructed by the present application comprises a tire data model, a tire mechanism model, and a whole vehicle dynamics model. The construction method comprises: acquiring real vehicle driving data of a vehicle under a first driving control parameter; inputting the first driving control parameter into the tire data model and the tire mechanism model, respectively, and obtaining first tire stress data and second tire stress data; fusing the first tire stress data and the second tire stress data, and obtaining fused tire stress data; inputting the fused tire stress data into the whole vehicle dynamics model, and obtaining simulated driving data; and according to the simulated driving data and the real vehicle driving data, performing parameter adjustment on the tire data model. By using the present application, a vehicle dynamics model that can accurately reflect the dynamic performance of a vehicle can be constructed.

Description

Method, device, equipment and storage medium for constructing vehicle dynamics model

Technical Field

The present application relates to the field of automobile technology, and in particular to a method, device, equipment and storage medium for constructing a vehicle dynamics model.

Background technique

With the continuous development of autonomous vehicles, the control algorithms of autonomous driving systems are also being updated and iterated. In order to more efficiently verify the accuracy of the control algorithm during the iterative update process, a vehicle dynamics model is usually built in the simulation system. In a specified driving scenario, the control algorithm generates driving control parameters (such as steering wheel angle, accelerator pedal opening, brake pedal opening, etc.), which are input into the vehicle dynamics model to obtain simulated driving data (such as vehicle speed, acceleration, tire speed, etc.). The control accuracy of the control algorithm is verified based on the simulated driving data.

The construction of the vehicle dynamics model plays a vital role in the verification process of the control algorithm. However, due to the limitation that automobile manufacturers cannot provide accurate parameters of various vehicle components, the constructed vehicle dynamics model cannot accurately reflect the dynamic performance of the vehicle. Therefore, how to build a vehicle dynamics model that can accurately reflect the dynamic characteristics of the vehicle has become an urgent problem to be solved.

Summary of the invention

The embodiments of the present application provide a method, apparatus, device and storage medium for constructing a vehicle dynamics model, which can construct a vehicle dynamics model that can more accurately reflect the dynamic performance of the vehicle.

In a first aspect, a method for constructing a vehicle dynamics model is provided, wherein the vehicle dynamics model includes a tire data model, a tire mechanism model, and a whole vehicle dynamics model, and the construction method includes: obtaining actual vehicle driving data of the vehicle under a first driving control parameter. Inputting the first driving control parameter into the tire data model and the tire mechanism model respectively to obtain first tire force data and second tire force data. Fusing the first tire force data and the second tire force data to obtain fused tire force data. Inputting the fused tire force data into the whole vehicle dynamics model to obtain simulated driving data. Adjusting the tire data model according to the simulated driving data and the actual vehicle driving data.

In the solution provided by the present application, a tire data model is introduced on the basis of the existing vehicle dynamics model. The tire data model is a machine learning model obtained by training and adjusting parameters based on real vehicle driving data. The tire force data output by the tire data model and the tire force data output by the tire mechanism model are fused and input into the whole vehicle dynamics model, and then the whole vehicle dynamics model outputs simulated driving data. In the construction of the vehicle dynamics model, the real vehicle driving data is combined, so that the constructed vehicle dynamics model is more in line with the dynamic performance of the real vehicle.

In a possible implementation, the first tire force data includes first lateral force data and first longitudinal force data, and the second tire force data includes second lateral force data and second longitudinal force data. Accordingly, the first tire force data and the second tire force data may be fused as follows:

The first lateral force data and the second lateral force data are fused to obtain fused lateral force data. The first longitudinal force data and the second longitudinal force data are fused to obtain fused longitudinal force data.

The above process of inputting the fused tire force data into the vehicle dynamics model may be as follows:

The fused lateral force data and the fused longitudinal force data are input into the vehicle dynamics model to obtain simulated driving data.

In a possible implementation, when the tire force data is fused, the lateral force data output by the two models can be weighted summed, and the longitudinal force models output by the two models can be weighted summed. In addition, the lateral force fusion weight and the longitudinal force fusion weight used in the weighted summation are obtained according to the driving control parameters. That is, the lateral force fusion weight and the longitudinal force fusion weight change adaptively with the change of the driving control parameters.

In addition to the driving control parameters, the lateral force fusion weight and the longitudinal force fusion weight can also be determined according to the actual vehicle driving data. That is, the lateral force fusion weight and the longitudinal force fusion weight are adaptively changed with the changes in the driving control parameters and the actual vehicle driving data. Thus, the constructed vehicle dynamics model can better adapt to various working conditions.

In one possible implementation, a correspondence between driving control parameters, lateral force fusion weights, and longitudinal force fusion weights may be established in advance, and then, when determining the first lateral force fusion weight and the first longitudinal force fusion weight corresponding to the first driving control parameter, this may be achieved by querying the above correspondence.

In a possible implementation, a first lateral force fusion weight and a first longitudinal force fusion weight corresponding to the first driving control parameter may also be obtained through a fusion weight decision model. The fusion weight decision model may be a machine learning model, specifically, a clustering model, such as a K-means clustering model, a density clustering (Density-Based Spatial Clustering of Applications with Noise, DBSCAN), and the like.

In a possible implementation, the data dimension of the simulated driving data is the same as the data dimension of the actual vehicle driving data, and the data dimension includes at least one of lateral speed, longitudinal speed, tire speed, lateral acceleration, longitudinal acceleration, and yaw angle.

In one possible implementation, in order to improve the efficiency of model training, in the solution provided in the present application, when it is determined that the first error between the data of the same data dimension in the actual vehicle driving data and the simulated driving data is less than the corresponding error threshold, training parameters can be adjusted based on the actual vehicle driving data and the simulated driving data.

