Vehicle cooperative perception and fusion estimation method for over-the-horizon scene

Abstract

This invention presents a vehicle stability control method for vehicle-road cooperative perception and fusion estimation in beyond-line-of-sight scenarios, belonging to the interdisciplinary fields of intelligent transportation, vehicle control engineering, and automotive safety technology. It acquires vehicle dynamics data from vehicle-side sensors and estimates the dynamic road surface adhesion coefficient based on a seven-degree-of-freedom model and unscented Kalman filtering. After acquiring road surface images using roadside cameras, the MobileNetV3+ model identifies the road surface state and maps it to the visual road surface adhesion coefficient. A discrete-time Kalman filter is used to fuse the dynamic and visual estimation results, dynamically adjusting the fusion weights based on confidence levels. The fused road surface adhesion coefficient is transmitted to the vehicle-side controller. Upper and lower level controllers are designed based on sliding mode control and road surface adhesion coefficient recognition to achieve vehicle stability control under complex road conditions. This method effectively improves the accuracy and predictability of road surface adhesion coefficient estimation, enhancing vehicle handling stability under extreme conditions.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of intelligent transportation, vehicle control engineering and automotive safety technology, and in particular relates to a vehicle stability control method for vehicle-road cooperative perception and fusion estimation in beyond-line-of-sight scenarios. Background Technology

[0002] Traditional vehicle stability control systems typically rely on the vehicle's own sensors or onboard cameras to estimate the road surface adhesion coefficient and make control decisions based on this. However, these methods still have limitations.

[0003] Vehicle sensors struggle to anticipate road conditions ahead, especially in blind spots or complex road conditions, limiting their warning and control capabilities. Furthermore, there is an estimation lag; dynamic estimation depends on the vehicle's current state and exhibits a certain degree of lag, making feedforward control impossible.

[0004] Traditional control methods lack a coordination mechanism and fail to fully utilize the information exchange between roadside facilities and vehicles, thus failing to leverage the overall advantages of vehicle-road-cloud collaboration. They cannot provide more accurate and forward-looking estimates of the road adhesion coefficient for following vehicles by fusing multi-source information such as roadside cameras, preceding vehicle dynamics data, and cloud data processing, thereby improving the vehicle's handling stability under extreme conditions.

[0005] Therefore, a new technical solution is urgently needed in the existing technology to solve the above problems. Summary of the Invention

[0006] The technical problem to be solved by this invention is to provide a vehicle stability control method for vehicle-road cooperative perception and fusion estimation for beyond-line-of-sight scenarios, in order to solve the problem of lag and failure of handling stability control caused by the inability of vehicles to obtain the road surface adhesion coefficient in a timely and accurate manner in complex road environments, especially in blind spots or accident-prone curves.

[0007] A vehicle-road cooperative perception and fusion estimation method for vehicle stability control in beyond-line-of-sight scenarios is triggered in specific curved road sections determined by blind spots and historical accident data; it includes the following steps, which are performed sequentially:

[0008] Step 1: Obtain the vehicle dynamics road adhesion coefficient;

[0009] Step 2: Obtain the visual road surface adhesion coefficient from the roadside camera;

[0010] Step 3: Integrate visual road surface adhesion coefficient and dynamic road surface adhesion coefficient;

[0011] Step 4: Establish communication between the roadside workstation and the vehicle-side workstation;

[0012] Step 5: Assess the current risk of instability based on the road surface adhesion coefficient;

[0013] Step Six: Implement vehicle handling stability control strategies based on the assessment of the danger.

[0014] The method for calculating the vehicle dynamic adhesion coefficient in step one is as follows: Collect dynamic data of the vehicle during driving, establish a seven-degree-of-freedom model of the vehicle, obtain the longitudinal, lateral, and yaw motion equations, and the vertical load, slip angle, and slip ratio of the four wheels; obtain the normalized tire force using the Dugoff tire model, and use the unscented Kalman filter algorithm to obtain the estimated dynamic road adhesion coefficient μ. dyn .

[0015] The method for calculating the visual road surface adhesion coefficient of the roadside camera in step two is as follows: Acquire road surface image data in front of the vehicle and perform data preprocessing; input the preprocessed data into the lightweight convolutional neural network model MobileNetV3+, extract multi-level features using depthwise separable convolution and bottleneck structures, and output the road surface state category through a classifier; establish a lookup table of predefined road surface state categories and adhesion coefficient values, query and output the corresponding estimated visual road surface adhesion coefficient μ. vis The visual confidence C of this visual estimation is evaluated using the probability vector output by the final Softmax layer of the MobileNetV3+ model. vis .

