CN118220143A - A personalized human-machine collaborative lane keeping robust control method, system and medium - Google Patents

Abstract

The application discloses a robust control method, a system and a medium for cooperative lane keeping of a personalized man-machine, which firstly establish a linear time-varying man-vehicle-road model of a driver in a ring; secondly, comprehensively considering factors such as steering behaviors of a driver, transverse comprehensive deviation conditions of a vehicle and the like, designing a man-machine control right game model considering characteristics of a driving group, and carrying out weighting adjustment on input steering moment of a driver and an intelligent auxiliary system so as to realize explicit expression of man-machine interaction effect; considering factors such as road curvature disturbance, insufficient linear model adaptation and time-varying characteristics of model parameters under complex working conditions, and designing a state feedback gamma suboptimal H _∞ robust controller based on a T-S fuzzy control theory; the application can better consider the track tracking precision and the man-machine friendliness.

Description

A personalized human-machine collaborative lane keeping robust control method, system and medium

Technical Field

The present application relates to the field of vehicle control technology, and in particular to a personalized human-machine collaborative lane keeping robust control method, system and medium.

Background technique

Driving decisions in complex scenarios, new infrastructure construction, social ethics, and legislation related to the definition of rights and responsibilities have largely hindered the commercialization of driverless technology. Therefore, before achieving full-domain driverless driving, the industry and academia are working hard to study the human-machine sharing system with the driver in the loop, giving full play to the advantages of both humans and self-driving cars as the best choice for the transition period.

Human driving behavior is often to pay attention to the driving environment and make decisions through vision, hearing, touch, etc. Humans are good at making fuzzy judgments, handling uncertain systems, and perceiving and making decisions based on value and situation. On the contrary, machines are based on precise and deterministic judgments, facts and rules. How to integrate the two to form human-machine fusion perception, hybrid decision-making, and ultimately improve the intelligence of automobiles is a current research area.

Human-machine shared control is a joint effort between humans and machines to complete driving tasks, and through the hybrid enhancement of human-machine intelligence, it ensures driving safety and improves driving performance. On the one hand, humans have strong reasoning and decision-making capabilities in complex environments, and have an advantage in upper-level planning, but are easily affected by psychological and physiological factors and produce improper operations; on the other hand, machine systems have the ability to perform precise control, but their learning and adaptability are weak. By complementing the advantages and disadvantages of the two, the intelligence level of the car driving system can be maximized.

In recent years, human-machine shared control has been a hot topic in the field of autonomous driving. As vehicle longitudinal control technology has matured, such as adaptive cruise control (ACC) and automatic emergency braking (AEB), existing human-machine shared control research focuses on lateral control, which has the following key problems:

(1) How to meet the control characteristics requirements of different driving groups and make effective decisions on human-machine control rights while ensuring driving safety.

(2) Considering factors such as the uncertainty of driver control, differences in driving habits and skills, and the driver's changing trust in the intelligent assistance system, how to design a reasonable shared control structure and collaborative control strategy and select the optimal shared control algorithm are difficult issues that deserve in-depth study.

The interactive relationship between humans and machines forms a one-sided perception, supervision, and compensation and correction, which makes the manipulation of information interaction and perception reflect asymmetric characteristics.

In the current technology and application context, domestic and foreign scholars have proposed the concept of co-driving intelligent vehicles, incorporating people's advanced cognition of fuzzy and uncertain problems into the feedback loop to improve the overall intelligence level of the human-machine co-driving system. This is the instantiation of the human-machine hybrid enhanced intelligence theory in the field of intelligent driving. At present, the unsafe conversion and cooperative control between the driver and the conditional/highly automated driving system are hot research difficulties, and how to combine driver skills with new intelligent vehicle technologies to design a more cooperative driver-in-the-loop co-driving system to achieve efficient human-machine collaboration still faces huge challenges.

Summary of the invention

The present application provides a personalized human-machine collaborative lane keeping robust control method, system and medium, which has the advantage of taking into account the characteristics of the driving group, so that human-machine co-driving can take into account the characteristics of the driving group and improve human-machine friendliness.