In a possible implementation, the process of adjusting the parameters of the tire data model according to the simulated driving data and the actual vehicle driving data may be as follows:

According to the weights corresponding to the data dimensions, the first errors are weighted and summed to obtain the second error. If the second error is not less than the second threshold, it is determined that the error between the simulated driving data and the actual vehicle driving data does not meet the convergence condition, and the tire data model is adjusted. The parameter adjustment method can be a gradient descent method, a Bayesian extreme value search method, and a neural network (NN) fitting method.

In a second aspect, a method for generating simulated driving data is provided, the method comprising:

Inputting the first driving control parameter into the tire data model and the tire mechanism model respectively to obtain first tire force data and second tire force data;

fusing the first tire force data and the second tire force data to obtain fused tire force data;

The fused tire force data is input into the whole vehicle dynamics model to obtain simulated driving data.

In a possible implementation, the first tire force data includes first lateral force data and first longitudinal force data, the second tire force data includes second lateral force data and second longitudinal force data, and the first tire force data and the second tire force data are fused to obtain fused tire force data, including:

fusing the first lateral force data and the second lateral force data to obtain fused lateral force data;

fusing the first longitudinal force data and the second longitudinal force data to obtain fused longitudinal force data;

Inputting the fused tire force data into the vehicle dynamics model to obtain simulated driving data includes:

The fused lateral force data and the fused longitudinal force data are input into the whole vehicle dynamics model to obtain simulated driving data.

In a possible implementation, the method further includes:

determining a first lateral force fusion weight and a first longitudinal force fusion weight according to the first driving control parameter;

The fusing the first lateral force data and the second lateral force data to obtain fused lateral force data includes:

performing weighted summation on the first lateral force data and the second lateral force data according to the first lateral force fusion weight to obtain fused lateral force data;

The fusing the first longitudinal force data and the second longitudinal force data to obtain fused longitudinal force data includes:

According to the first longitudinal force fusion weight, the first longitudinal force data and the second longitudinal force data are weightedly summed to obtain fused longitudinal force data.

In a possible implementation manner, determining a first lateral force fusion weight and a first longitudinal force fusion weight according to the first driving control parameter includes:

According to the corresponding relationship between the driving control parameter, the lateral force fusion weight and the longitudinal force fusion weight, a first lateral force fusion weight and a first longitudinal force fusion weight corresponding to the first driving control parameter are determined.

The driving control parameter is input into a weight decision model to obtain a first lateral force fusion weight and a first longitudinal force fusion weight corresponding to the first driving control parameter.

In a third aspect, a device for constructing a vehicle dynamics model is provided, the device comprising modules for executing the method for constructing a vehicle dynamics model in the first aspect or any possible implementation of the first aspect.

In a fourth aspect, a device for generating simulated driving data is provided, the device comprising modules for executing the method for generating simulated driving data in the second aspect or any possible implementation of the second aspect.

In a fifth aspect, a computer device is provided, comprising a processor and a memory, wherein the memory is used to store instructions, and the instructions are loaded and executed by the processor to implement the method for constructing a vehicle dynamics model as described in the first aspect or any possible implementation of the first aspect.

In a sixth aspect, a computer device is provided, comprising a processor and a memory, wherein the memory is used to store instructions, and the instructions are loaded and executed by the processor to implement the method for generating simulated driving data as described in the second aspect or any possible implementation of the second aspect.

In a seventh aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and the instructions are loaded and executed by a processor to implement the method for constructing a vehicle dynamics model as described in the first aspect or any possible implementation of the first aspect.

In an eighth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and the instructions are loaded and executed by a processor to implement the method for generating simulated driving data as described in the second aspect or any possible implementation of the second aspect.

In a ninth aspect, a computer program product is provided, the computer program product comprising instructions, the instructions being loaded and executed by a processor to implement the method of constructing a vehicle dynamics model as described in the first aspect or any possible implementation of the first aspect.

In a tenth aspect, a computer program product is provided, the computer program product comprising instructions, the instructions being loaded and executed by a processor to implement the method for generating simulated driving data as described in the second aspect or any possible implementation of the second aspect.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic diagram of a vehicle dynamics model provided by an exemplary embodiment of the present application;

FIG2 is a flow chart of a method for constructing a vehicle dynamics model provided by an exemplary embodiment of the present application;

FIG3 is a schematic diagram of a vehicle dynamics model provided by an exemplary embodiment of the present application;

FIG4 is a flow chart of a method for generating simulated driving data provided by an exemplary embodiment of the present application;

FIG5 is a schematic diagram of a device structure for constructing a vehicle dynamics model provided by an exemplary embodiment of the present application;

FIG6 is a schematic diagram of a structure of a device for generating simulated driving data provided by an exemplary embodiment of the present application;

FIG. 7 is a schematic diagram of the structure of a computer device provided by an exemplary embodiment of the present application.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present application more clear, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

The embodiment of the present application provides a method for constructing a vehicle dynamics model. The method introduces a tire data model based on the existing vehicle dynamics model. The tire data model is obtained by training and adjusting parameters of real vehicle driving data. The tire force data output by the tire data model and the tire force data output by the tire mechanism model are fused and input into the whole vehicle dynamics model, and then the whole vehicle dynamics model outputs simulated driving data. In the construction of the vehicle dynamics model, the real vehicle driving data is combined, so that the constructed vehicle dynamics model is more consistent with the dynamic performance of the real vehicle.

The following is a comparison between the vehicle dynamics model constructed using the present application and the existing vehicle dynamics model in conjunction with the accompanying drawings.

Referring to FIG1 , an existing vehicle dynamics model is shown. The vehicle dynamics model includes a tire mechanism model and a vehicle dynamics model. The vehicle dynamics model includes a tire dynamics model and a vehicle body dynamics model. Both the tire mechanism model and the vehicle dynamics model are models constructed based on physical principles. The tire mechanism model and the vehicle dynamics model are coupled via tire force data, wherein the tire force data is output by the tire mechanism model. For ease of description, the tire force data output by the tire mechanism model is recorded as: F _model . After the driving control parameters are input into the tire mechanism model, the tire mechanism model outputs the tire force data F _model , and outputs F _model to the vehicle dynamics model, and then the vehicle dynamics model outputs the simulated driving data.

