Method, device, equipment, medium and product for estimating road adhesion coefficient

Abstract

This application discloses a method, apparatus, device, medium, and product for estimating road surface adhesion coefficient, relating to the field of driving safety technology. The method includes: obtaining road surface type and visual adhesion coefficient estimates based on road surface images and a pre-built visual recognition model; mapping the visual adhesion coefficient estimates to road condition parameters, and obtaining a model-predicted adhesion coefficient based on the road condition parameters and a tire dynamics model; when the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range, projecting the model-predicted adhesion coefficient back into the visual adhesion coefficient range to achieve state constraints on the model-predicted adhesion coefficient, wherein the visual adhesion coefficient range is determined based on the road surface type; fusing the visual adhesion coefficient estimates and the state-constrained model-predicted adhesion coefficient using a capacitive Kalman filter algorithm to obtain the road surface adhesion coefficient; enabling highly reliable prediction information to participate in subsequent fusion, thus improving the accuracy and rationality of road surface adhesion coefficient estimation.

Description

Technical Field

[0001] This application relates to the field of driving safety technology, and in particular to a method, apparatus, equipment, medium and product for estimating the road surface adhesion coefficient. Background Technology

[0002] The road surface adhesion coefficient is a crucial parameter for vehicle dynamics control and safe driving, directly affecting braking, steering, and stability control performance. In existing technologies, adhesion coefficient estimation methods based on tire dynamics models are easily affected by model parameter uncertainties and vehicle speed variations, while estimation methods based on visual perception are significantly influenced by factors such as lighting conditions, weather, and recognition errors.

[0003] Under complex operating conditions, methods relying solely on models or vision struggle to obtain stable and accurate estimates of the road adhesion coefficient. Therefore, to overcome the limitations of a single information source, visual information can be fused with dynamic models to calculate the adhesion coefficient. However, while existing technologies employ simple weighted fusion of visual information and model estimates, they often directly incorporate unreliable or distorted information, leading to decreased accuracy and increased volatility in the final estimation results. This makes it difficult to meet the high stability and reliability requirements of intelligent vehicle dynamics control for adhesion coefficient estimation. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, equipment, medium, and product for estimating the road surface adhesion coefficient, which can solve the problem that "unreliable or distorted information is often directly involved in the fusion, resulting in a decrease in the accuracy of the final estimation result and an increase in fluctuation, making it difficult to meet the high stability and high reliability requirements of intelligent vehicle dynamics control for adhesion coefficient estimation".

[0005] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a method for estimating the road surface adhesion coefficient, including: Acquire images of the road surface in front of the vehicle; Based on the road surface image and the pre-built visual recognition model, the road surface type and visual adhesion coefficient estimates are obtained; According to the preset mapping rules, the visual adhesion coefficient estimate is mapped to road condition parameters, and the model predicted adhesion coefficient is obtained based on the road condition parameters and the tire dynamics model. When the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range, the model-predicted adhesion coefficient is projected back into the visual adhesion coefficient range to achieve state constraints on the model-predicted adhesion coefficient, wherein the visual adhesion coefficient range is determined based on the road surface type. The visual adhesion coefficient estimate and the model-predicted adhesion coefficient after state constraints are fused using the capacitive Kalman filter algorithm to obtain the road adhesion coefficient of the road ahead.

[0006] In one embodiment, the step of projecting the model-predicted adhesion coefficient back into the visual adhesion coefficient range when the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range is calculated as follows: ; In the formula, Predict the adhesion coefficient for the model. This represents the lower limit of the visual adhesion coefficient range. This represents the upper limit of the visual adhesion coefficient range. Predict the adhesion coefficient for the constrained model.

[0007] In one embodiment, the method further includes: simultaneously correcting the prediction covariance in the capacitive Kalman filter algorithm during the process of the model predicting the adhesion coefficient state constraint; For the constrained model-predicted adhesion coefficient, the value of the predicted covariance is reduced according to the degree to which the model-predicted adhesion coefficient deviates from the lower limit or the upper limit of the visual adhesion coefficient interval.

