Method, device and vehicle for determining road adhesion coefficient based on multi-terrain recognition

Abstract

The present application relates to the field of intelligent vehicles, and provides a method, device and vehicle for determining a road adhesion coefficient based on multi-terrain recognition. The method includes: obtaining the current chassis data, current road image and current position information of the vehicle in the current driving section; performing multi-terrain recognition on the current driving section based on the current chassis data, current road image and current position information to determine a reference range value of the road adhesion coefficient corresponding to the current driving section; determining the slip rate of the vehicle when driving on the current driving section, and estimating the estimated value of the road adhesion coefficient corresponding to each wheel end based on the slip rate; determining the road adhesion coefficient output value corresponding to each wheel end according to the reference range value of the road adhesion coefficient and the estimated value of the road adhesion coefficient. The present application can improve the recognition accuracy of the terrain in which the vehicle is located and the recognition accuracy of the road adhesion coefficient during driving, thereby improving the accuracy and reliability of the control target of the control-by-wire chassis.

Description

Technical Field

[0001] The present application relates to the field of intelligent vehicles, and in particular to a method, device and vehicle for determining a road adhesion coefficient based on multi-terrain recognition. Background Art

[0002] During driving, the accuracy of identifying the terrain on which the vehicle is located and the road adhesion coefficient are key factors affecting whether the drive-by-wire chassis can achieve accurate and reliable following control goals.

[0003] In the related art, due to the lack of recognition of the terrain on which the vehicle is located during driving or the low recognition accuracy, the recognition accuracy of the road adhesion coefficient is low, which leads to reduced accuracy and reliability of the control target followed by the controlled chassis.

[0004] Therefore, how to improve the recognition accuracy of the terrain on which the vehicle is located during driving and how to improve the recognition accuracy of the road adhesion coefficient have become technical problems that need to be urgently solved by those skilled in the art. Summary of the invention

[0005] In view of this, an embodiment of the present application provides a method, device and vehicle for determining a road adhesion coefficient based on multi-terrain recognition to solve the problem of how to improve the recognition accuracy of the terrain in which the vehicle is located during driving, thereby improving the recognition accuracy of the road adhesion coefficient.

[0006] In a first aspect of an embodiment of the present application, a method for determining a road adhesion coefficient based on multi-terrain recognition is provided, comprising:

[0007] Obtain the current chassis data, current road image and current position information of the vehicle on the current driving section;

[0008] Perform multi-terrain recognition on the current driving section based on the current chassis data, the current road image and the current position information to determine the reference range value of the road adhesion coefficient corresponding to the current driving section;

[0009] Determine the slip rate of the vehicle when it is traveling on the current driving section, and estimate the estimated value of the road adhesion coefficient corresponding to each wheel end of the vehicle when it is traveling on the current driving section based on the slip rate;

[0010] According to the road adhesion coefficient reference range value and the road adhesion coefficient estimation value, the road adhesion coefficient output value corresponding to each wheel end when the vehicle is traveling on the current driving section is determined.

[0011] According to a second aspect of an embodiment of the present application, a device for determining a road adhesion coefficient based on multi-terrain recognition is provided, comprising:

[0012] An acquisition module is configured to acquire current chassis data, current road image and current position information of the vehicle on the current driving section;

[0013] A multi-terrain recognition module is configured to perform multi-terrain recognition on a current driving section based on current chassis data, a current road image and current position information to determine a reference range value of a road adhesion coefficient corresponding to the current driving section;

[0014] an estimation module, configured to determine a slip rate of the vehicle when traveling on a current driving section, and estimate a road adhesion coefficient estimation value corresponding to each wheel end of the vehicle when traveling on the current driving section based on the slip rate;

[0015] The output module is configured to determine the road adhesion coefficient output value corresponding to each wheel end when the vehicle is traveling on the current driving section according to the road adhesion coefficient reference range value and the road adhesion coefficient estimation value.

[0016] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0017] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0018] Compared with the prior art, the beneficial effects of the embodiments of the present application include at least the following: in the multi-terrain recognition stage, the three dimensions of the human-vehicle-road state of the vehicle during driving are comprehensively considered, and the multi-terrain recognition is performed in combination with the current chassis data, current road image and current position information of the vehicle in the current driving section, thereby improving the accuracy of recognition of the terrain in which the vehicle is located during driving and improving various performances of the vehicle under complex road conditions; after completing the multi-terrain recognition, the road adhesion coefficient reference range value is determined based on the multi-terrain recognition result, and then the road adhesion coefficient output value corresponding to each wheel end of the vehicle when driving on the current driving section is determined in combination with the road adhesion coefficient reference range value and the road adhesion coefficient estimation value, which can improve the recognition accuracy of the road adhesion coefficient, thereby improving the accuracy and reliability of the control target of the wire-controlled chassis following. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 It is a scenario diagram of an application scenario of an embodiment of the present application.

[0021] Figure 2 It is a flow chart of a method for determining a road adhesion coefficient based on multi-terrain recognition provided in an embodiment of the present application.

[0022] Figure 3 It is a structural schematic diagram of a road adhesion coefficient determination device based on multi-terrain recognition provided in an embodiment of the present application.

[0023] Figure 4 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application.

[0024] Among them, the reference numerals in the figures are as follows:

[0025] Vehicle 100; vehicle controller 101, instrument 102, chassis control system 103;

[0026] Cloud 200;

[0027] Road adhesion coefficient determination device 300; acquisition module 301, multi-terrain recognition module 302, estimation module 303, output module 304;

[0028] Electronic device 400 ; processor 401 , memory 402 , computer program 403 . DETAILED DESCRIPTION

[0029] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0030] A method and device for determining a road adhesion coefficient based on multi-terrain recognition according to an embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0031] With the intelligence and electrification of automobiles, automobile chassis are gradually developing towards wire control. The wire control chassis has the advantages of high integration, fast response speed, and good controllability. It can make more precise and reliable follow-up control targets according to the terrain where the vehicle is located (such as asphalt roads, snow, sand, rocks, etc.), thereby improving the driving safety and ride comfort of the vehicle.

[0032] Although not considering the impact of terrain on vehicle control during driving can reduce economic costs, this will significantly reduce the vehicle's lateral stability under complex road conditions, thereby affecting driving safety and user experience.

[0033] In the related technologies, the vehicle control schemes involving terrain recognition mainly include: 1) The driver manually selects the vehicle control strategy. For example, the driver can determine the actual terrain during driving based on experience, and then manually select to start the road condition driving mode corresponding to the actual terrain. However, due to the lack of automatic recognition of the terrain in which the vehicle is located during driving, this method requires the driver to always pay attention to the changes in the actual terrain during driving and actively switch the corresponding vehicle control strategy. This not only increases the complexity of the driver's operation and reduces the driving experience, but also requires the driver to frequently switch in the case of short-distance terrain changes, affecting driving safety. 2) Some other vehicle control strategies, although they have automatic recognition of the terrain in which the vehicle is located during driving, the factors considered for the automatic terrain recognition schemes in these vehicle control strategies are relatively single (usually only the slip rate or road image is considered), the stability for complex road conditions is poor, the accuracy of terrain recognition is low, and it is impossible to perform high-precision road adhesion coefficient recognition, which is not conducive to calculating the vehicle control target, resulting in reduced accuracy and reliability of the control target of the wire-controlled chassis following control.

[0034] In view of this, an embodiment of the present application proposes a method for determining a road adhesion coefficient based on multi-terrain recognition. In the multi-terrain recognition stage, the three dimensions of the human-vehicle-road state of the vehicle during driving are comprehensively considered, and multi-terrain recognition is performed in combination with the current chassis data, current road image and current position information of the vehicle in the current driving section, thereby improving the accuracy of identifying the terrain in which the vehicle is located during driving and improving various performances of the vehicle under complex road conditions. After completing the multi-terrain recognition, a road adhesion coefficient reference range value is determined based on the multi-terrain recognition result, and then the road adhesion coefficient reference range value and the road adhesion coefficient estimation value are combined to determine the road adhesion coefficient output value corresponding to each wheel end when the vehicle is driving on the current driving section, which can improve the recognition accuracy of the road adhesion coefficient, thereby improving the accuracy and reliability of the control target followed by the controlled chassis by wire.

[0035] Figure 1 is a schematic diagram of an application scenario provided by an embodiment of the present application. Figure 1, the application scenario includes a vehicle 100 and a cloud 200. The vehicle 100 is a smart vehicle, specifically a new energy vehicle. The new energy vehicles in the embodiment of the present application include but are not limited to the following types of vehicles: electric vehicles (EV), pure electric vehicles (BEV), fuel cell electric vehicles (FCEV), plug-in hybrid electric vehicles (PHEV) and hybrid electric vehicles (HEV). The cloud 200 can be a remote server, which is the core part of the cloud computing architecture and can be used to store, process and transmit data, and provide users with various computing services and resources.