Continuing to refer to FIG. 1 , a vehicle dynamics model constructed using the method provided by the present application is also shown. Compared with existing vehicle dynamics models, a tire data model is introduced into the vehicle dynamics model constructed using the method provided by the present application. After the driving control parameters are input into the tire mechanism model and the tire data model, the tire data model outputs the first tire force data, and the tire mechanism model outputs the second tire force data F _model . For ease of description, the first tire force data is recorded as: F _data . Then, the first tire force data F _data and the second tire force data F _model are fused to obtain fused force data F _adaptive . The fused force data F _adaptive is input into the whole vehicle dynamics model, and the whole vehicle dynamics model outputs simulated driving data. Among them, the tire data model is a machine learning model, which is obtained by training and adjusting parameters of real vehicle driving data, so that the constructed vehicle dynamics model can more accurately reflect the dynamic performance of the vehicle.

The method for constructing a vehicle dynamics model provided in the embodiment of the present application can be implemented by a computer device, which can be a terminal or a server. When the computer device is a terminal, it can be a desktop computer, a laptop computer, etc. When the computer device is a server, it can be a single server, a server cluster, or a virtual machine deployed in the server.

The computer equipment obtains the actual vehicle driving data of the vehicle and the simulated driving data output by the vehicle dynamics model under the same driving control parameters, and then trains and adjusts the tire data model according to the actual vehicle driving data and the simulated driving data to obtain the vehicle dynamics model including the adjusted tire data model.

The following is a description of the method for constructing a vehicle dynamics model provided by an embodiment of the present application in conjunction with the accompanying drawings. Referring to FIG2 , the method may include the following processing steps:

Step 201: Acquire actual vehicle driving data under driving control parameters.

In implementation, vehicles of the same model may use the same vehicle dynamics model. Therefore, when constructing the vehicle dynamics model, a corresponding vehicle dynamics model may be constructed for each vehicle model.

When building a vehicle dynamics model for any vehicle model, the vehicle can be driven manually first, and when driving the vehicle, the accelerator pedal opening, brake pedal opening and steering wheel angle need to be changed to make the vehicle drive under different working conditions. When the vehicle is driving, the driving control parameters and actual vehicle driving data of the vehicle can be collected through sensors.

Specifically, the driving control parameters may include multiple data dimensions, such as steering wheel angle, brake pedal opening, accelerator pedal opening, etc. Among them, the steering wheel angle can be collected by a steering wheel angle sensor (Steering Angle Sensor, SAS), and the brake pedal opening and accelerator pedal opening can be collected by corresponding pedal travel sensors (Pedal Travel Sensor, PTS).

The actual vehicle driving data may include multiple data dimensions, such as the vehicle's lateral speed, longitudinal speed, wheel speed, lateral acceleration, longitudinal acceleration and yaw angle, etc. Among them, the lateral speed and longitudinal speed can be collected by the RT sensor, the vehicle speed can be collected by the wheel speed sensor (WSS), and the lateral acceleration, longitudinal acceleration and yaw angle can be collected by the inertial measurement unit (IMU).

The computer device obtains the driving control parameters and the actual vehicle driving data collected by the sensor. In addition, the data collected by the sensor may have data glitches. After obtaining the driving control parameters and the actual vehicle driving data collected by the sensor, the computer device can perform data cleaning by filtering to eliminate the data glitches.

In order to facilitate data statistics and subsequent training of tire data models, the data collection period of each data dimension in the above-mentioned real vehicle driving data and the collection period of driving control parameters can be set to be the same. Specifically, the data collection period of the above-mentioned SAS, PTS, RT, WSS, IMU and other sensors can be set to be the same. In this way, the computer device can use the driving control parameters and the corresponding real vehicle driving data obtained in the same collection period as a set of benchmark training data.

In addition, it should be noted that the data dimension of the above-mentioned actual vehicle driving data is only an example. According to the different degrees of freedom requirements of the vehicle dynamics model, the data dimension of the actual vehicle driving data can be increased or decreased. This application does not limit the data dimension of the actual vehicle driving data.

Step 202: Input the driving control parameters into the tire data model and the tire mechanism model respectively to obtain the first tire force data and the second tire force data.

In implementation, for each set of benchmark training data obtained in the above step 201, the driving control parameters in the set of benchmark training data are respectively input into the tire data model and the tire mechanism model to obtain the first tire force data output by the tire data model and the second tire force data output by the tire mechanism model.

Specifically, the tire force data may include lateral force data and longitudinal force data. Accordingly, after the driving control parameters are respectively input into the tire data model and the tire mechanism model, the first lateral force data and the first longitudinal force data output by the tire data model, and the second lateral force data and the second longitudinal force data output by the tire mechanism model can be obtained.

Step 203: Fusing the first tire force data and the second tire force data to obtain fused tire force data.

In implementation, the first lateral force data and the second lateral force data are fused to obtain fused lateral force data. The first longitudinal force data and the second longitudinal force data are fused to obtain fused longitudinal force data.

Referring to FIG3 , the first lateral force data and the first longitudinal force data output by the tire data model are recorded as: F _x,data and F _y,data , and the second lateral force data and the second longitudinal force data output by the tire mechanism model are recorded as: F _x,model and F _y,model , respectively. The first lateral force data F _x,data and the second lateral force data F _x,model are fused to obtain fused lateral force data F _x,adaptive , and the first longitudinal force data F _y,data and the second longitudinal force data F _y,model are fused to obtain fused longitudinal force data F _y,adaptive .

The above fusion process is described below. In a possible implementation, the process may include the following steps:

Step 2031: Obtain the target horizontal fusion weight and the target vertical fusion weight.

The values of the lateral fusion weight and the longitudinal fusion weight are related to the driving control parameters and/or the actual vehicle driving data. Based on this, there are many methods for obtaining the target lateral fusion weight and the target longitudinal fusion weight. Several of these methods are described below as examples.