[0008] In one embodiment, the method further includes: during the process of fusing the visual adhesion coefficient estimate and the model-predicted adhesion coefficient after state constraints, adaptively adjusting the observation noise covariance in the capacitive Kalman filter algorithm based on the visual recognition confidence of the visual adhesion coefficient estimate.

[0009] In one embodiment, the step of adaptively adjusting the observation noise covariance in the capacitive Kalman filter algorithm based on the visual recognition confidence level of the visual attachment coefficient estimate is calculated using the following expression: ; In the formula, This is the adjusted observation noise covariance matrix. The preset benchmark observation noise covariance matrix, The confidence level for the visual recognition. To prevent the denominator from being zero or an extremely small positive number with unstable values.

[0010] In one embodiment, the tire dynamics model is the LuGre tire dynamics model.

[0011] Secondly, this application also provides a device for estimating the road surface adhesion coefficient, comprising: The visual perception module acquires images of the road surface in front of the vehicle; The road surface condition recognition module obtains road surface type and visual adhesion coefficient estimates based on the road surface image and a pre-built visual recognition model. The tire dynamics prediction module maps the visual adhesion coefficient estimate to road condition parameters according to a preset mapping rule, and obtains the model-predicted adhesion coefficient based on the road condition parameters and the tire dynamics model. The state constraint module projects the model-predicted adhesion coefficient back into the visual adhesion coefficient range when the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range, thereby achieving state constraint on the model-predicted adhesion coefficient. The visual adhesion coefficient range is determined based on the road surface type. The adaptive fusion module fuses the visual adhesion coefficient estimate and the model-predicted adhesion coefficient after state constraints using a capacitive Kalman filter algorithm to obtain the road adhesion coefficient of the road surface ahead.

[0012] Thirdly, this application also provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method.

[0013] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the above-described method.

[0014] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0015] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a method for estimating the road surface adhesion coefficient. Based on road surface images and a pre-built visual recognition model, the method obtains road surface type and visual adhesion coefficient estimates. According to a preset mapping rule, the visual adhesion coefficient estimates are mapped to road condition parameters. Based on these road condition parameters and a tire dynamics model, a model-predicted adhesion coefficient is obtained. Visual prior information is introduced into the tire dynamics model, improving the model's predictive capabilities and enhancing its relevance and accuracy. When the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range, it is projected back into the visual adhesion coefficient range to constrain its state. The visual adhesion coefficient range is determined based on the road surface type. Forcing the model-predicted adhesion coefficient exceeding the visual adhesion coefficient range back into the allowable range effectively prevents the tire dynamics model from outputting estimation results that violate physical laws under complex conditions, further ensuring the accuracy of the model-predicted adhesion coefficient. A capacitive Kalman filter algorithm is used to fuse the visual adhesion coefficient estimates and the state-constrained model-predicted adhesion coefficients to obtain the road surface adhesion coefficient of the road ahead. This achieves complementary advantages between visual and model information, ultimately outputting a high-precision and highly stable road surface adhesion coefficient estimation result. Therefore, this application enables highly reliable prediction information to participate in subsequent fusion, thereby improving the accuracy and rationality of road surface adhesion coefficient estimation. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a method for estimating the road surface adhesion coefficient according to an embodiment of this application. Figure 2 This is a diagram illustrating the results of estimating the road adhesion coefficient of smooth dry asphalt using a method for estimating the road adhesion coefficient according to an embodiment of this application. Figure 3 This is a diagram showing the results of estimating the road surface adhesion coefficient on a fresh snow surface, according to an embodiment of this application. Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] See Figure 1 This application provides a method for estimating the road surface adhesion coefficient, comprising the following steps: S100: Acquires images of the road surface in front of the vehicle; S200: Based on road surface images and a pre-built visual recognition model, estimates of road surface type and visual adhesion coefficient are obtained; S300: Based on the preset mapping rules, the estimated visual adhesion coefficient is mapped to road condition parameters, and the model-predicted adhesion coefficient is obtained based on the road condition parameters and the tire dynamics model. S400: When the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range, the model-predicted adhesion coefficient is projected back into the visual adhesion coefficient range to achieve state constraints on the model-predicted adhesion coefficient. The visual adhesion coefficient range is determined based on the road surface type. S500: The visual adhesion coefficient estimate and the model-predicted adhesion coefficient after state constraints are fused by the capacitive Kalman filter algorithm to obtain the road adhesion coefficient of the road ahead.