[0016] The method for fusing visual road adhesion coefficient and dynamic road adhesion coefficient described in step three is as follows:

[0017] Step 1: Use a discrete-time Kalman filter with the adhesion coefficient μ as the state to establish the system model:

[0018] ,

[0019] In the formula, w k It is process noise, which follows a Gaussian distribution with zero mean and variance Q;

[0020] Step 2: Substitute two observation sources: the estimated dynamic road adhesion coefficient μ dyn and the estimated visual road surface adhesion coefficient μ vis ;

[0021] ,

[0022] In the formula, and It is observation noise. and The observed noises follow Gaussian distributions with mean of zero and variance of , respectively, and are independent of each other. The covariance matrix of the observed noise is:

[0023] ,

[0024] Step 3: Dynamically adjust the observation noise variance based on the confidence level. and ;

[0025] Dynamic observation noise variance:

[0026] ;

[0027] In the formula, is the confidence level of the dynamic estimation, and S is the vehicle state, where The function representing the calculation of the variance of dynamic observation noise is specifically derived from the variance of the fundamental noise. , dynamic confidence The reciprocal of the value, and the adjustment factor based on vehicle status. Multiplying them together yields the result;

[0028] Specifically, it is expressed as follows:

[0029] ,

[0030] Visual observation noise variance:

[0031] ,

[0032] In the formula: The fundamental noise variance of visual observations represents the inherent uncertainty of visual estimation under ideal conditions.

[0033] : The confidence level of the visual estimation, output by the visual recognition model;

[0034] Adjustment factors based on environmental condition E, which includes light intensity, weather conditions, road surface texture, camera occlusion, and other factors affecting the accuracy of visual recognition;

[0035] Based on road surface change signs The adjustment function, when When =1, it indicates that a change in road surface type has been detected, and the function reflects higher uncertainty by increasing the noise variance;

[0036] E: Environmental conditions, including lighting, weather, road surface texture, and visual perception-related factors;

[0037] Step 4: Confidence calculation, dynamic confidence. and visual confidence Calculated using the following formula:

[0038] ,

[0039] In the formula, and It is the confidence score calculation function;

[0040] Step 5: Road Surface Change Detection Detects changes in road surface type by detecting abrupt changes in visual observations:

[0041] ,

[0042] when Exceeding the threshold and When the value is large, it is determined to be a change in the road surface and triggers [the event / event]. Mark 1;

[0043] Step 6: Kalman filtering and final fusion result:

[0044] Prediction steps: ,

[0045] Update steps:

[0046] ① Calculate the Kalman gain:

[0047] ,

[0048] In the formula, ;

[0049] ②Status Update:

[0050] ,

[0051] In the formula, ;

[0052] ③ Covariance update:

[0053] ,

[0054] Filter status output This is the estimated value of the adhesion coefficient after fusion.

[0055] The method for establishing communication between the roadside workstation and the vehicle end in step four is as follows:

[0056] The roadside unit sends a data packet to the vehicle-side unit, including the fused adhesion coefficient estimate, the road segment location information corresponding to the fused adhesion coefficient estimate, the data validity timestamp, and the roadside confidence level of the coefficient.

[0057] The roadside unit includes a camera and a roadside computing workstation. The camera is used to acquire road surface images. The roadside computing workstation is configured to perform the following operations: recognize the visual road surface adhesion coefficient based on the image and fuse it with the dynamic road surface adhesion coefficient received from the vehicle.

[0058] The vehicle-mounted unit includes onboard sensors, a communication module, and a vehicle controller. The onboard sensors are used to estimate the dynamic road surface adhesion coefficient, the communication module is used to interact with the roadside unit, and the vehicle controller is configured to receive the fused road surface adhesion coefficient and execute a handling stability control strategy.

[0059] It also includes a cloud server that communicates with roadside units and vehicle-mounted units to store and distribute historical road surface data and fusion results.

[0060] Step five describes the determination of current instability risks based on the road surface adhesion coefficient, including wheel lock-up, sideslip caused by centrifugal force, and oversteering caused by the vehicle under certain driving conditions.

[0061] Step six describes the implementation of vehicle handling stability control strategies based on hazard assessment, which includes the design of an upper-level controller based on sliding mode control and a lower-level torque distribution method based on road surface adhesion coefficient identification.