Method 1: Implement by presetting rules and looking up the table.

Considering that the values of the horizontal fusion weight and the vertical fusion weight may be related to multiple data, this method can have multiple situations.

Case 1:

Considering that the values of the lateral fusion weight and the longitudinal fusion weight are related to the driving control parameters, the corresponding relationship between the driving control parameters, the lateral fusion weight and the longitudinal fusion weight can be established in advance, as shown in the following Table 1:

Table 1

驾驶控制参数Driving control parameters	横向融合权重Horizontal fusion weight	纵向融合权重Vertical fusion weight
A1A1	w _x1 w _x1	w _y1 w _y1
A2A2	w _x2 w _x2	w _y2 w _y2
A3A3	w _x3 w _x3	w _y2 w _y2
……	……	……

In the above Table 1, each row of data in the column where the driving control parameter is located can represent a value range set, and the value range set can include at least one data dimension value range of the throttle pedal opening, the accelerator pedal opening, and the steering wheel angle. The specific data dimension value ranges included can be configured according to actual needs, for example, including the value ranges of the throttle pedal opening, the accelerator pedal opening, and the steering wheel angle.

In this case, according to the driving control parameters in the reference training data, the corresponding target lateral weight and target longitudinal fusion weight are determined in the corresponding relationship among the driving control parameters, the lateral fusion weight and the longitudinal fusion weight.

For example, if the values of the driving control parameters in the reference training data are within the value range A1 in the above Table 1, the target lateral fusion weight will be determined to be w _x1 , and the target longitudinal fusion weight will be determined to be w _y1 .

Case 2:

Considering that the values of the lateral fusion weight and the longitudinal fusion weight are related to the actual vehicle driving data, the corresponding relationship between the driving control parameter actual vehicle driving data, the lateral fusion weight and the longitudinal fusion weight can be established in advance, as shown in Table 2 below:

Table 2

实车行驶数据Real vehicle driving data	横向融合权重Horizontal fusion weight	纵向融合权重Vertical fusion weight
B1B1	w _x1 w _x1	w _y1 w _y1
B2B2	w _x2 w _x2	w _y2 w _y2
B3B3	w _x3 w _x3	w _y2 w _y2
……	……	……

In the above Table 2, the actual vehicle driving data can be represented in the form of a value range. Among them, each row of data in the column where the actual vehicle driving data is located can represent a value range set, and the value range set can include at least one data dimension value range of the lateral speed, the longitudinal speed, the wheel speed, the lateral acceleration, the longitudinal acceleration, and the yaw angle. The specific data dimension value ranges can be configured according to actual needs, for example, including the value ranges of the throttle pedal opening, the accelerator pedal opening, and the steering wheel angle.

In this case, according to the actual vehicle driving data in the benchmark training data, the corresponding target lateral weight and target longitudinal fusion weight are determined in the corresponding relationship among the actual vehicle driving data, the lateral fusion weight and the longitudinal fusion weight.

For example, if the value of the actual vehicle driving data in the benchmark training data is within the value range B2 in the above Table 2, the target lateral fusion weight will be determined to be w _x2 and the target longitudinal fusion weight will be determined to be w _y2 .

Case 3:

Considering that the values of the lateral fusion weight and the longitudinal fusion weight are related to the driving control parameters and the actual vehicle driving data, the corresponding relationship between the driving control parameters, the actual vehicle driving data, the lateral fusion weight and the longitudinal fusion weight is established in advance, as shown in the following Table 3:

table 3

驾驶控制参数Driving control parameters	实车行驶数据Real vehicle driving data	横向融合权重Horizontal fusion weight	纵向融合权重Vertical fusion weight
A1A1	B1B1	w _x1 w _x1	w _y1 w _y1
A2A2	B2B2	w _x2 w _x2	w _y2 w _y2
A3A3	B3B3	w _x3 w _x3	w _y2 w _y2
……	……	……	……

In this case, according to the driving control parameters and actual vehicle driving parameters in the benchmark training data, the corresponding target lateral fusion weight and target longitudinal fusion weight are determined in the correspondence between the driving control parameters, the actual vehicle driving data, the lateral fusion weight and the longitudinal fusion weight.

For example, if the driving control parameter value in the benchmark training data is within the value range A3 in Table 3, and the actual vehicle driving data value in the benchmark training data is within the value range B3, the target lateral fusion weight is determined to be w _x3 and the target longitudinal fusion weight is determined to be w _y3 .

Method 2: Implemented through fusion weight decision model.

The fusion weight decision model can be a machine learning model, specifically, a clustering model, such as K-means clustering model, density clustering (Density-Based Spatial Clustering of Applications with Noise, DBSCAN), etc.

Case 1:

Considering that the values of the lateral fusion weight and the longitudinal fusion weight are related to the driving control parameters, the driving control parameters in the benchmark training data can be input into the fusion weight decision model to obtain the target lateral fusion weight and the target longitudinal fusion weight.

Case 2:

Considering that the values of the lateral fusion weight and the longitudinal fusion weight are related to the actual vehicle driving data, the actual vehicle driving data in the benchmark training data can be input into the fusion weight decision model to obtain the target lateral fusion weight and the target longitudinal fusion weight.

Case 3:

Considering that the values of the lateral fusion weight and the longitudinal fusion weight are related to both the driving control parameters and the actual vehicle driving data, the driving control parameters and the actual vehicle driving data in the benchmark training data can be input into the fusion weight decision model to obtain the target lateral fusion weight and the target longitudinal fusion weight.

In addition, the target horizontal fusion weight and the target vertical fusion weight may be obtained using the same fusion weight decision model or different fusion weight decision models, and this application does not limit this.

Step 2032: Perform weighted summation on the first lateral force data and the second lateral force data according to the target lateral fusion weight to obtain fused lateral force data.

