A vehicle lateral assistance control method considering driver fatigue

Abstract

The application discloses a vehicle lateral auxiliary control method considering driver fatigue, and comprises the following steps: collecting inherent parameters of a vehicle and real-time parameters of the vehicle; establishing a vehicle lateral dynamics and kinematics model and a driver steering model by using the collected parameters; establishing a global man-vehicle-road system fuzzy model by using the established model; designing a fast self-adaptive observer according to the global man-vehicle-road system fuzzy model to obtain an estimated value of a driver torque; representing a driver fatigue parameter according to the estimated value of the driver torque in combination with an ideal value of the driver torque; designing an adaptive auxiliary controller according to the driver fatigue parameter; and performing auxiliary control on the vehicle according to the adaptive auxiliary controller. The interval two-type T-S fuzzy model method is adopted to describe the uncertainty of a membership function, the observer is used to represent the driver fatigue, the auxiliary control of the vehicle with the driver fatigue is realized, and the safety and stability of driving are improved.

Description

Technical Field

[0001] This invention belongs to the field of vehicle stability control technology and relates to a vehicle lateral auxiliary control method that takes driver fatigue into account. Background Technology

[0002] With the rapid development of mobile communication technology, artificial intelligence, and other technologies, autonomous driving technology has made tremendous progress. However, due to technological limitations, road infrastructure, and traffic environment issues, fully autonomous driving remains difficult to achieve in the short term. Vehicle assistance control systems, by providing auxiliary torque to help drivers suppress lane departure tendencies, play a crucial role in improving vehicle stability and safety. Therefore, in recent years, research on driver assistance systems has attracted increasing attention from scholars.

[0003] Fatigue driving can lead to driver errors, seriously affecting road safety. The requirements for driver assistance systems differ depending on the driver's level of fatigue. When a driver is fatigued, from a vehicle safety perspective, the driver assistance control system needs to reduce the driver's authority and increase the level of assistance from the controller. Conversely, when the driver is alert, the level of assistance from the controller needs to be reduced. Traditional fatigue detection methods often rely on facial expressions and other features, requiring external equipment to detect the driver, which significantly impacts the driving experience. Therefore, there is an urgent need to develop a driver fatigue assistance control method that does not require external equipment to acquire facial expressions and other features to characterize driver fatigue. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this invention is to provide a vehicle lateral assist control method that takes driver fatigue into account.

[0005] This invention provides a vehicle lateral assist control method that takes driver fatigue into account, comprising the following steps:

[0006] S1: Collect the vehicle's inherent parameters and real-time parameters;

[0007] S2: Use the parameters collected by S1 to establish the vehicle's lateral dynamics and kinematics model, as well as the driver's steering model;

[0008] S3: Using the model established in S2, establish a global human-vehicle-road system model and perform fuzzy processing to obtain a global human-vehicle-road system fuzzy model;

[0009] S4: Based on the global human-vehicle-road system fuzzy model, design a fast adaptive observer to obtain an estimate of the driver's torque;

[0010] S5: Based on the estimated value of the driver's torque and combined with the ideal value of the driver's torque, characterize the driver's fatigue parameters;

[0011] S6: Design an adaptive auxiliary controller based on driver fatigue parameters;

[0012] S7: Provides auxiliary control for the vehicle based on the adaptive assistance controller.

[0013] The lateral assist control method for automobiles that takes driver fatigue into account, as described in this invention, has the following beneficial effects:

[0014] 1. This invention uses the interval type II TS fuzzy model method to deal with the uncertainty caused by tire nonlinearity and vehicle longitudinal speed variation, and obtains a more effective control effect of the vehicle nonlinear dynamics system, thereby improving the vehicle's steering stability.

[0015] 2. This invention uses a fast adaptive observer to estimate the driver's steering torque, which effectively improves the speed and accuracy of estimating time-varying driver torque, thereby enabling a more effective characterization of driver fatigue characteristics.

[0016] 3. The present invention uses an adaptive controller, which effectively provides auxiliary steering torque and improves the safety and stability of the vehicle. Attached Figure Description

[0017] Figure 1 This is a flowchart of a vehicle lateral assist control method that takes driver fatigue into account, according to the present invention.

[0018] Figure 2 This is a vehicle lateral dynamics and kinematics model according to an embodiment of the present invention;

[0019] Figure 3 This is a driver steering model according to an embodiment of the present invention. Detailed Implementation

[0020] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings.