[0036] The vehicle 100 may include a vehicle controller 101, an instrument 102, and a chassis control system 103. The vehicle controller 101 may include a CAN (Controller Area Network) interface, an image interface, a communication interface, a first terrain recognition model connected to the CAN interface, a second terrain recognition model connected to the image interface, and a third terrain recognition model connected to the communication interface; a multi-terrain recognition fusion module connected to the first terrain recognition model, the second terrain recognition model, and the third terrain recognition model, and an output interface connected to the multi-terrain recognition fusion module, the output interface being connected to the instrument 102 and the chassis control system 103, respectively.

[0037] The CAN interface can be used to obtain the current chassis data of the vehicle 100 during the process of driving on the current driving section. The image interface can be used to obtain the current road image of the vehicle 100 during the process of driving on the current driving section. The vehicle 100 can communicate with the cloud 200 through the communication interface and send its current position information during the process of driving on the current driving section to the cloud 200. After obtaining the current position information sent by the vehicle 100, the cloud 200 can perform multi-terrain recognition on the current driving section based on the current position information and return the multi-terrain recognition result to the vehicle 100.

[0038] Figure 2 is a flow chart of a method for determining a road surface adhesion coefficient based on multi-terrain recognition provided by an embodiment of the present application. The method for determining a road surface adhesion coefficient based on multi-terrain recognition provided by an embodiment of the present application can be Figure 1 The vehicle controller 101 is used to execute.

[0039] like Figure 2 As shown, the method for determining the road adhesion coefficient based on multi-terrain recognition may include the following steps:

[0040] Step S201, obtaining the current chassis data, current road image and current position information of the vehicle on the current driving section.

[0041] The current chassis data includes at least one of the driver manipulation data, vehicle driving state data or vehicle model parameters. The driver manipulation data includes but is not limited to: at least one of the accelerator pedal opening, steering wheel angle or brake pedal opening. The vehicle driving state data includes but is not limited to: driving mode, four wheel speeds, angle, vertical acceleration, vehicle longitudinal speed, longitudinal acceleration, lateral acceleration, vertical acceleration, center of mass sideslip angle, yaw rate, body roll angle or body roll rate. Vehicle model parameters include but are not limited to: at least one of the vehicle inherent parameters such as wheelbase, mass, wind resistance or transmission ratio.

[0042] As an example, see Figure 1 The vehicle controller 101 can obtain the current chassis data of the vehicle 100 during the driving process on the current driving section through the CAN interface; obtain the current road image collected by the vehicle-mounted image sensor of the vehicle 100 through the image interface; and collect the current position information of the vehicle 100 when driving on the current driving section through the positioning system installed on the vehicle 100 (such as the vehicle-mounted GPS positioning system, etc.).

[0043] Step S202: performing multi-terrain recognition on the current driving section based on the current chassis data, the current road image and the current position information to determine a reference range value of the road adhesion coefficient corresponding to the current driving section.

[0044] Multi-terrain recognition refers to the use of specific technical systems and algorithms, using a variety of sensor equipment such as lidar, cameras, millimeter-wave radar, ultrasonic sensors, etc. to collect the current chassis data, current road images and current position information of the current driving section of the vehicle during driving, and then use machine learning, deep learning, image recognition and other intelligent algorithms to process and analyze these data, so as to accurately determine the road terrain corresponding to the current driving section of the vehicle.

[0045] Usually, when a vehicle is driving on a road section with different road terrain, its road adhesion coefficient is often different. Among them, the common road terrain includes two categories, one is paved road, including but not limited to dry asphalt road, wet asphalt road, dry cement road, wet cement road, ice and snow road; the other is unpaved road, including but not limited to sand, dirt road, mud, rock, and grass.

[0046] In practical applications, the reference range of the road adhesion coefficient when the vehicle is traveling on various common road terrains can be calibrated through actual vehicle testing, and a mapping relationship between various road terrains and the reference range of the road adhesion coefficient can be established.

[0047] Step S203, determining the slip rate of the vehicle when traveling on the current driving section, and estimating the estimated value of the road adhesion coefficient corresponding to each wheel end of the vehicle when traveling on the current driving section based on the slip rate.

[0048] As an example, the vehicle controller 101 can obtain the driving speed and wheel speed of the vehicle 100 when it is traveling on the current driving section through the vehicle speed sensor and wheel speed sensor installed on the vehicle 100, and then calculate the slip rate of the vehicle 100 when it is traveling on the current driving section based on the obtained driving speed and wheel speed, and then estimate the estimated value of the road adhesion coefficient corresponding to each wheel end of the vehicle when it is traveling on the current driving section based on the slip rate.

[0049] Step S204, determining the road adhesion coefficient output value corresponding to each wheel end when the vehicle is traveling on the current driving section according to the road adhesion coefficient reference range value and the road adhesion coefficient estimation value.

[0050] As an example, the vehicle controller 101 can use the road adhesion coefficient reference range value to correct the road adhesion coefficient estimation value corresponding to each wheel end when the vehicle is traveling on the current driving section, obtain the road adhesion coefficient output value corresponding to each wheel end, and output the road adhesion coefficient output value to the chassis control system 103 of the vehicle 100 through its output interface. The chassis control system 103 can use the road adhesion coefficient output value to adjust and optimize various vehicle function parameters of the vehicle 100 (such as ABS (Anti-lock Brake System), EBD (Electric Brake Distribution), VDC (Vehicle Dynamic Control), TCS (Traction Control System), electric power assistance intensity, etc.), so that the vehicle can better adapt to different driving environments and conditions, and at the same time provide the driver with a safer, more stable and comfortable driving experience.

[0051] As an example, the slip rate threshold value of the vehicle can be optimized and adjusted according to the road adhesion coefficient output value. For example, the target range of the slip rate can be linearly adjusted according to the road adhesion coefficient output value. Generally, as the road adhesion coefficient increases, the slip rate threshold value can be increased and the range can be expanded accordingly. In this way, the peak adhesion coefficient of the road surface can be fully utilized on a high adhesion coefficient road surface to obtain a better braking effect, and the risk of wheel locking can be reduced on a low adhesion coefficient road surface to maintain the stability and steering performance of the vehicle.

[0052] As another example, the braking pressure of the vehicle can be adjusted and optimized according to the output value of the road adhesion coefficient. Generally, as the road adhesion coefficient increases, the adjustment frequency of the braking pressure can be increased accordingly. In this way, the wheel can quickly reach the target slip rate on a high-adhesion road surface, improve the acceleration performance, and avoid frequent wheel instability on a low-adhesion road surface. As the road adhesion coefficient increases, the adjustment amplitude can be increased accordingly. In this way, the wheel can quickly correct the wheel slip state on a high-adhesion road surface, and on a low-adhesion road surface, prevent the wheel from losing grip due to excessive changes in braking pressure.

[0053] The technical solution provided in the embodiment of the present application comprehensively considers the three dimensions of the human-vehicle-road state of the vehicle during driving in the multi-terrain recognition stage, and performs multi-terrain recognition in combination with the current chassis data, current road image and current position information of the vehicle in the current driving section, thereby improving the recognition accuracy of the terrain in which the vehicle is located during driving and improving various performances of the vehicle under complex road conditions; after completing the multi-terrain recognition, the road adhesion coefficient reference range value is determined based on the multi-terrain recognition result, and then the road adhesion coefficient output value corresponding to each wheel end of the vehicle when driving on the current driving section is determined in combination with the road adhesion coefficient reference range value and the road adhesion coefficient estimation value, which can improve the recognition accuracy of the road adhesion coefficient, thereby improving the accuracy and reliability of the control target followed by the controlled chassis by wire.

[0054] In some embodiments, the road adhesion coefficient reference range value includes a first road adhesion coefficient reference range value, a second road adhesion coefficient reference range value, and a third road adhesion coefficient reference range value;

[0055] Based on the current chassis data, the current road image and the current position information, multi-terrain recognition is performed on the current driving section to determine the reference range value of the road adhesion coefficient corresponding to the current driving section, including:

[0056] Performing multi-terrain recognition on the current driving section based on the current chassis data to determine a first road surface adhesion coefficient reference range value corresponding to the current driving section; wherein the first road surface adhesion coefficient reference range value includes a first road surface adhesion coefficient reference lower limit value and a first road surface adhesion coefficient reference upper limit value;

[0057] Performing multi-terrain recognition on the current driving section based on the current road image to determine a second road surface adhesion coefficient reference range value corresponding to the current driving section; wherein the second road surface adhesion coefficient reference range value includes a second road surface adhesion coefficient reference lower limit value and a second road surface adhesion coefficient reference upper limit value;

[0058] Based on the current position information, multi-terrain identification is performed on the current driving section to determine the third road adhesion coefficient reference range value corresponding to the current driving section; wherein the third road adhesion coefficient reference range value includes a third road adhesion coefficient reference lower limit value and a third road adhesion coefficient reference upper limit value.