Specifically, the weighted summation process can be performed according to the following formula:

F _x，adaptive ＝w _x *F _x，model +(1-w _x )*F _x，data

Among them, Fx _,adaptive is the fused lateral force data, Fx _,data is the first lateral force data output by the tire data model, Fx _,model is the second lateral force data output by the tire mechanism model, and _wx is the target lateral fusion weight.

Step 2033: According to the target longitudinal fusion weight, weighted sum is performed on the first longitudinal force and the second longitudinal force to obtain longitudinal fusion force data.

F _y，adaptive ＝w _y *F _y，model +(1-w _y )*F _y，data

Among them, F _{y, adaptive} is the fused longitudinal force data, F _{y, data} is the first longitudinal force data output by the tire data model, F _{y, model} is the second longitudinal force data output by the tire mechanism model, and w _y is the target longitudinal fusion weight.

In a possible implementation, considering that driving control parameters of different data dimensions and actual vehicle driving data may have different degrees of influence on the lateral fusion weight and the longitudinal fusion weight, the driving control parameters and the actual vehicle driving data may be processed before determining the target lateral fusion weight and the target longitudinal fusion weight based on the driving control parameters and the actual vehicle driving data.

Specifically, the processing method can be as follows:

Before determining the target lateral fusion weight, the steering wheel angle and the longitudinal speed are multiplied by a first coefficient, wherein the first coefficient is less than 1. When obtaining the target longitudinal fusion weight, the brake pedal opening, the accelerator pedal opening, and the lateral speed are multiplied by a second coefficient, wherein the second coefficient is less than 1. The first coefficient and the second coefficient may be the same or different.

Step 204: Input the fused tire force data into the vehicle dynamics model to obtain simulated driving data.

In implementation, the fused lateral force and the fused longitudinal force are input into the vehicle dynamics model, and the vehicle dynamics model outputs simulated driving data, wherein the data dimension of the simulated driving data is the same as the data dimension of the real vehicle driving data.

In a possible implementation, in order to improve the training efficiency of the tire data model, after obtaining the simulated driving data, it is possible to first determine whether the simulated driving data satisfies the error constraint condition. If the simulated driving data satisfies the error constraint condition, then continue to execute step 205.

Specifically, the method for determining whether the simulated driving data meets the error constraint condition may be as follows:

The first error between the data of the same data dimension in the actual vehicle driving data and the simulated driving data is calculated. If the first errors are all less than the error threshold of the corresponding data dimension, it is determined that the simulated driving data meets the error constraint condition. The error threshold of each data dimension can be set according to actual needs. For example, the error threshold of the lateral acceleration can be 0.5m/ ^s2 , the error threshold of the longitudinal acceleration can be 0.5m/ ^s2 , the error threshold of the yaw angle is 0.1rad/s, the error threshold of the wheel speed is 0.5rad/s, the error threshold of the lateral speed is 1m/s, and the error threshold of the longitudinal speed is 0.2m/s.

The calculation of the first error can be expressed as follows:

ΔV _x = |V _{x, model} -V _{x, sensor} |

ΔV _y ＝|V _y，model −V _y，sensor |

ΔA _x =|A _{x, model} -A _{x, sensor} |

ΔA _y ＝|A _y，model −A _y，sensor |

Δγ＝|γ _model -γ _sensor |

Δω＝|ω _model -ω _sensor |

Among them, _ΔVx is the first error between the lateral velocity Vx _,model in the simulated driving data and the lateral velocity Vx _,sensor in the actual vehicle driving data; _ΔVy is the first error between the longitudinal velocity Vy _,model in the simulated driving data and the longitudinal velocity Vy _,sensor in the actual vehicle driving data; _ΔAx is the first error between the lateral acceleration _Ax,model in the simulated driving data and the lateral acceleration _Ax,sensor in the actual vehicle driving data; ΔAy _is the first error between the longitudinal acceleration _Ay,model in the simulated driving data and the longitudinal acceleration Ay _,sensor in the actual vehicle driving data; Δγ is the first error between the yaw angle _γmodel in the simulated driving data and the yaw angle _γsensor in the actual vehicle driving data; Δω is the first error between the vehicle speed _ωmodel in the simulated driving data and the wheel speed _ωsensor in the actual vehicle driving data.

Step 205: Determine whether the error of the simulated driving data relative to the actual vehicle driving data satisfies the convergence condition.

In implementation, different data dimensions may correspond to different weights. Accordingly, in step 205, the first errors may be weighted and summed according to the pre-configured weights corresponding to the data dimensions to obtain the second error. Then, it is determined whether the error between the simulated driving data and the actual vehicle driving data meets the convergence condition according to the second error.

In combination with the above step 204, the weighted summation of the first error can be expressed as the following formula:

loss＝ _fVx * _ΔVx + _fVy * _ΔVy + _fAx * _ΔAx + _fAy * _ΔAy + _fγ * _Δγ + _fω * _Δω

Wherein, loss is the second error, f _Vx is the weight corresponding to the lateral velocity, f _Vy is the weight corresponding to the longitudinal velocity, f _Ax is the weight corresponding to the lateral acceleration, f _Ay is the weight corresponding to the longitudinal velocity, f _γ is the weight corresponding to the yaw angle, and f _ω is the weight corresponding to the wheel speed. The sum of f _Vx , f _Vy , f _Ax , f _Ay , f _γ , and f _ω is 1.

Step 206: If the error between the simulated driving data and the actual vehicle driving data meets the convergence condition, a vehicle dynamics model is generated.

In implementation, if the second error calculated in step 205 is less than the convergence threshold, it is determined that the error of the simulated driving data relative to the actual vehicle driving data meets the convergence condition. Then, the vehicle dynamics model is generated using the current tire data model. The convergence threshold can be set according to the accuracy requirements of the vehicle dynamics model.

Step 207: If the error between the simulated driving data and the actual vehicle driving data does not meet the convergence condition, the tire data model is adjusted according to the simulated driving data and the actual vehicle driving data.