[0062] The lower-layer torque distribution method based on road surface adhesion coefficient identification is as follows:

[0063] Let the total yaw moment T = T1 + T2 + T3,

[0064] In the formula: T1 = W1 ( F x -F xd ) 2 ,

[0065] T2 = W2 ( M z -M zd ) 2 ,

[0066] T3 = W3 [(F] x11 D x ) 112 + (F x12 D x ) 122 + (F x21 D x ) 212 + (F x22 D x ) 222 ],

[0067] F x =F x11cosδ 11 +F x12 cosδ 12 +F x21 cosδ 21 +F x22 cosδ 22 ,

[0068] M z = F x11 h1+F x12 h2+F x21 h3+F x22 h4,

[0069] In the formula:

[0070] T: Total yaw moment, generated by the torque distribution of each wheel;

[0071] T1 is the weighted squared term of the longitudinal force tracking error, representing the actual total longitudinal force F. x With the expected total longitudinal force F xd The deviation is used to optimize longitudinal force tracking;

[0072] T2: The weighted squared term of the yaw moment tracking error, representing the actual yaw moment M. z With the desired yaw moment M zd The deviation is used to optimize yaw moment tracking;

[0073] T3: Weighted sum of squares of adhesion utilization rate of each wheel, representing the longitudinal force of each wheel. With maximum longitudinal force The sum of squares of the ratios is used to optimize tire grip margin and prevent wheel slippage or lock-up.

[0074] M z The actual yaw moment is caused by the longitudinal force of each wheel. The lever arm h1 from the center of the vehicle is calculated;

[0075] M zd The desired yaw moment is calculated by the upper sliding mode controller based on the vehicle condition and the road surface adhesion coefficient.

[0076] F x The actual total longitudinal force is composed of the longitudinal forces of each wheel. and its steering angle The cosine is obtained by projecting it onto the longitudinal axis of the vehicle and summing the results.

[0077] F xd The expected total longitudinal force is calculated by the upper-level controller based on driver input and vehicle status.

[0078] δ 11: Steering angle of the left front wheel (the first wheel), used to calculate the longitudinal force projection and yaw moment;

[0079] W1: Longitudinal force weighting coefficient;

[0080] W2: Weighting coefficient for yaw moment;

[0081] W3: Weighting coefficient for four-wheel adhesion coefficient;

[0082] hi: The effective distance from the longitudinal force of each wheel to the center of each vehicle subjected to the force;

[0083] D xij The maximum longitudinal force that the ground can provide to each wheel can be calculated using the following formula:

[0084] D xij = μ xij F zij ,

[0085] Where: μ xij It is the maximum coefficient of friction between the road surface and the wheel, calculated based on the PAC tire model; F zij For the vertical loads of each wheel;

[0086] The local optimum is obtained using the Lagrange multiplier method. Assume the objective function is convex, i.e., the Hessian matrix is ​​positive definite. The expression for the Hessian matrix H of the objective function is:

[0087] ,

[0088] make = 0, we can get:

[0089] ,

[0090] In the formula:

[0091] : The turning angle of each wheel;

[0092] L i Distances from the center of mass to each axis;

[0093] A: Cosine matrix of steering angle, elements Used to transfer longitudinal forces from each wheel Projected onto the vehicle's longitudinal axis, calculate the total longitudinal force. ;

[0094] B: Yaw moment contribution matrix, elements ,in The distance from the center of mass to each axle is given by , and j represents the left and right position of the wheel. This is used to calculate the longitudinal force on each wheel and the yaw moment. Contributions;

[0095] D: Attachment utilization weight matrix, a diagonal matrix with elements of... ,in It is the maximum longitudinal force that the ground can provide to each wheel, which is used to weight the adhesion utilization rate of each wheel in the objective function;

[0096] and These are elements in matrices A and B, respectively.

[0097] Through the above design scheme, the present invention can bring the following beneficial effects: a vehicle-road cooperative perception and fusion estimation method for vehicle stability control in beyond-line-of-sight scenarios, which solves the problem of lag and failure of vehicle handling stability control caused by the inability of the vehicle to obtain the road surface adhesion coefficient in a timely and accurate manner in complex road environments, especially in blind spots or accident-prone curves. It effectively improves the accuracy and foresight of road surface adhesion coefficient estimation and enhances the handling stability of the vehicle under extreme conditions. Attached Figure Description

[0098] The present invention will be further described below with reference to the accompanying drawings and specific embodiments:

[0099] Figure 1 This is a schematic diagram illustrating the applicable scenarios of the present invention.

[0100] Figure 2 This is a schematic diagram illustrating the dynamic estimation of road surface adhesion coefficient according to the present invention.

[0101] Figure 3 This is a schematic diagram illustrating the visual estimation of road surface adhesion coefficient according to the present invention.

[0102] Figure 4 This is a schematic diagram of the process of fusing dynamics and visual estimation of road adhesion coefficient according to the present invention.

[0103] Figure 5 This is a schematic diagram of the overall process of the vehicle stability control method for vehicle-road cooperative perception and fusion estimation in beyond-line-of-sight scenarios according to the present invention.

[0104] Figure 6 This is a schematic diagram of the communication between the roadside workstation and the vehicle-side workstation of the present invention. Detailed Implementation

[0105] Vehicle-road cooperative perception and fusion estimation vehicle stability control methods for beyond-line-of-sight scenarios, such as Figures 1-6 As shown, it includes the following steps:

[0106] Step 1: Obtain the vehicle dynamic adhesion coefficient.