[0021] like Figure 1 As shown, a lateral assist control method for automobiles that takes driver fatigue into account, according to the present invention, includes the following steps:

[0022] S1: Collects the vehicle's inherent parameters and real-time parameters.

[0023] The inherent parameters of the vehicle include: the total mass of the vehicle (m), and the distance from the front axle to the center of the vehicle (l). a The distance l from the rear axle to the center of the car b The moment of inertia of the vehicle body I z lateral stiffness C of the front wheel a And the lateral stiffness C of the rear wheel b .

[0024] The real-time parameters of the vehicle include: front wheel steering angle δ e longitudinal speed of the car v x lateral speed v of the car y and yaw rate

[0025] S2: Using the parameters collected in S1, establish the vehicle's lateral dynamics and kinematics model, as well as the driver's steering model, specifically:

[0026] S2.1: Establish the lateral dynamics and kinematics model of the vehicle, such as Figure 2 As shown. Assuming the left and right wheels are symmetrical, CG represents the car's center of gravity, and assuming the front wheel steering angle and sideslip angle are small, neglecting the vehicle's longitudinal dynamics and aerodynamics, the following lateral dynamics and kinematic model of the car is established:

[0027]

[0028] Among them, F ya F is the lateral force on the front wheel. yb The force is the lateral force on the rear wheel, m is the total mass of the car, and v is the lateral force. y v is the lateral velocity of the car. x For the longitudinal speed of the car, For the lateral acceleration of the car, The yaw rate is angular velocity. Let l be the yaw acceleration. a It is the distance from the front axle to the center of the car, l b It is the distance from the rear axle to the center of the car, I z Let y be the rotational inertia of the vehicle body. e ψ is the lateral offset. e It is the heading error between the vehicle and the desired path, l d Let ρ(t) be the aiming distance, ρ(t) be the road curvature, and C be the target distance. a C represents the lateral stiffness of the front wheel. b For the lateral stiffness of the rear wheel, δ e This is the front wheel steering angle. S2.2: Establish as follows Figure 3 The driver steering model shown is expressed as follows:

[0029]

[0030] Wherein, the model input is defined as u d (t)=[θ near θ far ] T x d (t) is the state vector of the system, T d (t) represents the control output.

[0031]

[0032] C d =[0 1]

[0033] Among them, T L and T I These are the lead time constant and the lag time constant, T. N K represents the neuromuscular lag time constant of the driver. A K represents the visual expectation control gain. C This represents the visual compensation control gain.

[0034] Drivers complete driving tasks through compensatory and anticipatory behaviors, which are guided by the near-field perspective (θ). near and far-angle θ far To guide:

[0035]

[0036] Among them, y p D is the lateral offset with a preview distance as the driver approaches the curve. far It is a constant, based on the road curvature radius being between 10-20m.

[0037] S3: Using the model established in S2, a global human-vehicle-road system model is built and fuzzy processed to obtain a global fuzzy model of the human-vehicle-road system, specifically:

[0038] S3.1: To describe the driver's feeling of the steering wheel, the relevant dynamics of the driver and the automated system are described as follows:

[0039]

[0040] Among them, I s and B s R represents the moment of inertia and damping coefficient of the steering system, respectively. s For steering ratio, η t δ is the tire contact width. e T represents the front wheel steering angle. d (t), T a (t) and T al (t) represent the driver torque, auxiliary torque, and self-aligning torque, respectively.

[0041] S3.2: Establish the following global human-vehicle-road system model:

[0042]

[0043] B s2 =[000-v x 00]T

[0044] in, ω r (t)=ρ(t).

[0045] S3.2: Considering the change in tire cornering stiffness with road friction, C a and C b Re-describe and establish an effective TS fuzzy model:

[0046] C a =σ1(t)C ao

[0047] C b =σ2(t)C bo

[0048] Among them, C ao and C bo These are the nominal cornering stiffnesses of the front and rear tires, σ1(t), σ2(t) = v x and σ3(t)=σ1(t) / v x Three prerequisite variables describe the time-varying parameter v x C a and C b The generated model parameters are used to establish a TS fuzzy model. In practical applications, due to sensor limitations, longitudinal velocity is difficult to measure accurately, leading to uncertainty in the membership function. Therefore, an interval type II fuzzy model is used to handle the uncertainty. The three premise variables are expressed as follows:

[0049]

[0050] Where k = 1, 2, 3, σ k The maximum and minimum values ​​of (t) are respectively and σ k The membership function corresponding to (t) is expressed as θ. ks (t) and θ kb (t), the calculation formula is as follows:

[0051]

[0052] For σ1(t) and σ2(t), the upper and lower membership functions are expressed as:

[0053]

[0054] Where k = 1, 2, the upper and lower membership functions are represented by the superscripts L and U, respectively, and the subscript +Δ indicates that at θ ks (t) and θ kb In the calculation of (t), σ is usedk (t)+Δ replaces σ k (t), the subscript -Δ indicates that at θ ks (t) and θ kb In the calculation of (t), σ is used k (t)-Δ replaces σ k (t).