In implementation, if the second error calculated in step 205 is not less than the convergence threshold, the tire data model is adjusted according to the actual vehicle driving data and the simulated driving data. The parameter adjustment method may be a gradient descent method, a Bayesian extreme value search method, a neural network (NN) fitting method, and the like.

In a possible implementation, the weights corresponding to the data dimensions in step 205 may also be adjusted according to the convergence of the error between the simulated driving data and the actual vehicle driving data during the training and parameter adjustment process. For example, when the number of parameter adjustments reaches the upper limit, if the error between the simulated driving data and the actual vehicle driving data has not yet met the convergence condition, the weights corresponding to the data dimensions may be adjusted, and the training and parameter adjustment may be continued after the adjustment.

The method for constructing a vehicle dynamics model in the embodiment of the present application introduces a tire data model based on the existing vehicle dynamics model. The tire data model is obtained by training and adjusting parameters of real vehicle driving data. The tire force data output by the tire data model and the tire force data output by the tire mechanism model are fused and input into the whole vehicle dynamics model, and then the whole vehicle dynamics model outputs simulated driving data. In the construction of the vehicle dynamics model, the real vehicle driving data is combined, so that the constructed vehicle dynamics model is more consistent with the dynamic performance of the real vehicle.

In conjunction with the above method for constructing a vehicle dynamics model, the embodiment of the present application further provides a method for generating simulated driving data, which can be implemented using the vehicle dynamics model constructed by the above method. The method can be applied in the control algorithm verification process of an autonomous driving vehicle, and can also be applied in the vehicle controller to perform online driving prediction. In conjunction with Figure 4, the processing of the method can include the following steps:

Step 401: Input the first driving control parameter into the tire data model and the tire mechanism model respectively to obtain the first tire force data and the second tire force data.

In implementation, during the control algorithm verification process, the first driving control parameter generated by the control algorithm is obtained, and the first driving control parameter is input into the tire data model and the tire mechanism model respectively to obtain the first tire force data and the second tire force data.

When the application is used in a vehicle controller to perform online driving prediction, the controller obtains a first driving control parameter input by a driver, and inputs the first driving control parameter into a tire data model and a tire mechanism model respectively to obtain first tire force data and second tire force data.

Step 402: Fusing the first tire force data and the second tire force data to obtain fused tire force data.

In practice, the processing of step 402 is similar to the processing of step 203, and will not be described in detail here. It should be noted that in step 402, since there is no real vehicle driving data, when obtaining the target lateral fusion weight and the target longitudinal fusion weight, the simulated driving data output by the vehicle dynamics model is used as the real vehicle driving data.

Step 403: input the fused tire force data into the vehicle dynamics model to obtain simulated driving data.

During implementation, the fused lateral force and the fused longitudinal force are input into the vehicle dynamics model, and the vehicle dynamics model outputs simulated driving data.

The embodiment of the present application also provides a device for constructing a vehicle dynamics model, which can be a computer device. As shown in FIG5 , the device includes an acquisition module 510, an input module 520, a fusion module 530, a simulation module 540 and a parameter adjustment module 550, wherein:

The acquisition module 510 is used to acquire the actual vehicle driving data of the vehicle under the first driving control parameter; specifically, it can implement the processing of the above step 201 and its implicit steps.

The input module 520 is used to input the first driving control parameter into the tire data model and the tire mechanism model respectively to obtain the first tire force data and the second tire force data; specifically, the processing of the above step 202 and its implicit steps can be realized.

The fusion module 530 is used to fuse the first tire force data and the second tire force data to obtain fused tire force data; specifically, it can implement the processing of the above step 203 and its implicit steps.

The simulation module 540 is used to input the fused tire force data into the whole vehicle dynamics model to obtain simulated driving data; specifically, it can implement the processing of the above step 204 and its implicit steps.

The parameter adjustment module 550 is used to adjust the parameters of the tire data model according to the simulated driving data and the real vehicle driving data. Specifically, the processing of the above steps 205-207 and the implicit steps thereof can be implemented.

In a possible implementation, the first tire force data includes first lateral force data and first longitudinal force data, and the first fusion module 530 is used to:

The simulation module 540 is used to:

In a possible implementation, the fusion module 530 is used to:

In a possible implementation, the fusion module 530 is further configured to:

It is determined that first errors between data of the same data dimension in the actual vehicle driving data and the simulated driving data are both smaller than an error threshold.

In a possible implementation, the parameter adjustment module 550 is used to:

According to the weights corresponding to the data dimensions, weighted summation is performed on the first errors to obtain the second error;

If the second error is not less than a second threshold, it is determined that the error between the simulated driving data and the actual vehicle driving data does not meet the convergence condition, and the tire data model is adjusted.

It should be noted that: the device for constructing a vehicle dynamics model provided in the above embodiment only uses the division of the above functional modules as an example when constructing a vehicle dynamics model. In actual applications, the above functional distribution can be completed by different functional modules as needed, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the device for constructing a vehicle dynamics model provided in the above embodiment and the method embodiment for constructing a vehicle dynamics model belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

The embodiment of the present application further provides a device for generating simulated driving data, which may be a computer device or a controller. Referring to FIG. 6 , the device includes an input module 610, a fusion module 620, and a simulation module 630, wherein:

The input module 610 is used to input the first driving control parameter into the tire data model and the tire mechanism model respectively to obtain the first tire force data and the second tire force data; specifically, the processing of the above step 401 and its implicit steps can be realized.

The fusion module 620 is used to fuse the first tire force data and the second tire force data to obtain fused tire force data; specifically, it can implement the processing of the above step 402 and its implicit steps.

The simulation module 630 is used to input the fused tire force data into the vehicle dynamics model to obtain simulated driving data. Specifically, the processing of the above step 403 and its implicit steps can be implemented.