[0059] As an example, see Figure 1 The vehicle controller 101 can obtain chassis data of the vehicle 100 (or other vehicles of the same model as the vehicle 100) during driving on various road terrains through the CAN interface, and then perform data preprocessing on the chassis data through the principal component analysis method to obtain main feature data; thereafter, the main feature data is input into the initial terrain recognition model as a training data set for training, and the hyperparameters of the initial terrain recognition model are optimized based on TPE (Tree-structured ParzenEstimator). When the training reaches a preset convergence condition (such as the model accuracy reaches a preset accuracy threshold or the training rounds reach a preset number threshold), a trained first terrain recognition model is obtained.

[0060] Among them, the principal component analysis method is a commonly used unsupervised dimensionality reduction technology, which aims to project the original data into a low-dimensional space through linear transformation while retaining the key information of the data to the greatest extent.

[0061] By preprocessing the acquired current chassis data using the principal component analysis method, the current chassis data can be projected into a low-dimensional space, thereby achieving feature dimensionality reduction, which is beneficial to improving the computational efficiency of subsequent model training and enhancing model performance.

[0062] The initial terrain recognition model can use classification algorithm models such as LightGBM (Light Gradient Boosting Machine, lightweight gradient boosting machine algorithm) model and naive Bayesian classification algorithm.

[0063] The hyperparameters of the initial terrain recognition model mainly include basic booster parameters, learning rate and iteration related parameters, regularization parameters, and data sampling and feature processing parameters.

[0064] TPE search is a hyperparameter optimization algorithm that is often used in the Bayesian optimization framework and can efficiently find the optimal hyperparameter combination for machine learning models.

[0065] In some embodiments, performing multi-terrain recognition on the current driving section based on the current chassis data to determine a first road adhesion coefficient reference range value corresponding to the current driving section includes:

[0066] Input the current chassis data into the first terrain recognition model to perform multi-terrain recognition, and output the first terrain probability value corresponding to the current driving section;

[0067] Determining a road surface terrain corresponding to a current driving section based on the first terrain probability value;

[0068] Based on a preset mapping relationship between a road surface topography and a road surface adhesion coefficient reference range value, a first road surface adhesion coefficient reference range value corresponding to the road surface topography is queried.

[0069] In practical applications, see Figure 1 The vehicle controller 101 can obtain the current chassis data of the vehicle 100 when it is traveling on the current driving section through the CAN interface, and then pre-process the current chassis data through principal component analysis to obtain main feature data (feature data obtained by projecting the driver's operation data, vehicle driving state data and vehicle model parameters into a low-dimensional space); then input the main feature data into the trained first terrain recognition model for multi-terrain recognition, and output the first terrain probability value corresponding to the current driving section.

[0070] As an example, the mapping relationship between the preset road surface terrain and the road surface adhesion coefficient reference range value is shown in Table 1:

[0071] Table 1 Mapping relationship between road topography and road adhesion coefficient reference range

[0072]

[0073] The first terrain probability value output by the first terrain recognition model is ;in, Indicates i The probability value corresponding to the road terrain, i =1,2,3,......,10, i The values ​​of correspond to the numbers in Table 1. For example, when i =1, it indicates the road surface topography corresponding to number 1 (i.e. dry asphalt road); i =2, it means the road surface topography corresponding to number 2 (i.e. wet asphalt road surface), ... and so on. i =10, it indicates the road terrain corresponding to number 10 (i.e. grass).

[0074] Assume that the first terrain probability value output by the first terrain recognition model is )= , it can be determined that the probability of the current driving section being a wet asphalt road is the highest, so it can be determined that the road terrain corresponding to the current driving section is a wet asphalt road. Then, by querying the above Table 1, it can be determined that the reference range value of the first road adhesion coefficient corresponding to the wet asphalt road is 0.6~0.7.

[0075] As an example, see Figure 1 The vehicle controller 101 can obtain road images of the vehicle 100 (or other vehicles of the same model as the vehicle 100) during driving on various road terrains through an image interface; then, road images with good discrimination are selected as training data sets, and the BiseNetV2 (Bilateral SegmentationNetwork V2, dual-channel segmentation network) semantic segmentation network is used to perform image segmentation on these road images to segment the upper sky part and the middle building in the road image, and only the lower road part is retained to obtain a road feature image; the road feature image is then used to train the to-be-trained model, and when the training reaches a preset convergence condition (such as the model accuracy reaches a preset accuracy threshold or the number of training rounds reaches a preset number threshold), a trained second terrain recognition model is obtained.

[0076] Among them, the model to be trained can be selected from the PULC (Practical Ultra Light Classification) model, etc.

[0077] The road images used to train the model to be trained need to cover various road terrains (including but not limited to dry asphalt roads, wet asphalt roads, dry cement roads, wet cement roads, ice and snow roads, sand, dirt roads, mud, rocks, and grass). This is conducive to improving the model generalization ability and recognition accuracy of the second terrain recognition model.

[0078] Road images have good discrimination, which means that there are obvious differences and features between different types of road images, and different types of road terrain can be clearly distinguished. For example, there are obvious differences in texture and surrounding environment between dry asphalt roads and wet asphalt roads. Road images should accurately reflect these differences, so that manual or machine annotation can be more accurately judged and annotated, reducing the possibility of misjudgment and confusion, thereby ensuring the annotation quality.

[0079] The installation height and angle of the vehicle-mounted image sensor are related to the discrimination of the collected training data set. In practical applications, the discrimination of the collected training data set can be improved by adjusting the installation height and angle of the vehicle-mounted image sensor.

[0080] In practical applications, see Figure 1The vehicle controller 101 can obtain the current road image collected by the on-board image sensor of the vehicle 100 through the image interface, and output the current road image to the trained second terrain recognition model, and output the second terrain probability value corresponding to the current driving section.

[0081] The second terrain probability value output by the second terrain recognition model is ;in, Indicates i The probability value corresponding to the road terrain, i =1,2,3,......,10, i The values ​​of correspond to the numbers in Table 1. For example, when i =1, it indicates the road surface topography corresponding to number 1 (i.e. dry asphalt road); i =2, it means the road surface topography corresponding to number 2 (i.e. wet asphalt road surface), ... and so on. i =10, it indicates the road terrain corresponding to number 10 (i.e. grass).

[0082] Assume that the second terrain probability value output by the second terrain recognition model is = , it can be determined that the probability of the current driving section being a dry asphalt road is the highest, so it can be determined that the road terrain corresponding to the current driving section is a dry asphalt road. Then, by querying the above Table 1, it can be determined that the second road adhesion coefficient reference range value corresponding to the dry asphalt road is 0.8~1.0.

[0083] As an example, the third terrain recognition model may be deployed in the vehicle controller 101 of the vehicle 100. Figure 1 , the vehicle controller 101 can collect the current position information of the vehicle 100 when it is traveling on the current driving section through the vehicle-mounted GPS (Global Positioning System) positioning system, and send the current position information to the cloud 200 through the communication interface. When the cloud 200 receives the current position information, it searches for terrain information corresponding to the current position information (including the lane line position, width, slope, curvature of the current driving section, as well as traffic signs, signal lights, lane height limits, sewer entrances, obstacles, etc.) according to the high-precision map, and transmits the terrain information to the communication interface of the vehicle controller 101, which transmits the terrain information to the third terrain recognition model through the communication interface, and then the third terrain recognition model analyzes the terrain information and outputs the third terrain probability value corresponding to the current driving section.

[0084] The third terrain recognition model can be a convolutional neural network model, a bird's-eye view network, etc.

[0085] The third terrain probability value output by the third terrain recognition model is ;in, Indicates i The probability value corresponding to the road terrain, i =1,2,3,......,10, i The values ​​of correspond to the numbers in Table 1. For example, when i =1, it indicates the road surface topography corresponding to number 1 (i.e. dry asphalt road); i =2, it means the road surface topography corresponding to number 2 (i.e. wet asphalt road surface), ... and so on. i =10, it indicates the road terrain corresponding to number 10 (i.e. grass).