[0107] Accelerometers and gyroscopes are used to collect dynamic data of the vehicle during driving, including vehicle acceleration, speed, steering angle, lateral acceleration, yaw rate, wheel speed, and center of gravity sideslip angle.

[0108] The collected data is preprocessed by removing noise and filtering.

[0109] A seven-degree-of-freedom model of the vehicle was established, and the equations of motion for the vehicle's longitudinal, lateral, and yaw motions, as well as the calculation formulas for the vertical load, sideslip angle, and slip ratio of the four wheels, were obtained.

[0110] Vehicle longitudinal motion equation:

[0111] ,

[0112] Equation of lateral motion of the vehicle:

[0113] ,

[0114] The formulas for the vehicle's longitudinal and lateral yaw motion equations are as follows:

[0115] ,

[0116] In the formula: the subscripts fl, fr, rl, and rr in the letters represent the front left wheel, front right wheel, rear left wheel, and rear right wheel of the car, respectively. For longitudinal vehicle speed, For lateral vehicle speed, F x_ij For the longitudinal force of the wheel, F y_ij For the lateral force of the wheel, I z Let be the moment of inertia about the Z-axis. For the left front wheel steering angle, The right front wheel steering angle, For the horizontal swing angle, The front wheel track. The rear wheel track. This is the distance from the front axle to the center of gravity. This is the distance from the rear axle to the center of mass.

[0117] Wheel vertical load:

[0118] ,

[0119] In the formula: These are the longitudinal forces on the front left wheel, front right wheel, rear left wheel, and rear right wheel, respectively. For tire weight, Where L is the vehicle weight and L is the wheelbase. For the height of the vehicle's center of gravity, For the longitudinal acceleration of the vehicle, This refers to the vehicle's lateral acceleration.

[0120] Tire slip angle:

[0121] ,

[0122] In the formula: These are the tire slip angles for the front left wheel, front right wheel, rear left wheel, and rear right wheel, respectively. These are the longitudinal speeds of the front left wheel, front right wheel, rear left wheel, and rear right wheel, respectively. These are the lateral speeds of the front left wheel, front right wheel, rear left wheel, and rear right wheel, respectively.

[0123] Tire slip ratio:

[0124] ,

[0125] Establish a Dugoff tire model to obtain the normalized tire force;

[0126] ,

[0127] In the formula, F represents the road surface adhesion coefficient. z For vertical load, λ is the slip ratio, α is the tire slip angle, ε is the speed influence factor, and C x It is the longitudinal stiffness of each tire, C y It is the lateral stiffness of each tire.

[0128] The formula for tire normalization is as follows:

[0129] ,

[0130] In the formula, , This represents the tire's normalized longitudinal force and normalized lateral force.

[0131] Estimating the road adhesion coefficient μ by combining a seven-degree-of-freedom vehicle model and an unscented Kalman filter algorithm 1。

[0132] Set state variables , measurement variables The control quantity u(t) = [δ]; the corresponding prediction equation is:

[0133] ,

[0134] The observation equation is:

[0135] ,

[0136] in:

[0137]

[0138] Step 2: Obtain the visual road surface adhesion coefficient from the roadside camera.

[0139] As a roadside sensing device, the roadside camera acquires and analyzes images of the road surface ahead to achieve precise identification of the road surface condition and obtain an estimate of the visual road surface adhesion coefficient. The specific steps are as follows:

[0140] (1) Roadside image acquisition and preprocessing: High-definition cameras deployed at key locations on the road (such as curves, hilltops, and accident-prone sections) continuously acquire image data of the road surface in front of the vehicle at a preset sampling frequency. The acquired raw images are preprocessed, including grayscale conversion, contrast enhancement, Gaussian filtering to suppress noise, and regions of interest (ROIs) are divided to focus on the effective road surface area along the vehicle's trajectory.

[0141] (2) Road surface condition recognition based on the MobileNetV3+ model: The lightweight convolutional neural network model MobileNetV3+ is used as the core classifier. The preprocessed road surface image is input into the model, which utilizes its depthwise separable convolution and bottleneck structure to efficiently extract multi-level features. Finally, the classifier outputs the road surface condition category. The category is based on a predefined classification system, including but not limited to dry asphalt pavement - minor damage, dry asphalt pavement - severe damage, wet asphalt pavement - minor damage, wet asphalt pavement - severe damage, snow-covered pavement, and icy pavement.

[0142] (3) Visual pavement adhesion coefficient mapping: Establish a lookup table that maps predefined pavement condition categories to typical adhesion coefficient values. Based on the specific pavement condition category identified by the MobileNetV3+ model in step (2), query and output the corresponding nominal visual pavement adhesion coefficient estimate from this lookup table. For example, when identified as "icy road surface", the mapped... The value is 0.15; when identified as "dry asphalt pavement - minor damage", the mapped value is... The value is 0.85.