In a possible implementation, the first tire force data includes first lateral force data and first longitudinal force data, the second tire force data includes second lateral force data and second longitudinal force data, and the fusion module 620 is used to:

The simulation module 630 is used to:

In a possible implementation, the fusion module 620 is configured to:

According to the corresponding relationship among the driving control parameter, the lateral force fusion weight and the longitudinal force fusion weight, a first lateral force fusion weight and a first longitudinal force fusion weight corresponding to the first driving control parameter are determined.

In a possible implementation, the fusion module 620 is configured to:

It should be noted that: the device for generating simulated driving data provided in the above embodiment only uses the division of the above functional modules as an example when generating simulated driving data. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device or controller is divided into different functional modules to complete all or part of the functions described above. In addition, the device for generating simulated driving data provided in the above embodiment and the method embodiment for generating simulated driving data belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

As shown in FIG7 , the computer device 700 may be implemented by a general bus architecture. The computer device 700 includes at least one processor 701 , a communication bus 702 , a memory 703 , and at least one network interface 704 .

The processor 701 is, for example, a general-purpose central processing unit (CPU), a network processor (NP), a graphics processing unit (GPU), a neural-network processing units (NPU), a data processing unit (DPU), a microprocessor, or one or more integrated circuits for implementing the solution of the present application. For example, the processor 701 includes an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The PLD is, for example, a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

The communication bus 702 is used to transmit information between the above components. The communication bus 702 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in FIG7 , but it does not mean that there is only one bus or one type of bus.

The memory 703 is, for example, a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, or a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 703 is, for example, independent and connected to the processor 701 through the communication bus 702. The memory 703 can also be integrated with the processor 701.

The network interface 704 uses any transceiver-like device for communicating with other devices or communication networks. The network interface 704 includes a wired network interface and may also include a wireless network interface. Among them, the wired network interface may be, for example, an Ethernet interface. The Ethernet interface may be an optical interface, an electrical interface, or a combination thereof. The wireless network interface may be a wireless local area network (WLAN) interface, a cellular network network interface, or a combination thereof, etc.

In a specific implementation, as an example, the processor 701 may include one or more CPUs.

In a specific implementation, as an example, the computer device 700 may include multiple processors. Each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here may refer to one or more devices, circuits, and/or processing cores for processing data (such as computer program instructions).

In a specific implementation, as an example, the computer device 700 may also include an output device and an input device. The output device communicates with the processor 701 and can display information in a variety of ways. For example, the output device may be a liquid crystal display (LCD), a light emitting diode (LED) display device, a cathode ray tube (CRT) display device, or a projector. The input device communicates with the processor 701 and receives user input in a variety of ways. For example, the input device may be a mouse, a keyboard, a touch screen device, or a sensor device.

In some embodiments, the memory 703 is used to store and execute the program code 7031 for detecting malicious files in the present application, and the processor 701 executes the program code 7031 stored in the memory 703. That is, the computer device 700 can implement the method for constructing a vehicle dynamics model or the method for generating simulated driving data provided by the method embodiment through the processor 701 and the program code 7031 in the memory 703.

The present application also provides a computer-readable storage medium, such as a memory including a program code, which can be executed by a processor in a device to complete the method of constructing a vehicle dynamics model and/or the method of generating simulated driving data in the above embodiment. The implementation of the computer-readable storage medium can refer to the memory 703 in FIG. 7 .

The present application also provides a computer program product or a computer program, which includes a program code, and the program code is stored in a computer-readable storage medium. A processor reads the program code from the computer-readable storage medium, and the processor executes the program code, so that the device where the processor is located executes the above-mentioned method for constructing a vehicle dynamics model and/or the method for generating simulated driving data.

In addition, the present application also provides a device, which can specifically be a chip, component or module, and the device may include a connected processor and memory; wherein the memory is used to store computer-executable instructions, and when the device is running, the processor can execute the computer-executable instructions stored in the memory, so that the chip executes the method of constructing a vehicle dynamics model and/or the method of generating simulated driving data in the above-mentioned method embodiment.

Among them, the devices, equipment, computer-readable storage media, computer program products or chips provided in this application are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, the process or function described in the embodiment of the present application of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD) or a semiconductor medium. The semiconductor medium can be a solid state disk (SSD).

The above is only a specific implementation of the present application. Those skilled in the art may conceive of changes or substitutions based on the specific implementation provided by the present application, which should all be included in the protection scope of the present application.

Claims

A method for constructing a vehicle dynamics model, characterized in that the vehicle dynamics model includes a tire data model, a tire mechanism model and a whole vehicle dynamics model, and the method includes:

Acquiring actual vehicle driving data of the vehicle under the first driving control parameter;

Inputting the first driving control parameter into the tire data model and the tire mechanism model respectively to obtain first tire force data and second tire force data;

fusing the first tire force data and the second tire force data to obtain fused tire force data;

Inputting the fused tire force data into the whole vehicle dynamics model to obtain simulated driving data;

The tire data model is adjusted according to the simulated driving data and the actual vehicle driving data.
The method according to claim 1, characterized in that the first tire force data includes first lateral force data and first longitudinal force data, the second tire force data includes second lateral force data and second longitudinal force data, and the fusing the first tire force data and the second tire force data to obtain the fused tire force data comprises:

fusing the first lateral force data and the second lateral force data to obtain fused lateral force data;

fusing the first longitudinal force data and the second longitudinal force data to obtain fused longitudinal force data;

The step of inputting the fused tire force data into the vehicle dynamics model to obtain simulated driving data includes:

The fused lateral force data and the fused longitudinal force data are input into the whole vehicle dynamics model to obtain simulated driving data.
The method according to claim 2, characterized in that the method further comprises:

determining a first lateral force fusion weight and a first longitudinal force fusion weight according to the first driving control parameter;

The fusing the first lateral force data and the second lateral force data to obtain fused lateral force data includes:

performing weighted summation on the first lateral force data and the second lateral force data according to the first lateral force fusion weight to obtain fused lateral force data;