On the one hand, the present application provides a personalized human-machine collaborative lane keeping robust control method, comprising the following steps:

Step 1: Establish a vehicle-road model that takes into account the driver's steering behavior;

Step 2: Establish a human-machine collaboration factor model that takes into account the characteristics of the driving group;

Step 3: Based on the above vehicle-road model, a T-S fuzzy model of the co-driving LKAS is established, and the steering torque of the intelligent assistance system output by the T-S fuzzy model of the co-driving LKAS is adjusted according to the human-machine cooperation factor obtained by the above human-machine cooperation factor model.

Furthermore, in step 1, based on the linear time-varying two-degree-of-freedom vehicle extension model, the motor-by-wire steering system model, and the near-far perspective model, a vehicle-road model that takes into account the driver's steering behavior is established:

In the formula, is the state vector, y _L is the lateral deviation of the vehicle's center of mass at the preview point in the vehicle coordinate system, v _y is the lateral velocity, /> is the yaw angle deviation, /> are the yaw angles at the vehicle center of mass and the preview point, ω _r is the yaw angular velocity, and δ _f is the front wheel steering angle; u is the control input, ω = [ρ, σ] ^T is the disturbance vector, ρ is the road curvature, and σ is the modeling error of the wire control steering system model; y = [θ _near ,θ _far ,δ _f ] ^T is the model output, θ _near is the near viewing angle, and θ _far is the far viewing angle; each coefficient matrix satisfies:

Wherein, _vx is the time-varying longitudinal velocity, _Lnear is the single-point preview distance, _Lfar is the far-view point preview distance, _Cf and _Cr are the cornering stiffness of the front and rear wheels respectively, a and b are the distances between the front and rear axles and the vehicle center of mass respectively, _is is the steering gear ratio, _Js is the equivalent moment of inertia of the steering system, and _Bs is the equivalent friction coefficient of the steering system.

Furthermore, in step 2, the human-machine collaborative factor model is mathematically described as follows:

In the formula, β is the human-machine synergy factor, α is the characteristic parameter of the driving group, which is used to distinguish the degree of intervention of the intelligent assistance system that different driving groups can accept; 2ε is the width of the comfortable driving area, which is related to the driving habits of the driving group; is the comprehensive preview deviation, where/> are the normalized values of the lateral deviation and yaw angle deviation respectively, λ ₁ and λ ₂ are design parameters, and satisfy λ ₁ +λ ₂ =1.

Furthermore, the driving groups are divided into an aggressive group, a normal group and a novice group, and it is agreed that the acceptable intervention degrees of the aggressive group, the normal group and the novice group increase successively, and the acceptable intervention timing is advanced successively.

Furthermore, the vehicle-road model described by equation (1) is described with uncertain parameters:

The uncertainty of the cornering stiffness C _f and _Cr is described as follows:

In the formula, are all constants, ξ _f and ξ _f are time-varying parameters, and satisfy ξ _f ≤1, ξ _r ≤1; thus, equation (1) can be expanded into:

In the formula,

Furthermore, in step 3, the T-S fuzzy model of the shared-driving LKAS is:

In the formula, A _i , B _i , G _i , and C _i represent the coefficient matrix of the state equation of the i-th subsystem. is a membership function and satisfies/> is the fuzzy antecedent, r represents the number of fuzzy rules, and the control input u(t) satisfies:

Where, _Ki is the feedback gain matrix of the i-th subsystem;

For the closed-loop control system described by equation (5) and equation (6), a quadratic performance index function is established:

In the formula, Q and R are the state weight matrix and control weight coefficient respectively.

Furthermore, the antecedent variable v _x ∈[v _min ,v _max ], v _x 、/> Expressed as:

In the formula,

Therefore, for the T-S fuzzy model shown in equation (5), r = 4, and satisfies:

By solving _Ki , the optimal control quantity is obtained.