[0143] (4) Visual estimation confidence assessment: At the same time, the confidence of this visual estimation is assessed based on the probability vector output by the final Softmax layer of the MobileNetV3+ model. Confidence level The calculation formula is:

[0144] ,

[0145] Among them, P i This represents the probability value by which the model classifies the road surface image into the i-th category. The higher the value, the higher the credibility of the visual recognition result. This confidence level will be used for dynamic adjustment of the fusion weights in step three.

[0146] (5) Data output: The roadside calculation unit will output the calculated visual pavement adhesion coefficient estimate. and their corresponding confidence levels The information is sent to the relevant vehicles through the vehicle-road cooperative communication link described in step four.

[0147] Step 3: Integrate the visual road surface adhesion coefficient and the dynamic road surface adhesion coefficient.

[0148] Step 1: Use a discrete-time Kalman filter with the adhesion coefficient μ as the state to establish the system model:

[0149] ,

[0150] In the formula, w k It is process noise, which follows a Gaussian distribution with zero mean and variance Q;

[0151] Step 2: Substitute two observation sources: the estimated dynamic road adhesion coefficient μ dyn and the estimated visual road surface adhesion coefficient μ vis ;

[0152] ,

[0153] In the formula, and It is observation noise. and The observed noises follow Gaussian distributions with mean of zero and variance of , respectively, and are independent of each other. The covariance matrix of the observed noise is:

[0154] ,

[0155] Step 3: Dynamically adjust the observation noise variance based on the confidence level. and ;

[0156] Dynamic observation noise variance:

[0157] ;

[0158] In the formula, is the confidence level of the dynamic estimation, and S is the vehicle state, where The function representing the calculation of the variance of dynamic observation noise is specifically derived from the variance of the fundamental noise. , dynamic confidence The reciprocal of the value, and the adjustment factor based on vehicle status. Multiplying them together yields the result;

[0159] Specifically, it is expressed as follows:

[0160] ,

[0161] Visual observation noise variance:

[0162] ,

[0163] In the formula: The fundamental noise variance of visual observations represents the inherent uncertainty of visual estimation under ideal conditions.

[0164] : The confidence level of the visual estimation, output by the visual recognition model;

[0165] Adjustment factors based on environmental condition E, which includes light intensity, weather conditions, road surface texture, camera occlusion, and other factors affecting the accuracy of visual recognition;

[0166] Based on road surface change signs The adjustment function, when When =1, it indicates that a change in road surface type has been detected, and the function reflects higher uncertainty by increasing the noise variance;

[0167] E: Environmental conditions, including lighting, weather, road surface texture, and visual perception-related factors;

[0168] Step 4: Confidence calculation, dynamic confidence. and visual confidence Calculated using the following formula:

[0169] ,

[0170] In the formula, and It is the confidence score calculation function;

[0171] Step 5: Road Surface Change Detection Detects changes in road surface type by detecting abrupt changes in visual observations:

[0172] ,

[0173] when Exceeding the threshold and When the value is large, it is determined to be a change in the road surface and triggers [the event / event]. Mark 1;

[0174] Step 6: Kalman filtering and final fusion result:

[0175] Prediction steps: ,

[0176] Update steps:

[0177] ① Calculate the Kalman gain:

[0178] ,

[0179] In the formula, ;

[0180] ②Status Update:

[0181] ,

[0182] In the formula, ;

[0183] ③ Covariance update:

[0184] ,

[0185] The final fusion result is the filter's state output. This is the estimated value of the adhesion coefficient after fusion.

[0186] Step 4: Establish communication between the roadside workstation and the vehicle.

[0187] Intelligent connected systems can be mainly divided into three information modules: "vehicle-road-cloud". Through information interaction among all parties, the coordination of vehicles, roads and cloud is realized, and the estimated value of road surface adhesion coefficient is ultimately transmitted.

[0188] The data packet sent from the roadside unit to the vehicle-side unit includes at least: the fused road surface adhesion coefficient value, the road segment location information corresponding to the coefficient, the valid timestamp of the data, and the confidence level of the roadside unit for the coefficient.

[0189] The roadside unit includes at least one camera and a roadside computing workstation. The camera is used to acquire road surface images, and the roadside computing workstation is configured to perform the following operations: identify the visual road surface adhesion coefficient based on the image and fuse it with the dynamic road surface adhesion coefficient received from the vehicle.

[0190] The vehicle-side unit includes onboard sensors, a communication module, and a vehicle controller. The onboard sensors are used to estimate the dynamic road surface adhesion coefficient, the communication module is used to interact with the roadside unit, and the vehicle controller is configured to receive the fused road surface adhesion coefficient and execute a handling stability control strategy based on it.