[0055] For the premise variable σ3(t), the upper and lower membership functions are:

[0056]

[0057] Wherein, the subscript σ 1b This indicates that in calculating θ 3s (t) and θ 3b When (t), replace σ1(t) with σ 1b ; Subscript σ 1s This indicates that in calculating θ 3s (t) and θ 3b When (t), replace σ1(t) with σ 1s .

[0058] The interval type II fuzzy model of the human-vehicle-road system is obtained as follows:

[0059] Fuzzy rule i: When σ1(t) is the "minimum or maximum value", σ2(t) is the "minimum or maximum value", and σ3(t) is the "minimum or maximum value", then:

[0060]

[0061] Among them, i=1,2,...,8, A si B s1i and B s2i By respectively dividing matrix A of the above global human-vehicle-road system model s B s1 B s2 The corresponding values ​​in the equation are obtained by replacing them with σ1(t), σ2(t), and σ3(t); for each fuzzy rule, the activation strength is:

[0062]

[0063] Where, φ s (t) is the lower bound of the activation strength, φ b (t) is the upper bound of the activation strength. The following fuzzy model of the global human-vehicle-road system is obtained:

[0064]

[0065] Where, φ i (t) simplifies to φi , It is a nonlinear function that depends on parameter uncertainty, φ is (t) is the lower bound of the activation strength of the i-th fuzzy rule, φ ib The i-th fuzzy rule is the upper bound of the activation strength.

[0066] S4: Based on the global fuzzy model of the human-vehicle-road system, design a fast adaptive observer to obtain an estimate of the driver's torque, specifically:

[0067] Construct the following fast adaptive observer:

[0068]

[0069] in, It is an estimate of the vehicle's condition. It is an estimate of the output. This is an estimate of the driver's torque, L. j and F j Let Γ be the observer gain matrix, Γ be the observer learning rate, τ(t) be the time delay, and C be the observer gain matrix. s =[0 5×1 I 5×5 ] is the coefficient matrix.

[0070] By defining the estimation error as: Based on the designed observer and the global fuzzy model of the human-vehicle-road system, the following error dynamics system is obtained:

[0071]

[0072] Among them, e x (t) Vehicle state error, e y (t) Output error, by making the error dynamics system asymptotically stable and satisfying H with a decay level of γ1. ∞ Performance, by minimizing the H of the error dynamics system ∞ Norms are used to ensure the stability of error dynamic systems, i.e.:

[0073]

[0074] Among them, e T (t) represents the driver torque error, ω r (t) serves as the disturbance input to the error dynamics system, and γ1 is H ∞ Performance degradation level.

[0075] For the observer gain matrix F j and L jThe value of is determined by the following method: For the above error dynamics system, if there exist matrices P > 0, Q > 0, W > 0, Y i The Lyapunov function is established as follows:

[0076]

[0077] Where δ is represented as: δ = f max ×(Γ -1 W -1 Γ -1 ) > 0.

[0078] Differentiating the Lyapunov function yields the matrix inequality that makes the derivative of the Lyapunov function negative definite:

[0079]

[0080] Υ ii <0, i = 1, 2, ..., r

[0081] Υ ij +Υ ji <0, i, j = 1, 2, ..., r, i < j

[0082]

[0083] The matrix mentioned above is a symmetric positive definite matrix, and the asterisk indicates the transpose of the symmetric part of the matrix. This can be transformed into the following optimization problem:

[0084] Minimizeηsubject to

[0085]

[0086] The gain matrix of the observer can be expressed as:

[0087] L j =P -1 Y j

[0088] Where η = 10 -6 If the above conditions are met, then the gain matrix F of the observer is obtained. j and L j This allows us to obtain an estimate of the driver's torque.