[0021] In S100 and S200, onboard cameras capture images of the road surface in front of the vehicle. After preprocessing such as cropping and normalization, these images are input into a pre-built visual recognition model to identify the type of road surface in front of the vehicle, such as smooth dry asphalt, dry gravel, waterlogged concrete, and fresh snow, and output an estimated visual adhesion coefficient.

[0022] For example, the construction process of the visual recognition model mainly includes: collecting a large amount of image data containing various typical road surfaces, labeling each type of road surface data with its corresponding road surface type, and assigning a typical visual adhesion coefficient value to each road surface type based on prior knowledge or measured data to form a training dataset; using this training dataset to supervise the training of a multi-task deep learning model based on a convolutional neural network, enabling the model to learn the mapping relationship from road surface images to road surface types and visual adhesion coefficients; thus, after training, this multi-task deep learning model serves as a visual recognition module, capable of processing real-time collected images of the road surface in front of the vehicle and outputting the identified road surface type and estimated visual adhesion coefficient value.

[0023] In S300, this step defines the mapping rules between the visual adhesion coefficient estimate and road condition parameters, as shown in the following mapping table: Table 1. Mapping rules between estimated visual adhesion coefficient and road condition parameters

[0024] The table above exemplifies the mapping rules between visual adhesion coefficient estimates and road condition parameters. These mapping rules are pre-calibrated based on extensive experimental data or domain experience. Here, θ represents the visual adhesion coefficient estimate, and θ is a road condition parameter. The corresponding relationships are as follows: high-adhesion road surfaces correspond to larger values. Smaller values ​​and smaller θ values ​​correspond to lower adhesion pavement surfaces. The road condition parameters are correction coefficients that adjust the internal characteristics of the tire dynamics model. They quantify the influence of different road surface types on tire friction characteristics. In other words, the road condition parameters can be used to quantify the degree to which road surface conditions hinder the establishment of tire friction. The larger the θ value, the more difficult it is for the road surface to establish friction.

[0025] Taking smooth dry asphalt as an example: when the road surface type is identified as smooth dry asphalt and the visual adhesion coefficient estimate is 0.75, this type of road surface has high adhesion ability. By looking up the table, the visual adhesion coefficient estimate is mapped to the corresponding road condition parameter θ = 0.9.

[0026] It should be noted that the mapping table provided in this application is an illustrative example, which only lists the mapping relationship between some road surface types, estimated values ​​of visual adhesion coefficients and corresponding road condition parameters. In actual engineering applications, the mapping table can be expanded or refined through experimental calibration or data-driven methods to expand the coverage and adaptation accuracy of the mapping table.

[0027] Preferably, the tire dynamics model used in this application is the LuGre tire dynamics model. The LuGre tire dynamics model is a physical model used to accurately simulate the dynamic friction behavior between the tire and the road surface. It is mainly used to predict the coefficient of friction based on the vehicle's dynamic state, obtaining the model-predicted coefficient of friction. The inputs to the tire dynamics model include vehicle state parameters (such as tire relative velocity, angular velocity, and normal force) and road condition parameters, and the output is the model-predicted coefficient of friction. In other embodiments, other types of tire models may be used, and those skilled in the art can choose according to the actual situation.

[0028] This application transforms the visual adhesion coefficient estimate into road condition parameters of the tire dynamics model through a mapping relationship, allowing visual information to directly participate in the prediction process of the tire dynamics model. This mapping process introduces visual prior information into the tire dynamics model, giving the model's prediction results better road condition adaptability. Thus, visual information is no longer treated as an isolated external observation, but directly participates in the internal dynamic calculation process of the tire dynamics model.