[0086] Assume that the third terrain probability value output by the third terrain recognition model is = , it can be determined that the probability of the current driving section being a dry asphalt road is the highest, so it can be determined that the road terrain corresponding to the current driving section is a dry asphalt road. Then, by querying the above Table 1, it can be determined that the third road adhesion coefficient reference range value corresponding to the dry asphalt road is 0.8~1.0.

[0087] As another example, the third terrain recognition model can be deployed in the cloud 200. The vehicle controller 101 can collect the current position information of the vehicle 100 when it is traveling on the current driving section through the vehicle-mounted GPS positioning system, and send the current position information to the cloud 200 through the communication interface. When the cloud 200 receives the current position information, it searches for the terrain information corresponding to the current position information according to the high-precision map, and then inputs the terrain information into the third terrain recognition model, outputs the third terrain probability value corresponding to the current driving section, and transmits the third terrain probability value to the vehicle controller 101. The vehicle controller 101 determines the road surface terrain according to the third terrain probability value returned by the cloud 200, and then determines the third road adhesion coefficient reference range value corresponding to the road surface terrain by querying the above-mentioned Table 1.

[0088] In some embodiments, estimating the road adhesion coefficient estimate corresponding to each wheel end when the vehicle is traveling on the current driving section based on the slip rate includes:

[0089] Determine the longitudinal force of the vehicle when traveling on the current driving section based on the slip ratio and the tire characteristic parameters of the vehicle;

[0090] Determine the loads on each wheel end of the vehicle when it is traveling on the current driving section;

[0091] Based on the longitudinal force and the loads on each wheel end, an estimated value of the road adhesion coefficient corresponding to each wheel end when the vehicle is traveling on the current driving section is determined.

[0092] As an example, the longitudinal force of the vehicle when traveling on the current driving section can be calculated based on the tire model based on the slip rate and tire characteristics. For example, the longitudinal force of the vehicle when traveling on the current driving section can be calculated according to formula (1): .

[0093] (1);

[0094] In formula (1), s is the slip rate; B, C, D, and E are tire characteristic parameters, which can be obtained by fitting test data based on actual vehicle tests. Among them, B is a parameter related to the balance performance of the tire; C is a parameter related to the skeleton structure of the tire; D is a parameter related to the drainage performance of the tire; and E is a parameter related to the elasticity of the tire.

[0095] After the longitudinal forces have been calculated, the wheel loads are calculated based on the vehicle parameters and driving conditions.

[0096] Under acceleration and deceleration conditions, the front axle load and rear axle load of the vehicle are shown in the following equations (2) and (3):

[0097] (2);

[0098] (3);

[0099] In formula (2) and (3), is the front axle load, is the rear axle load, is the wheelbase, is the distance from the center of gravity to the front axle, is the distance from the center of gravity to the rear axle, is the vehicle mass, is the height of the vehicle's center of gravity, is the vehicle longitudinal acceleration, is the acceleration due to gravity.

[0100] When lateral acceleration (also called vehicle lateral acceleration) intervenes, the load distribution of the left and right wheels of the vehicle is shown in the following equations (4) and (5):

[0101] (4);

[0102] (5);

[0103] In formula (4) and (5), is the total load on the left wheel, is the total load on the right wheel, is the vehicle mass, is the height of the vehicle's center of gravity, is the vehicle width, is the vehicle lateral acceleration, is the acceleration due to gravity.

[0104] Next, based on the lateral acceleration and the calibrated lateral acceleration threshold , , calculate the single wheel load, where, Less than .

[0105] In the first case, if the lateral acceleration is less than ,Right now < , then the single wheel load (including the left front wheel load) can be calculated according to equations (6) and (7) , right front wheel load , Left rear wheel load , right rear wheel load ).

[0106] (6);

[0107] (7);

[0108] In formula (6) and (7), represents the left front wheel load, represents the right front wheel load, represents the front axle load, represents the left rear wheel load, represents the right rear wheel load, Represents the rear axle load.

[0109] In the second case, if the lateral acceleration is greater than or equal to and less than ,Right now , then the single wheel load (including the left front wheel load) can be calculated according to equations (8) to (11) , right front wheel load , Left rear wheel load , right rear wheel load ).

[0110] (8);

[0111] (9);

[0112] (10);

[0113] (11);

[0114] In formulas (8) to (11), represents the total load on the left wheel, is the total load on the right wheel, represents the front axle load, Represents the rear axle load.

[0115] In the third case, if the lateral acceleration is greater than or equal to ,Right now , then the single wheel load (including the left front wheel load) can be calculated according to equations (12) to (15) , right front wheel load , Left rear wheel load , right rear wheel load ).

[0116] (12);

[0117] (13);

[0118] (14);

[0119] (15);

[0120] In formula (12) to formula (15), represents the total load on the left wheel, is the total load on the right wheel, represents the front axle load, Represents the rear axle load.

[0121] Finally, based on the longitudinal force and each wheel end load (including the left front wheel load, the right front wheel load, the left rear wheel load and the right rear wheel load), determine the estimated value of the road adhesion coefficient corresponding to each wheel end (including the left front wheel, the right front wheel, the left rear wheel and the right rear wheel) of the vehicle in the current driving section.

[0122] For example, the estimated value of the road adhesion coefficient corresponding to each wheel end when the vehicle is traveling on the current driving section can be calculated according to equations (16) to (19).

[0123] (16);

[0124] = (17);

[0125] = (18);

[0126] = (19);

[0127] In formulas (16) to (19), Respectively represent the estimated values ​​of road adhesion coefficients corresponding to the left front wheel, right front wheel, left rear wheel, and right rear wheel; Indicates the longitudinal force of the vehicle when traveling on the current driving section; , 、 、 They represent the left front wheel load, right front wheel load, left rear wheel load and right rear wheel load respectively.

[0128] In some embodiments, according to the road adhesion coefficient reference range value and the road adhesion coefficient estimation value, determining the road adhesion coefficient output value corresponding to each wheel end when the vehicle is traveling on the current driving section includes:

[0129] According to the first road adhesion coefficient reference range value, the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section is corrected to obtain the first road adhesion coefficient correction value corresponding to each wheel end;

[0130] According to the second road adhesion coefficient reference range value, the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section is corrected to obtain the second road adhesion coefficient correction value corresponding to each wheel end;

[0131] According to the third road adhesion coefficient reference range value, the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section is corrected to obtain the third road adhesion coefficient correction value corresponding to each wheel end;

[0132] Based on the first road adhesion coefficient correction value, the second road adhesion coefficient correction value and the third road adhesion coefficient correction value corresponding to each wheel end, a road adhesion coefficient output value corresponding to each wheel end is determined.

[0133] In some embodiments, according to the first road adhesion coefficient reference range value, the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section is corrected to obtain the first road adhesion coefficient correction value corresponding to each wheel end, including:

[0134] For each wheel end, if the estimated road adhesion coefficient value corresponding to the wheel end is less than the first road adhesion coefficient reference lower limit value, then based on the first road adhesion coefficient reference lower limit value, the tolerance coefficient and the sensitivity coefficient, the estimated road adhesion coefficient value of the wheel end is corrected to obtain the first road adhesion coefficient correction value corresponding to the wheel end;

[0135] If the estimated value of the road adhesion coefficient corresponding to the wheel end is greater than or equal to the first road adhesion coefficient reference lower limit value and less than the first road adhesion coefficient reference upper limit value, the estimated value of the road adhesion coefficient corresponding to the wheel end is determined as the first road adhesion coefficient correction value corresponding to the wheel end;

[0136] If the estimated value of the road adhesion coefficient corresponding to the wheel end is greater than or equal to the first road adhesion coefficient reference upper limit value, the estimated value of the road adhesion coefficient is corrected based on the first road adhesion coefficient reference upper limit value, the tolerance coefficient and the sensitivity coefficient to obtain the first road adhesion coefficient correction value corresponding to the wheel end.

[0137] As an example, according to formula (20), the estimated road adhesion coefficient value corresponding to the left front wheel of the vehicle can be corrected using the first road adhesion coefficient reference range value to obtain the first road adhesion coefficient correction value corresponding to the left front wheel.

[0138] (20);

[0139] In formula (20), Indicates the first road adhesion coefficient correction value corresponding to the left front wheel; Indicates the reference lower limit value of the first road adhesion coefficient; Indicates the reference upper limit value of the first road adhesion coefficient; Indicates the estimated value of the road adhesion coefficient corresponding to the left front wheel; represents the tolerance factor; Represents the sensitivity coefficient.

[0140] Tolerance coefficient, representing the tolerance to the value beyond the first road adhesion coefficient reference range The larger the value, the greater the acceptance; the smaller the value, the smaller the acceptance.