[0089] S5: Based on the estimated value of the driver's torque and combined with the ideal value of the driver's torque, the driver fatigue parameters are characterized as follows:

[0090]

[0091] in, For driver fatigue parameters, T is an estimated value for the driver's torque. d (t) represents the ideal value of the driver's torque.

[0092] S6: Design an adaptive assistance controller based on driver fatigue parameters.

[0093] The adaptive auxiliary controller is:

[0094]

[0095] Among them, K j For the controller gain matrix, x is a parameter for driver fatigue. s (t) represents the state variables of the global human-vehicle-road system.

[0096] Next, we determine the control performance. The conditions that the global human-vehicle-road system control output must meet are:

[0097]

[0098] Among them, y s (t) represents the global human-vehicle-road output, and γ2 represents the H... ∞ Given positive scalars γ2, α1, α2, if there exist positive definite matrices R, W1, W2, W3, N1, N2 and general matrices U, G, i, j = 1, 2, ..., 8 such that the following linear matrix inequality holds, then there exists an auxiliary controller that makes the global human-vehicle-road system asymptotically stable and satisfies H ∞ performance:

[0099] Θ ii <0

[0100] Θ ij +Θ ji <0, i<j

[0101]

[0102] in:

[0103]

[0104] By solving the above matrix inequalities, the controller gain matrix K can be obtained. j The adaptive assistance controller is obtained, which, based on the vehicle's state parameters, can output the auxiliary torque T required by a fatigued driver to steer. a It provides auxiliary control for the car.

[0105] S7: Provides auxiliary control for the vehicle based on the adaptive assistance controller.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the ideas of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for lateral assist control of a vehicle considering driver fatigue, characterized in that, Includes the following steps: S1: Collect the vehicle's inherent parameters and real-time parameters; S2: Use the parameters collected by S1 to establish the vehicle's lateral dynamics and kinematics model, as well as the driver's steering model; S3: Using the model established in S2, establish a global human-vehicle-road system model and perform fuzzy processing to obtain a global human-vehicle-road system fuzzy model; S4: Based on the global human-vehicle-road system fuzzy model, design a fast adaptive observer to obtain an estimate of the driver's torque; S5: Based on the estimated value of the driver's torque and combined with the ideal value of the driver's torque, characterize the driver's fatigue parameters; S6: Design an adaptive auxiliary controller based on driver fatigue parameters; S7: Provides auxiliary control for the vehicle based on the adaptive assistance controller; Step S2 includes: S2.1: Establish the following lateral dynamics and kinematic model of the vehicle: in, The lateral force on the front wheel, Let m be the lateral force on the rear wheel, and m be the total mass of the car. For the lateral speed of the car, For the longitudinal speed of the car, For the lateral acceleration of the car, The yaw rate is angular velocity. The yaw acceleration is... It is the distance from the front axle to the center of the car. It is the distance from the rear axle to the center of the car. Let the moment of inertia of the vehicle body be _____. This is the lateral offset. It is the heading error between the vehicle and the desired path. Pre-aiming distance, For road curvature, For the lateral stiffness of the front wheel, For the lateral stiffness of the rear wheel, It is the front wheel steering angle; S2.2: Establish the following driver steering model: Wherein, the model input is defined as , Let be the state vector of the system. To control the output; in, and These are the lead time constant and the lag time constant, respectively. This represents the neuromuscular lag time constant of the driver. Represents visual expectation control gain. Represents visual compensation control gain; Drivers complete driving tasks through compensatory and anticipatory behaviors, which are guided by near-field perspectives. and distant view To guide: in, This is a lateral offset with a preview distance, as the driver approaches a curve. It is a constant, based on the fact that the radius of curvature of the road is between 10 and 20 m; Step S3 specifically involves: S3.1: To describe the driver's feeling of the steering wheel, the relevant dynamics of the driver and the automated system are described as follows: in, and These represent the moment of inertia and damping coefficient of the steering system, respectively. For steering ratio, This refers to the tire contact width. Represents the front wheel steering angle. , and These represent the driver's torque, auxiliary torque, and self-aligning torque, respectively. S3.2: Establish the following global human-vehicle-road system model: in, , ; S3.3: Considering the change in tire cornering stiffness with road surface friction, and Re-description for building an effective TS fuzzy model: in, and It is the nominal cornering stiffness of the front and rear tires, selected , and Three prerequisite variables to describe the time-varying parameters , and The generated model parameters are used to establish a TS fuzzy model. Since the membership function exhibits uncertainty, an interval type II fuzzy model is used to handle this uncertainty. The three premise variables are expressed as follows: in, , The maximum and minimum values ​​are respectively and , The corresponding membership function is expressed as: and The calculation formula is as follows: for and The membership functions for upper and lower membership are expressed as follows: in, The superscripts L and U represent the membership functions, respectively, and the subscripts U represent the membership functions. Indicates in and In the calculation, use replace subscript Indicates in and In the calculation, use replace ; For premise variables The upper and lower membership functions are: Among them, subscript Indicates in calculation and At that time, Replace with ; Subscript Indicates in calculation and At that time, Replace with ; The interval type II fuzzy model of the human-vehicle-road system is obtained as follows: Fuzzy rule i: when It refers to "minimum or maximum value". It is "minimum or maximum value" and If it is "minimum or maximum value", then: in, , , and By respectively combining the matrices of the above global human-vehicle-road system model , , Replace the corresponding value in with , , This is obtained from; for each fuzzy rule, the activation strength is: in, It is the lower bound of the activation strength. This is the upper bound of the activation intensity; the following global fuzzy model of the human-vehicle-road system is obtained: in, Simplified to , , It is a nonlinear function that depends on parameter uncertainty. It is the lower bound of the activation strength of the i-th fuzzy rule. It is the upper bound of the activation strength of the i-th fuzzy rule.