[0029] In S400, this step introduces a visual adhesion coefficient range as a physical constraint boundary to perform real-time compliance checks on the model-predicted adhesion coefficient: when the model-predicted adhesion coefficient exceeds the reasonable range defined by the visual prior, it is projected back to the nearest interval boundary. This effectively suppresses non-physical estimates caused by model mismatch or dynamic disturbances, ensuring that the adhesion coefficient always remains within a reasonable range consistent with the current road surface type, thus guaranteeing the physical validity of the estimation results.

[0030] The visual adhesion coefficient range is a predefined, physically reasonable range determined based on the road surface type identified in the above steps. For example, if the identified road surface is smooth dry asphalt, the corresponding visual adhesion coefficient range is 0.65-0.85. When the model-predicted adhesion coefficient obtained through tire dynamics model estimation exceeds the above range, a projection method is used for correction.

[0031] Specifically, the step of projecting the model-predicted adhesion coefficient back into the visual adhesion coefficient range when the model predicts the adhesion coefficient exceeds the range of the visual adhesion coefficient is calculated as follows: ; In the formula, To predict the adhesion coefficient for the model, This represents the lower limit of the visual adhesion coefficient range. This represents the upper limit of the visual adhesion coefficient range. Predict the adhesion coefficient for the constrained model.

[0032] It should be noted that the state constraint projection rule expressed by this formula is: when the model predicts the adhesion coefficient... Below the lower limit of the visual adhesion coefficient range When, force it to be corrected to When the value exceeds the upper limit of the visual adhesion coefficient range, it is corrected to... If it is within the interval, it remains unchanged.

[0033] In S500, specifically, during the process of model prediction of adhesion coefficient state constraints, the prediction covariance in the volumetric Kalman filter algorithm is simultaneously corrected; for the constrained model prediction of adhesion coefficient, the value of the prediction covariance is reduced accordingly based on the degree to which the model prediction of adhesion coefficient deviates from the lower limit or upper limit of the visual adhesion coefficient interval.

[0034] For example, for the constrained model to predict the adhesion coefficient, the value of the prediction covariance is reduced according to the magnitude of its deviation from the lower or upper limit of the visual adhesion coefficient range. The larger the deviation, the greater the reduction. By reducing the covariance, the uncertainty is reduced, so that the subsequent fusion update step of the volumetric Kalman filter algorithm can be carried out within a reasonable covariance range.

[0035] Autonomous vehicles rely on precise sensor data to understand their surroundings and make decisions. The Cumulative Kalman Filter (CKF) algorithm has shown potential in handling vehicle dynamics modeling and sensor fusion, contributing to improved safety and reliability of autonomous driving systems. This application utilizes the prediction and update steps of the CKF algorithm to integrate model prediction information with visual observation information. Specifically, the CKF algorithm fuses the visual adhesion coefficient estimate with the state-constrained model prediction adhesion coefficient to obtain the road surface adhesion coefficient.

[0036] The capacitive Kalman filter uses the state prediction values ​​obtained by recursion from the tire dynamics model as prior information in the prediction step, and introduces the visual adhesion coefficient estimate output by the visual recognition model as the observation vector into the filtering framework in the update step. During the update process, the model prediction values ​​and visual observation values ​​are weighted and fused by calculating the Kalman gain, so that the final output posterior state estimate not only follows the evolution law of tire dynamics, but is also corrected in real time by visual perception information, thus realizing the organic integration of model prediction information and visual observation information.

[0037] In this embodiment of the application, during the process of fusing the visual adhesion coefficient estimate and the model prediction adhesion coefficient after state constraints, the observation noise covariance in the volumetric Kalman filter algorithm is adaptively adjusted according to the visual recognition confidence of the visual adhesion coefficient estimate.

[0038] Specifically, the step of adaptively adjusting the observation noise covariance in the capacitive Kalman filter algorithm based on the visual recognition confidence level of the visual adhesion coefficient estimate is expressed as follows: ; In the formula, This is the adjusted observation noise covariance matrix. The preset benchmark observation noise covariance matrix, For visual recognition confidence, To prevent the denominator from being zero or an extremely small positive number with unstable values.