[0191] The cloud server communicates with the roadside unit and vehicle-mounted unit to store and distribute historical road surface data and fusion results.

[0192] Step 5: Determine the current risk of instability based on the road surface adhesion coefficient.

[0193] (1) Wheel lock-up. According to the single-wheel model, to prevent wheel lock-up, the braking force of the brakes must be less than the frictional force exerted on the wheels by the ground:

[0194] ,

[0195] To prevent wheel lock-up, the braking force of the brakes must be less than the frictional force exerted on the wheels by the ground.

[0196] (2) Centrifugal force causes sideslip. The vehicle travels at a speed V x Moving on a circle of radius r, once the centrifugal force exceeds the lateral force F provided by the tire... y This will cause the vehicle to become unstable.

[0197] (3) Oversteer caused by certain driving conditions. Understeer vehicles have a certain degree of high-speed stability, but under normal circumstances, the vehicle tends to veer inward when braking in a curve. The reason for this tendency is that when the vehicle brakes, the load on the rear wheels is transferred to the front wheels to a greater extent, making the forward torque greater than the reverse torque.

[0198] Step Six: Implement vehicle handling stability control strategies based on the assessment of the danger.

[0199] (1) Design of Upper-Level Controller Based on Sliding Mode Control According to vehicle dynamics theory, the formula for calculating the total longitudinal resultant force ∑ Fx of the vehicle is as follows:

[0200] ,

[0201] In the formula: γ is the longitudinal acceleration of the vehicle; γ is the yaw rate; v is the lateral speed; and m is the mass of the vehicle.

[0202] Longitudinal speed sliding mode controller sliding surface change rate It can be represented by a function with a saturation function:

[0203] ,

[0204] In the formula: For the longitudinal vehicle speed sliding mode controller sliding surface, H v = u -u d Where u is the actual longitudinal vehicle speed, u d For the desired longitudinal speed; sat( H v ) is a saturation function, and

[0205] ,

[0206] In the formula: v represents the boundary layer thickness designed for the longitudinal vehicle speed sliding mode controller; C v ε v This is the convergence law coefficient.

[0207] The desired longitudinal resultant force ∑ F is obtained xd The expression is:

[0208] ∑ F xd = m[-ε v sat( H v -C v H v -γ v + u(·) d ],

[0209] In the formula: u(·) d The desired longitudinal acceleration of the vehicle.

[0210] Yaw angular velocity sliding mode controller

[0211] According to vehicle dynamics theory, the total yaw force of the vehicle can be obtained.

[0212] The expression for ∑ M is:

[0213] ∑ M = I z γ

[0214] In the formula I z Let H(·) be the moment of inertia about the Z-axis at the vehicle's center of mass. The yaw rate is the rate of change of the sliding surface of the sliding mode controller. γ It can be expressed as a function with a saturation function as follows:

[0215] H(·) γ = -ε γ sat( H γ -k γ H γ

[0216] In the formula H γ H is the sliding surface for the yaw rate sliding mode controller. γ = γ -γ d Where γ is the actual yaw rate, γ d For the desired yaw rate; sat( H γ ) is a saturation function, and

[0217] ,

[0218] In the formula: k γ ε γ C is the coefficient of the reaching law; satγ is the boundary layer thickness designed for the yaw rate sliding mode controller.

[0219] The expression for the desired total yaw moment ∑ M is:

[0220] ∑ M = I z [-ε] γ sat( H γ -k γ H γ +γ(·) d ]

[0221] In the formula: γ(·) d Let be the desired rate of change of yaw rate.

[0222] Lower-layer torque distribution method based on road surface adhesion coefficient identification

[0223] T = T1 + T2 + T3

[0224] In the formula: T1 = W1 ( F x -F xd ) 2 ,

[0225] T2 = W2 ( M z -M zd ) 2 ,

[0226] T3 = W3 [(F] x11 D x ) 112 + (F x12 D x ) 122 + (F x21 D x ) 212 + (F x22 D x ) 222 ],

[0227] F x =F x11 cosδ 11 +F x12 cosδ 12 +F x21 cosδ 21 +F x22 cosδ 22 ,

[0228] M z = F x11 h1+F x12 h2+F x21 h3+Fx22 h4,

[0229] In the formula: T is the total yaw moment; W1, W2, and W3 are the weighting coefficients for longitudinal force, yaw moment, and four-wheel adhesion coefficient, respectively; F xij D represents the longitudinal force to be distributed to each wheel; hi is the effective distance from the longitudinal force of each wheel to the center of the vehicle on which the force is applied; xij The formula for determining the maximum longitudinal force that the ground can provide to each wheel is:

[0230] D xij = μ xij F zij

[0231] Where: μ xij It is the maximum coefficient of friction between the road surface and the wheel, calculated based on the PAC tire model; F zij The vertical load is for each wheel.