[0141] Sensitivity coefficient, representing the sensitivity to the value exceeding the first road adhesion coefficient reference range The larger the value, the more sensitive the response; the smaller the value, the less sensitive the response. Sensitivity often refers to the degree to which a statistic or model responds to small changes in the data.

[0142] As an example, assuming that the first terrain probability value output by the first terrain recognition model determines that the road terrain of the current driving section is a wet asphalt road, then by querying the above Table 1, it can be determined that the reference range value of the first road adhesion coefficient is 0.6~0.7, that is, =0.6, =0.7. If the road adhesion coefficient corresponding to the left front wheel is estimated Requirements: , based on the first road adhesion coefficient reference lower limit value, tolerance coefficient and sensitivity coefficient, the estimated value of the road adhesion coefficient corresponding to the left front wheel is corrected to obtain the first road adhesion coefficient correction value corresponding to the left front wheel, that is, the first road adhesion coefficient correction value corresponding to the left front wheel is = If the estimated road adhesion coefficient of the left front wheel is Requirements: , then the estimated value of the road adhesion coefficient is determined as the first road adhesion coefficient correction value, that is, the first road adhesion coefficient correction value corresponding to the left front wheel is = If the estimated road adhesion coefficient of the left front wheel is Requirements: , based on the first road adhesion coefficient reference upper limit value, tolerance coefficient and sensitivity coefficient, the road adhesion coefficient estimation value is corrected to obtain the first road adhesion coefficient correction value, that is, the first road adhesion coefficient correction value is = .

[0143] It can be understood that the first road adhesion coefficient correction value corresponding to the right front wheel can be determined by referring to the above method. , the first road adhesion coefficient correction value corresponding to the left rear wheel , the first road adhesion coefficient correction value corresponding to the right rear wheel , I will not go into details here.

[0144] As an example, according to formula (21), the estimated road adhesion coefficient value corresponding to the left front wheel of the vehicle can be corrected using the second road adhesion coefficient reference range value to obtain the second road adhesion coefficient correction value corresponding to the left front wheel.

[0145] (twenty one);

[0146] In formula (21), Indicates the second road adhesion coefficient correction value corresponding to the left front wheel; Indicates the reference lower limit value of the second road adhesion coefficient; Indicates the second road adhesion coefficient reference upper limit value; Indicates the estimated value of the road adhesion coefficient corresponding to the left front wheel; represents the tolerance factor; Represents the sensitivity coefficient.

[0147] As an example, assuming that the second terrain probability value output by the second terrain recognition model determines that the road terrain of the current driving section is a dry asphalt road, then by querying the above Table 1, it can be determined that the reference range value of the second road adhesion coefficient is 0.8~1.0, that is, =0.8, =1.0. If the estimated road adhesion coefficient of the left front wheel meets the following conditions: , based on the second road adhesion coefficient reference lower limit value, tolerance coefficient and sensitivity coefficient, the estimated value of the road adhesion coefficient corresponding to the left front wheel is corrected to obtain the second road adhesion coefficient correction value corresponding to the left front wheel, that is, the second road adhesion coefficient correction value is = If the estimated road adhesion coefficient of the left front wheel meets the following conditions: , then the estimated value of the road adhesion coefficient is determined as the second road adhesion coefficient correction value, that is, the second road adhesion coefficient correction value corresponding to the left front wheel is = If the estimated road adhesion coefficient of the left front wheel meets the following conditions: , based on the second road adhesion coefficient reference upper limit value, tolerance coefficient and sensitivity coefficient, the estimated value of the road adhesion coefficient corresponding to the left front wheel is corrected to obtain the second road adhesion coefficient correction value corresponding to the left front wheel, that is, the second road adhesion coefficient correction value is = .

[0148] It can be understood that the second road adhesion coefficient correction value corresponding to the right front wheel can be determined by referring to the above method. , the second road adhesion coefficient correction value corresponding to the left rear wheel , the second road adhesion coefficient correction value corresponding to the right rear wheel , I will not go into details here.

[0149] As an example, according to formula (22), the estimated road adhesion coefficient value corresponding to the left front wheel of the vehicle can be corrected using the third road adhesion coefficient reference range value to obtain the third road adhesion coefficient correction value corresponding to the left front wheel of the vehicle.

[0150] (twenty two);

[0151] In formula (22), Indicates the third road adhesion coefficient correction value corresponding to the left front wheel of the vehicle; Indicates the reference lower limit value of the third road adhesion coefficient; Indicates the reference upper limit value of the third road adhesion coefficient; Indicates the estimated road adhesion coefficient corresponding to the left front wheel of the vehicle; represents the tolerance factor; Represents the sensitivity coefficient.

[0152] As an example, assuming that the third terrain probability value output by the third terrain recognition model determines that the road terrain of the current driving section is a dry asphalt road, then by querying the above Table 1, it can be determined that the reference range value of the third road adhesion coefficient is 0.8~1.0, that is, =0.8, =1.0. If the estimated road adhesion coefficient of the left front wheel meets the following conditions: , based on the third road adhesion coefficient reference lower limit value, tolerance coefficient and sensitivity coefficient, the estimated value of the road adhesion coefficient corresponding to the left front wheel is corrected to obtain the third road adhesion coefficient correction value corresponding to the left front wheel, that is, the third road adhesion coefficient correction value corresponding to the left front wheel is = If the estimated road adhesion coefficient of the left front wheel meets the following conditions: , then the estimated value of the road adhesion coefficient is determined as the third road adhesion coefficient correction value, that is, the third road adhesion coefficient correction value is = If the estimated road adhesion coefficient of the left front wheel meets the following conditions: , based on the third road adhesion coefficient reference upper limit value, tolerance coefficient and sensitivity coefficient, the estimated value of the road adhesion coefficient corresponding to the left front wheel is corrected to obtain the third road adhesion coefficient correction value corresponding to the left front wheel, that is, the third road adhesion coefficient correction value corresponding to the left front wheel is = .

[0153] It can be understood that the third road adhesion coefficient correction value corresponding to the right front wheel can be determined by referring to the above method. , the third road adhesion coefficient correction value corresponding to the left rear wheel , the third road adhesion coefficient correction value corresponding to the right rear wheel , I will not go into details here.

[0154] In some embodiments, based on the first road adhesion coefficient correction value, the second road adhesion coefficient correction value and the third road adhesion coefficient correction value corresponding to each wheel end, determining the road adhesion coefficient output value corresponding to each wheel end when the vehicle is traveling on the current driving section includes:

[0155] For each wheel end, determining a first weight value corresponding to a first road surface adhesion coefficient correction value of the wheel end, a second weight value corresponding to a second road surface adhesion coefficient correction value of the wheel end, and a third weight value corresponding to a third road surface adhesion coefficient correction value of the wheel end;

[0156] Based on the first weight value, the second weight value and the third weight value, the first road adhesion coefficient correction value, the second road adhesion coefficient correction value and the third road adhesion coefficient correction value of the wheel end are weighted to obtain the road adhesion coefficient output value corresponding to the wheel end.

[0157] As an example, the road adhesion coefficient output value corresponding to the left front wheel of the vehicle when it is traveling on the current driving section can be calculated according to formula (23).

[0158] (twenty three);

[0159] In formula (23), Indicates the road adhesion coefficient output value corresponding to the left front wheel when the vehicle is traveling on the current driving section; , , Respectively represent the first weight value, the second weight value and the third weight value; They respectively represent the first road adhesion coefficient correction value, the second road adhesion coefficient correction value and the third road adhesion coefficient correction value corresponding to the left front wheel.

[0160] It can be understood that the road adhesion coefficient output value corresponding to the right front wheel when the vehicle is traveling on the current driving section can be determined by referring to the above method. , the road adhesion coefficient output value corresponding to the left rear wheel , the road adhesion coefficient output value corresponding to the right rear wheel , I will not go into details here.

[0161] In some embodiments, determining a first weight value corresponding to a first road surface adhesion coefficient correction value of a wheel end, a second weight value corresponding to a second road surface adhesion coefficient correction value of a wheel end, and a third weight value corresponding to a third road surface adhesion coefficient correction value of a wheel end includes:

[0162] Obtaining a first initial weight value corresponding to a first terrain recognition model, a second initial weight value corresponding to a second terrain recognition model, and a third initial weight value corresponding to a third terrain recognition model; wherein the first terrain recognition model is used to perform multi-terrain recognition on current chassis data; the second terrain recognition model is used to perform multi-terrain recognition on a current road image; and the third terrain recognition model is used to perform multi-terrain recognition on current location information;

[0163] According to the preset test cycle, a chassis data test set, a road image test set and a position information test set of the vehicle during driving are obtained;

[0164] Using the chassis data test set to test the first terrain recognition model, obtaining a first test result;

[0165] Using the road image test set to test the second terrain recognition model, obtaining a second test result;

[0166] Sending the location information test set to the cloud, so that the cloud uses the location information test set to test the third terrain recognition model, obtain a third test result, and return the third test result;

[0167] Based on the first test result, the second test result and the third test result, the first initial weight value, the second initial weight value and the third initial weight value are fine-tuned to obtain the first weight value, the second weight value and the third weight value.