[0039] This application utilizes the recognition confidence score output by the visual recognition model to adaptively adjust the observation noise covariance matrix in the capacitive Kalman filter algorithm: the higher the confidence score, the smaller the observation noise covariance value, and the greater the weight of visual observation information in the fusion; conversely, the weight is reduced. This mechanism enables the filter to dynamically balance the contributions of model prediction and visual observation, improving the robustness and accuracy of the fusion estimation under complex conditions.

[0040] The following section describes the workflow of the Adaptive Constrained Volumetric Kalman Filter (AC-CKF) algorithm using formulas: (1) Time update: ; ; ; ; ; In the formula, The volume point at the previous moment. This is the posterior estimate from the previous time step. To estimate the error covariance matrix a priori, For the basic volume point set, It is a unit vector. Let be the dimension of the state vector. For the volume point after propagation, This is a priori state estimate (predicted value). To estimate the error covariance matrix a priori, Let be the process noise covariance matrix.

[0041] (2) Measurement update: ; ; In the formula, The volume point after resampling. To predict the observed values, This is the observation volume point.

[0042] (3) State constraints: ; ; In the formula, This is a priori state estimate (predicted value). It is the set of prior state estimates.

[0043] (4) Adaptive noise conditioning: ; (5) Status update: ; ; ; ; ; In the formula, To observe the autocovariance matrix, Let be the cross-covariance matrix of the state and the observation. For Kalman gain, This is the posterior state estimate (filtered value). This is the actual measurement value at the current moment. For the posterior estimation of the error covariance matrix, This is the prior estimate of the error covariance matrix.

[0044] This application overcomes the problem of unstable estimation under complex road conditions by integrating visual perception information with tire dynamics models, achieving reliable estimation of the road adhesion coefficient. By constructing a dual mapping framework of visual perception and tire dynamics, the physical consistency and interpretability of visual perception results are effectively improved. Thus, visual prior information is introduced into the tire dynamics model, giving the model prediction results better road condition adaptability. Furthermore, the capacitive Kalman filter algorithm is used to further fuse the visual observation results and model prediction results at the state space level.

[0045] Meanwhile, an adaptive observation noise adjustment mechanism based on visual confidence enables the filter to dynamically adjust information weights, reducing the interference of low-quality visual information on the estimation results. By introducing state constraints and covariance correction mechanisms, the occurrence of non-physical adhesion coefficient estimation results is avoided, thereby improving the overall numerical stability and engineering applicability of the system.

[0046] Figure 2 and Figure 3 middle, The true value adhesion coefficient refers to the pre-defined actual road surface adhesion coefficient used as the evaluation benchmark. This is an estimate of the visual adhesion coefficient. To predict the adhesion coefficient for the model, The coefficient of adhesion of the road surface after fusion.

[0047] Based on the same inventive concept, this application also provides a device for estimating the road surface adhesion coefficient. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the road surface adhesion coefficient estimation device provided below can be found in the limitations of the road surface adhesion coefficient estimation method described above, and will not be repeated here.

[0048] This application provides a device for estimating the road surface adhesion coefficient, comprising: The visual perception module acquires images of the road surface in front of the vehicle; The road surface condition recognition module obtains road surface type and visual adhesion coefficient estimates based on road surface images and pre-built visual recognition models; The tire dynamics prediction module maps the visual adhesion coefficient estimate to road condition parameters according to the preset mapping rules, and obtains the model-predicted adhesion coefficient based on the road condition parameters and the tire dynamics model. The state constraint module projects the model-predicted adhesion coefficient back into the visual adhesion coefficient range when the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range, thereby achieving state constraints on the model-predicted adhesion coefficient. The visual adhesion coefficient range is determined based on the road surface type. The adaptive fusion module fuses the visual adhesion coefficient estimate and the model-predicted adhesion coefficient after state constraints using a capacitive Kalman filter algorithm to obtain the road adhesion coefficient of the road ahead.