[0232] The solution obtained using the Lagrange multiplier method is the local optimum. To obtain the global optimum, we need to assume that the objective function is convex, i.e., the Hessian matrix is ​​a positive definite matrix. The expression for the Hessian matrix H of the objective function is:

[0233] ,

[0234] make = 0, we can get,

[0235] ,

[0236] In the formula: L represents the turning angle of each wheel. i This represents the distance from the center of mass to each axis.

Claims

1. A vehicle-road cooperative perception and fusion estimation vehicle stability control method for over-the-horizon scenes, triggered at specific curved road sections determined by visual blind areas and accident history data; characterized in that: Comprising the following steps, And the following steps are sequentially carried out: Step one: obtain the vehicle dynamics road adhesion coefficient; Step two: obtain the visual road adhesion coefficient of the road end camera; Step three: fuse the visual road adhesion coefficient and the dynamics road adhesion coefficient; Step four: the road end workstation establishes communication contact with the vehicle end; Step five: judge the current instability risk based on the road adhesion coefficient; Step six: execute the vehicle handling stability control strategy according to the risk judgment; The method for fusing the visual road adhesion coefficient and the dynamics road adhesion coefficient in step three is: Step 1: use a discrete-time Kalman filter with state as adhesion coefficient μ to establish system model: ， where w k is process noise, subject to a Gaussian distribution with zero mean and variance Q; Step 2: Substitute in two observation sources: estimated dynamic road adhesion coefficient μ dyn and estimated visual road adhesion coefficient μ vis ; ， wherein and is the observation noise, σ2 dynand σ2 visrespectively follow a Gaussian distribution with zero mean and variance σ2 dynand σ2 vis, respectively, and are independent of each other, and the covariance matrix of the observation noise is: ， Step 3: dynamically adjust observation noise variance σ2 dyn and σ2 vis according to confidence level; Dynamics observation noise variance: ； wherein is a confidence of the dynamics estimate, S is a vehicle state, wherein denotes a computation function of the dynamics observation noise variance, specifically resulting from a base noise variance , an inverse of the dynamics confidence , and an adjustment factor based on the vehicle state multiplication; Specifically expressed as: ， Visual observation noise variance: ， In the formula: : the variance of the noise underlying visual observations, representing the inherent uncertainty of visual estimates under ideal conditions; : confidence of visual estimation, output by visual recognition model; : adjustment factor based on environmental conditions E, including light intensity, weather conditions, road texture, camera occlusion, and factors affecting visual recognition accuracy; Based on road surface change signs The adjustment function, when When =1, it indicates that a change in road surface type has been detected, and the function reflects higher uncertainty by increasing the noise variance; E: environmental conditions, including light, weather, road texture and visual perception related factors; Step 4: Confidence calculation, Kinetic Confidence and Visual Confidence calculated by the following equation: , In the formula, and is a confidence calculation function; Step 5: road change detection detects road type change through sudden change of visual observation value: , When exceeds the threshold and is large, it is determined that the road surface changes, triggering flag 1; Step 6: Kalman filtering and final fusion result: Prediction step: , Update step: ① Calculate Kalman gain: ， In the formulae, ; ② State update: ， In the formulae, ; ③ Covariance update: ， State output of filter is the fused estimate of the attachment coefficient.

2. The method of claim 1, wherein the method is used for over-the-horizon scene vehicle-road cooperative perception and fusion estimation vehicle stability control. The vehicle dynamics adhesion coefficient calculation method in step one is: collecting vehicle dynamics data in driving, establishing a seven-degree-of-freedom model of the vehicle, obtaining longitudinal, lateral and yaw motion equations, four wheel vertical loads, side slip angles and slip rates; through the Dugoff tire model, obtaining the normalized tire force, and using the unscented Kalman filter algorithm to obtain the estimated dynamics road adhesion coefficient μ dyn .

3. The method of claim 1, wherein the method further comprises: determining a target speed of the ego vehicle based on the target speed of the ego vehicle and the target speed of the target vehicle; and determining a target acceleration of the ego vehicle based on the target acceleration of the ego vehicle and the target acceleration of the target vehicle. The method for calculating the visual road surface adhesion coefficient of the road end camera in step two is: collecting road surface image data in front of the vehicle, performing data preprocessing; inputting the preprocessed data into a lightweight convolutional neural network model MobileNetV3+, extracting multi-level features by using depth separable convolution and bottleneck structure, and outputting the road surface state category through a classifier; establishing a mapping relationship lookup table between the predefined road surface state category and the adhesion coefficient value, and querying and outputting the corresponding estimated visual road surface adhesion coefficient μ vis The visual confidence C vis of the visual estimation is evaluated through the probability vector output by the final Softmax layer of the MobileNetV3+ model.