[0168] The preset test period can be flexibly set according to actual conditions, for example, it can be set to 1 week, 1 month, 3 months, etc., and this embodiment does not impose specific restrictions on this.

[0169] After the first terrain recognition model, the second terrain recognition model and the third terrain recognition model are trained, a first initial weight value corresponding to the first terrain recognition model, a second initial weight value corresponding to the second terrain recognition model, and a third initial weight value corresponding to the third terrain recognition model can be set according to actual model performance (such as model accuracy).

[0170] After loading, when the preset test cycle is reached, the chassis data test set, road image test set and location information test set of the vehicle during driving are re-collected, and then Calculate new , , ;in, They are respectively the old data set (i.e., the chassis data test set, road image test set, and position information test set collected in the previous test cycle) and the new data set (i.e., the chassis data test set, road image test set, and position information test set collected in the current test cycle).

[0171] Exemplarily, the first initial weight value, the second initial weight value, and the third initial weight value may be fine-tuned according to formula (24) to obtain the first weight value, the second weight value, and the third weight value.

[0172] (twenty four);

[0173] In formula (24), Indicates the current test cycle after fine-tuning. i Weight value; Indicates the first i Initial weight value, i =1,2,3; Indicates the firsti Update weight values; Indicates the first j Initial weight value, j =1,2,3; Indicates the first j Update weight values; Indicates the 1st to 2nd data sets determined based on the old data set collected in the previous test cycle. j The sum of the initial weight values; Indicates the first to the second data sets determined based on the new data sets collected in the current test cycle. j Update the sum of weight values; To adjust the parameters.

[0174] As an example, suppose that in the last test cycle, the first terrain recognition model is deduced using the chassis data set collected during driving, and the model accuracy of the first terrain recognition model is determined to be a(0≤a≤1) The second terrain recognition model is deduced using the road image set collected during the driving process, and the model accuracy of the second terrain recognition model is determined to be b (0≤b≤1) The third terrain recognition model is deduced using the vehicle position information set collected during driving, and the model accuracy of the third terrain recognition model is determined to be c(0≤c≤ 1） Then the first initial weight value can be set according to the model accuracy of the first terrain recognition model, the second terrain recognition model and the third terrain recognition model. (corresponding to the first terrain recognition model), the second initial weight value (corresponding to the second terrain recognition model) and the third initial weight value (corresponding to the third terrain recognition model) are = 、 = 、 = .

[0175] Similarly, in the current test cycle, the first initial weight value determined in the previous test cycle can be referred to. , the second initial weight value and the third initial weight value The first update weight value is determined by using the newly collected chassis data test set, road image test set and position information test set. (corresponding to the first terrain recognition model), the second update weight value (corresponding to the second terrain recognition model), the third update weight value (Corresponding to the third terrain recognition model).

[0176] Then, the first initial weight value , the second initial weight value , the third initial weight value , first update weight value , the second update weight value , Third update weight value Substitute into the above formula (24) to calculate the 1st to 2nd test cycle after fine-tuning. i Weight value.

[0177] By collecting a new data set of the vehicle during driving according to a preset test cycle, the first initial weight value , the second initial weight value , the third initial weight value Fine-tuning can improve the accuracy of model fusion, thereby improving the accuracy of identifying the vehicle's terrain during driving, and further improving the accuracy of identifying the road adhesion coefficient, and improving the accuracy and reliability of the controlled chassis following the control target.

[0178] All the above optional technical solutions can be arbitrarily combined to form optional embodiments of the present application, which will not be described one by one here.

[0179] See also Figure 1During the driving process of the vehicle 100, the vehicle controller 101 can obtain the current chassis data (including driver operation data, vehicle driving status data and vehicle model parameters) of the vehicle 100 when it is driving on the current driving section through the CAN interface, and input the current chassis data into the first terrain recognition model, and output the first terrain recognition probability to the terrain recognition fusion module; obtain the current road image collected by the on-board image sensor of the vehicle 100 through the image interface, and input the current road image into the second terrain recognition model, and output the second terrain recognition probability to the terrain recognition fusion module; transmit the current position information collected by the on-board GPS positioning system to the cloud 200 through the communication interface, and the cloud 200 returns the corresponding information of the current position information according to the high-precision map. Terrain information; the vehicle controller 101 inputs the terrain information returned by the cloud 200 into the third terrain recognition model, and outputs the third terrain recognition probability to the multi-terrain recognition fusion module; the multi-terrain recognition fusion module determines the road terrain corresponding to the current driving section according to the first, second and third terrain recognition probabilities, and queries the road adhesion coefficient reference range value corresponding to the road terrain according to the mapping relationship between the road terrain and the preset road adhesion coefficient reference range value, and then corrects the road adhesion coefficient estimation value corresponding to each wheel end according to the road adhesion coefficient reference range value, obtains the road adhesion coefficient output value corresponding to each wheel end when the vehicle is driving on the current driving section, and outputs it to the chassis control system 103 via the output interface, and outputs the road terrain to the instrument 102.

[0180] The difference between the driver's intention (mainly reflected in the driving operation behavior) and the vehicle's driving state reflects the current driving condition (such as the terrain condition of the vehicle) to a certain extent. The technical solution provided in the embodiment of the present application establishes a first terrain recognition model based on the vehicle chassis data while considering the driver's behavior and the vehicle's driving state; comprehensively considers the human-vehicle-road state when the vehicle is driving, and uses the current chassis data of the vehicle when driving (including the driver's operation data, the vehicle's driving state data and the vehicle model parameters) as the input of the first terrain recognition model; uses the current road image when the vehicle is driving as the input of the second terrain recognition model; and interacts with the cloud to obtain high-precision terrain. Figure 3The state information of the three dimensions is used as the input of the third terrain recognition model, and the multi-terrain recognition is performed on the current driving section through the first, second and third terrain recognition models respectively, and the first, second and third terrain probability values ​​are output; based on the current development of vehicle-mounted communication technology and computing power, a fusion recognition algorithm integrating the first, second and third terrain recognition models is proposed, and fusion recognition is performed based on the first, second and third terrain probability values ​​to determine the road terrain of the current driving section; after completing the multi-terrain recognition, the road adhesion coefficient estimation value is corrected using the road adhesion coefficient reference range value corresponding to the road terrain recognized by the first, second and third terrain recognition models to obtain a high-precision road adhesion coefficient output value, and then the vehicle control strategy can be adjusted and optimized based on the road terrain of the current driving section and the road adhesion coefficient output value, thereby improving the accuracy and reliability of the control target of the wire-controlled chassis following, so that the vehicle can better adapt to different driving environments and conditions, and provide the driver with a safer, more stable and comfortable driving experience.

[0181] In some embodiments, the multi-terrain recognition fusion module fuses the first, second, and third terrain probabilities output by the first, second, and third terrain recognition models, outputs the maximum terrain probability corresponding to the current driving section, and outputs the maximum terrain probability to the instrument 102 through the output interface.

[0182] As an example, assume that the first terrain probability output by the first terrain recognition model is ,in, ~ They respectively represent the probability values ​​of the road terrain corresponding to the numbers 1 to 10 in Table 1 for the current driving section. For example, Indicates the probability value that the current driving section is the dry asphalt road corresponding to number 1, It indicates the probability value that the current driving section is the wet asphalt road corresponding to number 2, ... and so on. It indicates the probability value that the current driving section is the grass corresponding to number 10. The second terrain probability output by the second terrain recognition model is ,in, ~ They respectively represent the probability values ​​of the road terrain corresponding to the numbers 1 to 10 in Table 1 for the current driving section. For example, Indicates the probability value that the current driving section is the dry asphalt road corresponding to number 1, It indicates the probability value that the current driving section is the wet asphalt road corresponding to number 2, ..., and so on. It indicates the probability value that the current driving section is the grass corresponding to number 10. The third terrain probability output by the third terrain recognition model is ,in, ~ They respectively represent the probability values ​​of the road terrain corresponding to the numbers 1 to 10 in Table 1 for the current driving section. For example, Indicates the probability value that the current driving section is the dry asphalt road corresponding to number 1, It indicates the probability value that the current driving section is the wet asphalt road corresponding to number 2, ... and so on. Indicates the probability value that the current driving section is the grass corresponding to number 10. After the multi-terrain recognition fusion module fuses the first, second, and third terrain probabilities output by the first, second, and third terrain recognition models, the maximum terrain probability output is ;in, are the first, second and third terrain probabilities output by the first, second and third terrain recognition models respectively, are the weight values ​​corresponding to the first, second, and third terrain recognition models after fine-tuning in the current test cycle, is the maximum terrain probability.