[0049] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 4 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media to run. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection.

[0050] Those skilled in the art will understand that Figure 4The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0051] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0052] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0053] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0054] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0055] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0056] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0057] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0058] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for estimating the road surface adhesion coefficient, characterized in that, include: Acquire images of the road surface in front of the vehicle; Based on the road surface image and the pre-built visual recognition model, the road surface type and visual adhesion coefficient estimates are obtained; According to the preset mapping rules, the visual adhesion coefficient estimate is mapped to road condition parameters, and the model predicted adhesion coefficient is obtained based on the road condition parameters and the tire dynamics model. When the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range, the model-predicted adhesion coefficient is projected back into the visual adhesion coefficient range to achieve state constraints on the model-predicted adhesion coefficient, wherein the visual adhesion coefficient range is determined based on the road surface type. The visual adhesion coefficient estimate and the model-predicted adhesion coefficient after state constraints are fused using the capacitive Kalman filter algorithm to obtain the road adhesion coefficient of the road ahead.

2. The method for estimating the road surface adhesion coefficient according to claim 1, characterized in that, The step of projecting the model-predicted adhesion coefficient back into the visual adhesion coefficient range when the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range is calculated as follows: ； In the formula, Predict the adhesion coefficient for the model. This represents the lower limit of the visual adhesion coefficient range. This represents the upper limit of the visual adhesion coefficient range. Predict the adhesion coefficient for the constrained model.

3. The method for estimating the road surface adhesion coefficient according to claim 2, characterized in that, The method further includes: simultaneously correcting the prediction covariance in the capacitive Kalman filter algorithm during the process of the model predicting the state constraints of the adhesion coefficient; For the constrained model-predicted adhesion coefficient, the value of the predicted covariance is reduced according to the degree to which the model-predicted adhesion coefficient deviates from the lower limit or the upper limit of the visual adhesion coefficient interval.

4. The method for estimating the road surface adhesion coefficient according to claim 1, characterized in that, The method further includes: during the process of fusing the estimated visual adhesion coefficient and the model-predicted adhesion coefficient after state constraints, adaptively adjusting the observation noise covariance in the capacitive Kalman filter algorithm based on the visual recognition confidence of the estimated visual adhesion coefficient.

5. The method for estimating the road surface adhesion coefficient according to claim 4, characterized in that, The step of adaptively adjusting the observation noise covariance in the capacitive Kalman filter algorithm based on the visual recognition confidence level of the visual attachment coefficient estimate is calculated using the following expression: ； In the formula, This is the adjusted observation noise covariance matrix. The preset benchmark observation noise covariance matrix, The confidence level for the visual recognition. To prevent the denominator from being zero or an extremely small positive number with unstable values.

6. The method for estimating the road surface adhesion coefficient according to claim 1, characterized in that, The tire dynamics model used is the LuGre tire dynamics model.

7. A device for estimating the road surface adhesion coefficient, characterized in that, include: The visual perception module acquires images of the road surface in front of the vehicle; The road surface condition recognition module obtains road surface type and visual adhesion coefficient estimates based on the road surface image and a pre-built visual recognition model. The tire dynamics prediction module maps the visual adhesion coefficient estimate to road condition parameters according to a preset mapping rule, and obtains the model-predicted adhesion coefficient based on the road condition parameters and the tire dynamics model. The state constraint module projects the model-predicted adhesion coefficient back into the visual adhesion coefficient range when the model-predicted adhesion coefficient exceeds the visual adhesion coefficient range, thereby achieving state constraint on the model-predicted adhesion coefficient. The visual adhesion coefficient range is determined based on the road surface type. The adaptive fusion module fuses the visual adhesion coefficient estimate and the model-predicted adhesion coefficient after state constraints using a capacitive Kalman filter algorithm to obtain the road adhesion coefficient of the road surface ahead.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the method for estimating the road surface adhesion coefficient according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the method for estimating the road surface adhesion coefficient as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the method for estimating the road surface adhesion coefficient as described in any one of claims 1-6.

Images