On the other hand, the present application provides a personalized human-machine collaborative lane keeping robust control system, including a processor and a memory, wherein the memory stores a computer program, and when the computer program is called and executed by the processor, the personalized human-machine collaborative lane keeping robust control method as described above is implemented.

On the other hand, the present application provides a computer-readable medium, which stores a computer program. When the computer program is called and executed by a computer, it implements the personalized human-machine collaborative lane keeping robust control method as described above.

In summary, the beneficial effects of this application are:

1. This application takes into account the characteristics of the driving group, so that human-machine co-driving can take into account the characteristics of the driving group and improve human-machine friendliness;

2. This application takes into account factors such as road curvature disturbance under complex working conditions, insufficient adaptation of linear models, and time-varying characteristics of model parameters, and designs a state feedback γ suboptimal H _∞ robust controller based on TS fuzzy control theory, which better balances trajectory tracking accuracy and human-machine friendliness.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic diagram of a vehicle-road reference model based on single-point preview;

FIG2 is a schematic diagram of human-machine collaboration factors for different driving groups;

FIG. 3 is a schematic diagram of a human-machine collaborative steering control system.

Detailed ways

The specific implementation of the present application is described in detail below with reference to the accompanying drawings.

In a specific embodiment of the present application, a personalized human-machine collaborative lane keeping robust control method is provided, comprising the following steps:

Step 1: Establish a vehicle-road model that takes into account the driver's steering behavior;

Step 2: Establish a human-machine collaboration factor model that takes into account the characteristics of the driving group;

Step 3: Based on the above vehicle-road model, a T-S fuzzy model of the co-driving LKAS is established, and the steering torque of the intelligent assistance system output by the T-S fuzzy model of the co-driving LKAS is adjusted according to the human-machine cooperation factor obtained by the above human-machine cooperation factor model.

The vehicle-road reference model based on single-point preview is shown in Figure 1, where XOY is the inertial coordinate system and xoy is the vehicle coordinate system.

In order to better simulate the steering characteristics of actual drivers, a vehicle-road model that takes into account the driver's steering behavior is established based on the linear time-varying two-degree-of-freedom vehicle extension model, the motor-by-wire steering system model, and the near-far perspective model:

In the formula, is the state vector, y _L is the lateral deviation of the vehicle's center of mass at the preview point in the vehicle coordinate system, v _y is the lateral velocity, /> is the yaw angle deviation, /> are the yaw angles at the vehicle center of mass and the preview point, ω _r is the yaw angular velocity, and δ _f is the front wheel steering angle; u is the control input, ω = [ρ, σ] ^T is the disturbance vector, ρ is the road curvature, and σ is the modeling error of the wire control steering system model; y = [θ _near , θ _far , δ _f ] ^T is the model output, θ _near is the near viewing angle, and θ _far is the far viewing angle; each coefficient matrix satisfies:

Wherein, _vx is the time-varying longitudinal velocity, _Lnear is the single-point preview distance, _Lfar is the far-view point preview distance, _Cf and _Cr are the cornering stiffness of the front and rear wheels respectively, a and b are the distances between the front and rear axles and the vehicle center of mass respectively, _is is the steering gear ratio, _Js is the equivalent moment of inertia of the steering system, and _Bs is the equivalent friction coefficient of the steering system.

The local expected path of the intelligent system is always the center line of the lane, but different driving groups have different sensitivities to side deviation, that is, their local expected paths are different. This further shows that different driving groups have different lane keeping habits, and they can accept different intervention times and intervention degrees of the intelligent system.

To this end, taking into account the development trend of the vehicle motion state at the preview position and the lane keeping habits of the driving group, in step 2, a human-machine collaboration factor model considering the characteristics of the driving group is established, and its mathematical description is as follows:

In the formula, β is the human-machine synergy factor, α is the characteristic parameter of the driving group, which is used to distinguish the degree of intervention of the intelligent assistance system that different driving groups can accept; 2ε is the width of the comfortable driving area, which is related to the driving habits of the driving group; is the comprehensive preview deviation, where/> are the normalized values of the lateral deviation and yaw angle deviation respectively, λ ₁ and λ ₂ are design parameters, and satisfy λ ₁ +λ ₂ =1.