[0183] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.

[0184] Figure 3 Schematic diagram of a road adhesion coefficient determination device based on multi-terrain recognition provided by an embodiment of the present application. Figure 3 As shown, the road adhesion coefficient determination device 300 based on multi-terrain recognition includes:

[0185] The acquisition module 301 is configured to acquire the current chassis data, current road image and current position information of the vehicle on the current driving section;

[0186] The multi-terrain recognition module 302 is configured to perform multi-terrain recognition on the current driving section based on the current chassis data, the current road image and the current position information to determine a reference range value of the road adhesion coefficient corresponding to the current driving section;

[0187] The estimation module 303 is configured to determine the slip rate of the vehicle when it is traveling on the current driving section, and estimate the estimated value of the road adhesion coefficient corresponding to each wheel end of the vehicle when it is traveling on the current driving section based on the slip rate;

[0188] The output module 304 is configured to determine the road adhesion coefficient output value corresponding to each wheel end when the vehicle is traveling on the current driving section according to the road adhesion coefficient reference range value and the road adhesion coefficient estimation value.

[0189] In some embodiments, the road adhesion coefficient reference range value includes a first road adhesion coefficient reference range value, a second road adhesion coefficient reference range value, and a third road adhesion coefficient reference range value.

[0190] The above-mentioned multi-terrain recognition module 302 includes:

[0191] A first multi-terrain recognition unit is configured to perform multi-terrain recognition on a current driving section based on current chassis data to determine a first road adhesion coefficient reference range value corresponding to the current driving section; wherein the first road adhesion coefficient reference range value includes a first road adhesion coefficient reference lower limit value and a first road adhesion coefficient reference upper limit value;

[0192] A second multi-terrain recognition unit is configured to perform multi-terrain recognition on a current driving section based on a current road image to determine a second road adhesion coefficient reference range value corresponding to the current driving section; wherein the second road adhesion coefficient reference range value includes a second road adhesion coefficient reference lower limit value and a second road adhesion coefficient reference upper limit value;

[0193] The third multi-terrain recognition unit is configured to perform multi-terrain recognition on the current driving section based on the current position information to determine the third road adhesion coefficient reference range value corresponding to the current driving section; wherein the third road adhesion coefficient reference range value includes a third road adhesion coefficient reference lower limit value and a third road adhesion coefficient reference upper limit value.

[0194] In some embodiments, the first multi-terrain recognition unit includes:

[0195] An input component is configured to input current chassis data into a first terrain recognition model for multi-terrain recognition, and output a first terrain probability value corresponding to a current driving section;

[0196] A terrain determination component is configured to determine a road surface terrain corresponding to a current driving section based on the first terrain probability value;

[0197] The query component is configured to query a first road adhesion coefficient reference range value corresponding to the road surface terrain based on a preset mapping relationship between the road surface terrain and the road adhesion coefficient reference range value.

[0198] In some embodiments, the output module 304 includes:

[0199] A first correction unit is configured to correct the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section according to the first road adhesion coefficient reference range value, so as to obtain a first road adhesion coefficient correction value corresponding to each wheel end;

[0200] The second correction unit is configured to correct the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section according to the second road adhesion coefficient reference range value, so as to obtain a second road adhesion coefficient correction value corresponding to each wheel end;

[0201] A third correction unit is configured to correct the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section according to the third road adhesion coefficient reference range value, so as to obtain a third road adhesion coefficient correction value corresponding to each wheel end;

[0202] The determination unit is configured to determine the road adhesion coefficient output value corresponding to each wheel end when the vehicle is traveling on the current driving section based on the first road adhesion coefficient correction value, the second road adhesion coefficient correction value and the third road adhesion coefficient correction value corresponding to each wheel end.

[0203] In some embodiments, the first correction unit includes:

[0204] A first correction component is configured to correct the estimated road adhesion coefficient value corresponding to each wheel end based on the first road adhesion coefficient reference lower limit value, a tolerance coefficient and a sensitivity coefficient to obtain a first road adhesion coefficient correction value corresponding to the wheel end if the estimated road adhesion coefficient value corresponding to the wheel end is less than a first road adhesion coefficient reference lower limit value;

[0205] A second correction component is configured to determine the estimated road adhesion coefficient value corresponding to the wheel end as the first road adhesion coefficient correction value corresponding to the wheel end if the estimated road adhesion coefficient value corresponding to the wheel end is greater than or equal to the first road adhesion coefficient reference lower limit value and less than the first road adhesion coefficient reference upper limit value;

[0206] The third correction component is configured to correct the estimated value of the road adhesion coefficient corresponding to the wheel end based on the first road adhesion coefficient reference upper limit value, the tolerance coefficient and the sensitivity coefficient if the estimated value of the road adhesion coefficient corresponding to the wheel end is greater than or equal to the first road adhesion coefficient reference upper limit value, so as to obtain the first road adhesion coefficient correction value corresponding to the wheel end.

[0207] In some embodiments, the above-mentioned determining unit includes:

[0208] a weight determination component configured to determine, for each wheel end, a first weight value corresponding to a first road adhesion coefficient correction value of the wheel end, a second weight value corresponding to a second road adhesion coefficient correction value of the wheel end, and a third weight value corresponding to a third road adhesion coefficient correction value of the wheel end;

[0209] The coefficient output component is configured to weight the first road adhesion coefficient correction value, the second road adhesion coefficient correction value and the third road adhesion coefficient correction value of the wheel end based on the first weight value, the second weight value and the third weight value to obtain the road adhesion coefficient output value corresponding to the wheel end.

[0210] In some embodiments, the weight determination component may be specifically configured as follows:

[0211] Obtaining a first initial weight value corresponding to a first terrain recognition model, a second initial weight value corresponding to a second terrain recognition model, and a third initial weight value corresponding to a third terrain recognition model; wherein the first terrain recognition model is used to perform multi-terrain recognition on current chassis data; the second terrain recognition model is used to perform multi-terrain recognition on a current road image; and the third terrain recognition model is used to perform multi-terrain recognition on current location information;

[0212] According to the preset test cycle, a chassis data test set, a road image test set and a position information test set of the vehicle during driving are obtained;

[0213] Using the chassis data test set to test the first terrain recognition model, obtaining a first test result;

[0214] Using the road image test set to test the second terrain recognition model, obtaining a second test result;

[0215] Sending the location information test set to the cloud, so that the cloud uses the location information test set to test the third terrain recognition model, obtain a third test result, and return the third test result;

[0216] Based on the first test result, the second test result and the third test result, the first initial weight value, the second initial weight value and the third initial weight value are fine-tuned to obtain the first weight value, the second weight value and the third weight value.

[0217] In some embodiments, the estimation module 303 includes:

[0218] a longitudinal force determination unit configured to determine a longitudinal force of the vehicle when traveling on a current driving section based on a slip ratio and a tire characteristic parameter of the vehicle;

[0219] A load determination unit, configured to determine each wheel end load when the vehicle is traveling on a current driving section;

[0220] The adhesion coefficient determination unit is configured to determine, based on the longitudinal force and the loads on each wheel end, an estimated value of the road adhesion coefficient corresponding to each wheel end when the vehicle is traveling on a current driving section.

[0221] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0222] An embodiment of the present application also provides a vehicle, including a vehicle controller, an instrument and a chassis control system; the vehicle controller is connected to the instrument and the chassis control system respectively; the vehicle controller includes the road adhesion coefficient determination device based on multi-terrain recognition of the second aspect mentioned above.

[0223] Figure 4 Schematic diagram of an electronic device 400 provided in an embodiment of the present application. Figure 4 As shown, the electronic device 400 of this embodiment includes: a processor 401, a memory 402, and a computer program 403 stored in the memory 402 and executable on the processor 401. When the processor 401 executes the computer program 403, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor 401 executes the computer program 403, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0224] The electronic device 400 may be a desktop computer, a notebook, a PDA, a cloud server, or other electronic device. The electronic device 400 may include, but is not limited to, a processor 401 and a memory 402. Those skilled in the art will appreciate that Figure 4 The electronic device 400 is merely an example and does not limit the electronic device 400 . The electronic device 400 may include more or fewer components than those shown in the figure, or different components.