4. The method of claim 1, wherein the method further comprises: The method for the road end workstation to establish communication contact with the vehicle end in step four is: The road end unit sends data packets to the vehicle end unit, including the fused adhesion coefficient estimate value, the road section position information corresponding to the fused adhesion coefficient estimate value, the data effective timestamp, and the confidence level of the road end to the coefficient; The road end unit includes a camera and a road end computing workstation, and the camera is used to collect road images; The road end computing workstation is configured to perform the following operations: identify the visual road adhesion coefficient based on the image, and fuse it with the dynamics road adhesion coefficient received from the vehicle; The vehicle end unit includes vehicle-mounted sensors, a communication module and a vehicle controller, the vehicle-mounted sensors are used to estimate the dynamics road adhesion coefficient, the communication module is used for data interaction with the road end unit, and the vehicle controller is configured to receive the fused road adhesion coefficient and execute the handling stability control strategy.

5. The method of claim 4, wherein the method further comprises: It also includes a cloud server in communication connection with the road end unit and the vehicle end unit, used for storing and distributing historical road data and fusion results.

6. The method of claim 1, wherein the method further comprises: The judgment of current instability risk based on the road adhesion coefficient in step five includes wheel lock, centrifugal force induced side slip and excessive steering caused by the vehicle under certain driving conditions.

7. The method of claim 1, wherein the method further comprises: The execution of the vehicle handling stability control strategy according to the risk judgment in step six includes the design of the upper controller based on sliding mode control and the lower torque distribution method based on road adhesion coefficient identification.

8. The method of claim 7, wherein the method further comprises: The lower torque distribution method based on road adhesion coefficient identification is: Let total yaw moment T = T1 + T2+ T3, where: T1 = W1 ( F x - F xd ) 2 , T2 = W2 ( M z -M zd ) 2 , T3 = W3 [ ( F x11 D x ) 112 + ( F x12 D x ) 122 + ( F x21 D x ) 212 + ( F x22 D x ) 222 ], F x =F x11 cosδ 11 +F x12 cosδ 12 +F x21 cosδ 21 +F x22 cosδ 22 , M z = F x11 h1+F x12 h2+F x21 h3+F x22 h4, In the formula: T: total yaw moment, generated by each wheel torque distribution; T1 weighted square term of the longitudinal force tracking error, representing the deviation of the actual total longitudinal force F x from the desired total longitudinal force F xd for optimizing the longitudinal force tracking; T2: weighted square term of yaw moment tracking error, representing deviation of actual yaw moment M z from desired yaw moment M zd for optimizing yaw moment tracking; T3: Weighted sum of square of each wheel adhesion utilization, representing longitudinal force of each wheel to the maximum longitudinal force ratio, for optimizing tire adhesion margin and preventing wheel slip or lockup; M z : actual yaw moment, calculated from the longitudinal forces of the individual wheels and their lever arms hi to the center of the vehicle ; M zd : desired yaw moment, calculated by the upper layer sliding mode controller according to vehicle states and road adhesion coefficient; F x : actual total longitudinal force, summed from the longitudinal forces of the individual wheels and the cosine of their steering angle projected onto the longitudinal vehicle axis F xd : desired total longitudinal force, calculated by the upper controller from driver input and vehicle state; δ 11 : steering angle of the front left wheel, used to calculate the longitudinal force projection and the yaw moment; W1: longitudinal force weight coefficient; W2: yaw moment weight coefficient; W3: four-wheel adhesion coefficient weight coefficient; hi: the effective distance from each wheel longitudinal force to the center of force of each vehicle; D xij : The maximum longitudinal force that the ground can provide to each wheel is calculated using the following formula: D xij = μ xij F zij , where: μ xij is the maximum friction factor between the road and the wheel calculated according to the PAC tire model; F zij is the vertical load of each wheel; The local optimal solution is obtained by the Lagrange multiplier method, assuming that the objective function is a convex function, i.e. the Hessian matrix is positive definite; the Hessian matrix H of the objective function is expressed as: ， Let = 0, we have: ， In the formula: : wheel angle; L i : centroid-to-axis distance; A: Cosine matrix of the steering angle, elements , for projecting the individual wheel longitudinal forces onto the vehicle longitudinal axis direction, calculating the total longitudinal force ; B: yaw moment contribution matrix, element where is the distance from the center of mass to each axle, j indicates the left or right position of the wheel, and is used to calculate the contribution of each wheel's longitudinal force to the yaw moment ​ D: the adhesion utilization rate weight matrix, is a diagonal matrix, and the elements are wherein is the maximum longitudinal force that the ground can provide to each wheel, used to weight the adhesion utilization rate of each wheel in the objective function; and are elements in matrices A and B, respectively.

Images