2. The vehicle lateral assist control method considering driver fatigue according to claim 1, characterized in that: The inherent parameters of the vehicle include: total vehicle mass, distance from the front axle to the center of the vehicle, distance from the rear axle to the center of the vehicle, moment of inertia of the vehicle body, lateral stiffness of the front wheels and lateral stiffness of the rear wheels. The real-time parameters of the vehicle include: front wheel steering angle, vehicle longitudinal speed, vehicle lateral speed, and yaw rate.

3. The vehicle lateral assist control method considering driver fatigue according to claim 1, characterized in that, The observer is constructed in step S4 as follows: in, It is an estimate of the vehicle's condition. It is an estimate of the output. This is an estimate of the driver's torque. and The observer gain matrix is... The observer learning rate; It's a time delay. It is a coefficient matrix; By defining the estimation error as: , Based on the designed observer and the global fuzzy model of the human-vehicle-road system, the following error dynamics system is obtained: in, This indicates the vehicle's condition error. The error representing the output quantity is achieved by making the error dynamics system asymptotically stable and satisfying the following conditions: Attenuation level Performance, by minimizing the error dynamics of the system Norms are used to ensure the stability of error dynamic systems, i.e.: in, For driver torque error, As an external disturbance input to the error dynamics system for Performance degradation level; definition: The time derivative is norm-bounded. : yes The maximum value for the observer gain matrix and The value of is determined by the following method: For the above error dynamics system, if there exists a matrix , , Y i Then the Lyapunov function is established as follows: in, Represented as: ; Differentiating the Lyapunov function yields the matrix inequality that makes the derivative of the Lyapunov function negative definite: , , The matrix mentioned above is a symmetric positive definite matrix, and the asterisk indicates the transpose of the symmetric part of the matrix. This can be transformed into the following optimization problem: Minimize subject to The gain matrix of the observer can be expressed as: in, If the above matrix inequality has a solution, then the gain matrix of the observer is obtained. and This allows us to obtain an estimate of the driver's torque. .

4. The vehicle lateral assist control method considering driver fatigue according to claim 3, characterized in that, The driver fatigue parameters in step S5 are characterized as follows: in, For driver fatigue parameters, This is an estimate of the driver's torque. This is the ideal value for the driver's torque.

5. The vehicle lateral assist control method considering driver fatigue according to claim 4, characterized in that, In step S6, based on the driver fatigue parameters, the adaptive assistance controller is designed as follows: in, For the controller gain matrix, For driver fatigue parameters, This refers to the state variables of the global human-vehicle-road system. Next, we determine the control performance. The conditions that the global human-vehicle-road system control output must meet are: in, For the global human-vehicle-road system control output, for Performance degradation level, given a positive scalar , , If a positive definite matrix exists , , , , , and general matrices , , If the following linear matrix inequality holds, then there exists an auxiliary controller that makes the global human-vehicle-road system asymptotically stable and satisfies the following conditions. performance: in: By solving the above matrix inequalities, the controller gain matrix can be obtained. This yields an adaptive assistance controller, which, based on the vehicle's state parameters, can output the auxiliary torque required for a fatigued driver to steer. It provides auxiliary control for the car.

Images