[0225] The processor 401 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0226] The memory 402 may be an internal storage unit of the electronic device 400, for example, a hard disk or memory of the electronic device 400. The memory 402 may also be an external storage device of the electronic device 400, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 400. The memory 402 may also include both an internal storage unit of the electronic device 400 and an external storage device. The memory 402 is used to store computer programs and other programs and data required by the electronic device.

[0227] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments 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 above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units.

[0228] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, and the computer program code may be in source code form, object code form, executable file or some intermediate form. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electric carrier signals and telecommunication signals.

[0229] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for determining road adhesion coefficient based on multi-terrain recognition, characterized in that: include: Obtain the current chassis data, current road image and current position information of the vehicle on the current driving section; Inputting the current chassis data into a first terrain recognition model for multi-terrain recognition to determine a first road surface adhesion coefficient reference range value corresponding to the current driving section; Inputting the current road image into a second terrain recognition model for multi-terrain recognition to determine a second road surface adhesion coefficient reference range value corresponding to the current driving section; Inputting terrain information corresponding to the current position information into a third terrain recognition model for multi-terrain recognition to determine a third road surface adhesion coefficient reference range value corresponding to the current driving section; Determining a slip rate when the vehicle is traveling on a current driving section, and estimating an estimated value of a road adhesion coefficient corresponding to each wheel end of the vehicle when the vehicle is traveling on the current driving section based on the slip rate; According to the first road adhesion coefficient reference range value, correcting the road adhesion coefficient estimation value corresponding to each wheel end when the vehicle is traveling on the current driving section, to obtain the first road adhesion coefficient correction value corresponding to each wheel end; According to the second road adhesion coefficient reference range value, correcting the road adhesion coefficient estimation value corresponding to each wheel end when the vehicle is traveling on the current driving section, to obtain the second road adhesion coefficient correction value corresponding to each wheel end; According to the third road adhesion coefficient reference range value, the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section is corrected to obtain the third road adhesion coefficient correction value corresponding to each wheel end; For each wheel end, determining a first weight value corresponding to a first road surface adhesion coefficient correction value of the wheel end, a second weight value corresponding to a second road surface adhesion coefficient correction value of the wheel end, and a third weight value corresponding to a third road surface adhesion coefficient correction value of the wheel end; Based on the first weight value, the second weight value and the third weight value, the first road adhesion coefficient correction value, the second road adhesion coefficient correction value and the third road adhesion coefficient correction value of the wheel end are weighted to obtain the road adhesion coefficient output value corresponding to the wheel end.

2. The method according to claim 1, characterized in that The first road adhesion coefficient reference range value includes a first road adhesion coefficient reference lower limit value and a first road adhesion coefficient reference upper limit value; The second road adhesion coefficient reference range value includes a second road adhesion coefficient reference lower limit value and a second road adhesion coefficient reference upper limit value; The third road adhesion coefficient reference range value includes a third road adhesion coefficient reference lower limit value and a third road adhesion coefficient reference upper limit value.

3. The method according to claim 2, characterized in that Inputting the current chassis data into a first terrain recognition model for multi-terrain recognition to determine a first road surface adhesion coefficient reference range value corresponding to the current driving section, including: Inputting the current chassis data into a first terrain recognition model to perform multi-terrain recognition, and outputting a first terrain probability value corresponding to the current driving section; Determining a road surface terrain corresponding to the current driving section based on the first terrain probability value; Based on a preset mapping relationship between a road surface topography and a road surface adhesion coefficient reference range value, a first road surface adhesion coefficient reference range value corresponding to the road surface topography is queried.

4. The method according to claim 2, characterized in that According to the first road adhesion coefficient reference range value, the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section is corrected to obtain the first road adhesion coefficient correction value corresponding to each wheel end, including: For each wheel end, if the estimated road adhesion coefficient value corresponding to the wheel end is less than the first road adhesion coefficient reference lower limit value, then based on the first road adhesion coefficient reference lower limit value, the tolerance coefficient and the sensitivity coefficient, the estimated road adhesion coefficient value corresponding to the wheel end is corrected to obtain the first road adhesion coefficient correction value corresponding to the wheel end; If the estimated value of the road adhesion coefficient corresponding to the wheel end is greater than or equal to the first road adhesion coefficient reference lower limit value and less than the first road adhesion coefficient reference upper limit value, then determining the estimated value of the road adhesion coefficient corresponding to the wheel end as the first road adhesion coefficient correction value corresponding to the wheel end; If the estimated value of the road adhesion coefficient corresponding to the wheel end is greater than or equal to the first road adhesion coefficient reference upper limit value, the estimated value of the road adhesion coefficient corresponding to the wheel end is corrected based on the first road adhesion coefficient reference upper limit value, the tolerance coefficient and the sensitivity coefficient to obtain the first road adhesion coefficient correction value corresponding to the wheel end.

5. The method according to claim 1, characterized in that: Determining a first weight value corresponding to a first road surface adhesion coefficient correction value of the wheel end, a second weight value corresponding to a second road surface adhesion coefficient correction value of the wheel end, and a third weight value corresponding to a third road surface adhesion coefficient correction value of the wheel end, comprises: Acquire a first initial weight value corresponding to a first terrain recognition model, a second initial weight value corresponding to a second terrain recognition model, and a third initial weight value corresponding to a third terrain recognition model; wherein the first terrain recognition model is used to perform multi-terrain recognition on current chassis data; the second terrain recognition model is used to perform multi-terrain recognition on a current road image; and the third terrain recognition model is used to perform multi-terrain recognition on current location information; According to a preset test cycle, a chassis data test set, a road image test set and a position information test set of the vehicle during driving are obtained; Using the chassis data test set to test the first terrain recognition model, to obtain a first test result; Using the road image test set to test the second terrain recognition model, to obtain a second test result; Sending the location information test set to the cloud, so that the cloud uses the location information test set to test the third terrain recognition model, obtain a third test result, and return the third test result; Based on the first test result, the second test result and the third test result, the first initial weight value, the second initial weight value and the third initial weight value are fine-tuned to obtain a first weight value, a second weight value and a third weight value.

6. The method according to claim 1, characterized in that The estimated value of the road adhesion coefficient corresponding to each wheel end of the vehicle when the vehicle is traveling on the current driving section is estimated based on the slip rate, including: Determining a longitudinal force of the vehicle when traveling on a current driving section based on the slip ratio and a tire characteristic parameter of the vehicle; Determining each wheel end load of the vehicle when traveling on a current driving section; Based on the longitudinal force and each wheel end load, an estimated value of a road adhesion coefficient corresponding to each wheel end when the vehicle is traveling on a current driving section is determined.

7. A road adhesion coefficient determination device based on multi-terrain recognition, characterized in that: include: An acquisition module is configured to acquire current chassis data, current road image and current position information of the vehicle on the current driving section; a multi-terrain recognition module, configured to input the current chassis data into a first terrain recognition model for multi-terrain recognition, so as to determine a first road surface adhesion coefficient reference range value corresponding to the current driving section; Inputting the current road image into a second terrain recognition model for multi-terrain recognition to determine a second road surface adhesion coefficient reference range value corresponding to the current driving section; Inputting terrain information corresponding to the current position information into a third terrain recognition model for multi-terrain recognition to determine a third road surface adhesion coefficient reference range value corresponding to the current driving section; an estimation module, configured to determine a slip rate when the vehicle is traveling on a current driving section, and estimate a road adhesion coefficient estimation value corresponding to each wheel end of the vehicle when the vehicle is traveling on the current driving section based on the slip rate; an output module configured to correct the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on a current driving section according to the first road adhesion coefficient reference range value, so as to obtain a first road adhesion coefficient correction value corresponding to each wheel end; According to the second road adhesion coefficient reference range value, the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section is corrected to obtain the second road adhesion coefficient correction value corresponding to each wheel end; according to the third road adhesion coefficient reference range value, the estimated road adhesion coefficient value corresponding to each wheel end when the vehicle is traveling on the current driving section is corrected to obtain the third road adhesion coefficient correction value corresponding to each wheel end; for each wheel end, determine a first weight value corresponding to the first road adhesion coefficient correction value of the wheel end, a second weight value corresponding to the second road adhesion coefficient correction value of the wheel end, and a third weight value corresponding to the third road adhesion coefficient correction value of the wheel end; based on the first weight value, the second weight value and the third weight value, weight the first road adhesion coefficient correction value, the second road adhesion coefficient correction value and the third road adhesion coefficient correction value of the wheel end to obtain the road adhesion coefficient output value corresponding to the wheel end.

8. A vehicle, characterized in that: It includes a vehicle controller, an instrument and a chassis control system; the vehicle controller is connected to the instrument and the chassis control system respectively; The vehicle controller includes the road adhesion coefficient determination device based on multi-terrain recognition as described in claim 7.

Images