The fusing the first longitudinal force data and the second longitudinal force data to obtain fused longitudinal force data includes:

According to the first longitudinal force fusion weight, the first longitudinal force data and the second longitudinal force data are weightedly summed to obtain fused longitudinal force data.
The method according to claim 3, characterized in that the determining, according to the first driving control parameter, a first lateral force fusion weight and a first longitudinal force fusion weight comprises:

According to the corresponding relationship between the driving control parameter, the lateral force fusion weight and the longitudinal force fusion weight, a first lateral force fusion weight and a first longitudinal force fusion weight corresponding to the first driving control parameter are determined.
The method according to claim 3, characterized in that the determining, according to the first driving control parameter, a first lateral force fusion weight and a first longitudinal force fusion weight comprises:

The driving control parameter is input into a fusion weight decision model to obtain a first lateral force fusion weight and a first longitudinal force fusion weight corresponding to the first driving control parameter.
The method according to claim 1 is characterized in that the data dimension of the simulated driving data is the same as the data dimension of the real vehicle driving data, and the data dimension includes at least one of lateral speed, longitudinal speed, tire speed, lateral acceleration, longitudinal acceleration, and yaw angle.
The method according to claim 6, characterized in that before fusing the first tire force data and the second tire force data, the method further comprises:

It is determined that first errors between data of the same data dimension in the actual vehicle driving data and the simulated driving data are both smaller than an error threshold.
The method according to claim 7, characterized in that the step of adjusting the parameters of the tire data model according to the simulated driving data and the real vehicle driving data comprises:

According to the weights corresponding to the data dimensions, weighted summation is performed on the first errors to obtain the second error;

If the second error is not less than a second threshold, it is determined that the error between the simulated driving data and the actual vehicle driving data does not meet the convergence condition, and the tire data model is adjusted.
A method for generating simulated driving data, characterized in that the method comprises:

Inputting the first driving control parameter into the tire data model and the tire mechanism model respectively to obtain first tire force data and second tire force data;

fusing the first tire force data and the second tire force data to obtain fused tire force data;

The fused tire force data is input into the whole vehicle dynamics model to obtain simulated driving data.
The method according to claim 9, characterized in that the first tire force data includes first lateral force data and first longitudinal force data, the second tire force data includes second lateral force data and second longitudinal force data, and the fusing the first tire force data and the second tire force data to obtain the fused tire force data comprises:

fusing the first lateral force data and the second lateral force data to obtain fused lateral force data;

fusing the first longitudinal force data and the second longitudinal force data to obtain fused longitudinal force data;

Inputting the fused tire force data into the vehicle dynamics model to obtain simulated driving data includes:

The fused lateral force data and the fused longitudinal force data are input into the whole vehicle dynamics model to obtain simulated driving data.
The method according to claim 10, characterized in that the method further comprises:

determining a first lateral force fusion weight and a first longitudinal force fusion weight according to the first driving control parameter;

The fusing the first lateral force data and the second lateral force data to obtain fused lateral force data includes:

performing weighted summation on the first lateral force data and the second lateral force data according to the first lateral force fusion weight to obtain fused lateral force data;

The fusing the first longitudinal force data and the second longitudinal force data to obtain fused longitudinal force data includes:

According to the first longitudinal force fusion weight, the first longitudinal force data and the second longitudinal force data are weightedly summed to obtain fused longitudinal force data.
The method according to claim 11, characterized in that the determining, according to the first driving control parameter, a first lateral force fusion weight and a first longitudinal force fusion weight comprises:

According to the corresponding relationship among the driving control parameter, the lateral force fusion weight and the longitudinal force fusion weight, a first lateral force fusion weight and a first longitudinal force fusion weight corresponding to the first driving control parameter are determined.
The method according to claim 12, characterized in that the determining, according to the first driving control parameter, a first lateral force fusion weight and a first longitudinal force fusion weight comprises:

The driving control parameter is input into a fusion weight decision model to obtain a first lateral force fusion weight and a first longitudinal force fusion weight corresponding to the first driving control parameter.
A device for constructing a vehicle dynamics model, characterized in that the vehicle dynamics model includes a tire data model, a tire mechanism model and a whole vehicle dynamics model, and the device includes:

An acquisition module, used for acquiring actual vehicle driving data of the vehicle under the first driving control parameter;

An input module, used for inputting the first driving control parameter into the tire data model and the tire mechanism model respectively, to obtain first tire force data and second tire force data;

a fusion module, used for fusing the first tire force data and the second tire force data to obtain fused tire force data;

A simulation module, used for inputting the fused tire force data into the whole vehicle dynamics model to obtain simulated driving data;

A parameter adjustment module is used to adjust the parameters of the tire data model according to the simulated driving data and the actual vehicle driving data.
A device for generating simulated driving data, characterized in that the device comprises:

An input module, used for inputting the first driving control parameter into the tire data model and the tire mechanism model respectively, to obtain first tire force data and second tire force data;

a fusion module, used for fusing the first tire force data and the second tire force data to obtain fused tire force data;

The simulation module is used to input the fused tire force data into the whole vehicle dynamics model to obtain simulated driving data.
A computer device, characterized in that the computer device includes a processor and a memory, the memory is used to store instructions, and the instructions are loaded and executed by the processor to implement the method for constructing a vehicle dynamics model according to any one of claims 1 to 8.
A computer device, characterized in that the computer device includes a processor and a memory, the memory is used to store instructions, and the instructions are loaded and executed by the processor to implement the method for generating simulated driving data described in any one of claims 9-14.
A computer-readable storage medium, characterized in that the computer-readable storage medium stores instructions, and the instructions are loaded and executed by a processor to implement the method for constructing a vehicle dynamics model according to any one of claims 1 to 8.
A computer-readable storage medium, characterized in that the computer-readable storage medium stores instructions, and the instructions are loaded and executed by a processor to implement the method for generating simulated driving data according to any one of claims 9 to 14.