The driving groups are divided into an aggressive group, a normal group, and a novice group. It is agreed that the acceptable intervention degree of the aggressive group, the normal group, and the novice group increases in sequence, and the acceptable intervention timing is advanced in sequence. The human-machine collaboration factors for different driving groups are designed, as shown in Figure 2. The larger α is, the smaller the acceptable intervention degree of the driving group is, and vice versa; the larger ε is, the later the acceptable intervention timing of the driving group is, and vice versa.

When the tire cornering characteristic is in the linear region (tire cornering angle ≤ 5°), the tire cornering stiffness is considered to be a constant, which may be recorded as C _f0 and C _r0 . However, under complex working conditions (such as icy and snowy roads, side winds, etc.), the tire cornering characteristic can easily enter the nonlinear region, resulting in the tire cornering stiffness no longer being a constant, but presenting a complex nonlinear mapping relationship, that is, the vehicle-road model shown in formula (1) is no longer a linear model. The vehicle-road model described by formula (1) is described by parameter uncertainty, specifically:

The uncertainty of the cornering stiffness C _f and _Cr is described as follows:

In the formula, are all constants, ξ _f and ξ _f are both time-varying parameters, and satisfy ξ _f ≤1, ξ _r ≤1;

In this way, formula (1) can be expanded into:

In the formula,

The T-S fuzzy model of the shared-driving LKAS shown in formula (4) is established as:

In the formula, _Ai , _Bi , _Gi , _Ci represent the coefficient matrix of the state equation of the i-th subsystem, is a membership function and satisfies/> is the fuzzy antecedent, r represents the number of fuzzy rules, and the control input u(t) satisfies:

Where, _Ki is the feedback gain matrix of the i-th subsystem;

For the closed-loop control system described by equation (5) and equation (6), a quadratic performance index function is established:

In the formula, Q and R are the state weight matrix and control weight coefficient respectively.

In order to optimize the quadratic performance index function J(t), Theorem 1 is given.

Theorem 1 The above closed-loop control system remains stable when ω(t) = 0, and for any ω(t) ≠ 0, there exists a state feedback controller that satisfies the system H _∞ performance. The necessary and sufficient condition is that given γ > 0, there exists a symmetric positive definite matrix P and a matrix V _i of any suitable dimension such that the following LMI holds.

In the formula, Ω _i ＝A _i P+B _i V _i +(A _i P+B _i V _i ) ^T , V _i ＝K _i P, I is the unit matrix, i∈[1,r].

The stability analysis of the above system is as follows:

Create the Lyapunov function:

V＝x ^T P ^-1 x (9)

According to Schur's complement lemma, formula (8) is equivalent to:

Multiplying equation (10) by diag{P ^-1 , I} on the left and right, we can get:

Multiplying equation (11) by [x ^T , ω ^T ] on the left and [x ^T , ω ^T ] ^T on the right respectively, we can simplify it to get:

From formula (12), we know that when ω(t) = 0, That is, the system is stable.

Similarly, from formula (12), when the initial condition of the system satisfies x(0) = 0, it is easy to obtain:

By minimizing γ, J(t) is minimized, thus Theorem 1 is proved.

Due to the influence of factors such as road curvature and side wind, it is difficult for v _x to remain constant, that is, it has time-varying characteristics. Suppose the antecedent variable v _x ∈ [v _min ,v _max ], v _x 、/> Expressed as:

In the formula,

Therefore, for the T-S fuzzy model shown in equation (5), r = 4, and satisfies:

The LMI toolbox in Matlab is used to solve _Ki and obtain the optimal control quantity.

The design of the human-machine collaborative steering control system based on the above method is shown in Figure 3, including: T-S robust controller, human-machine collaborative factor decision, CarSim/Simulink simulation platform, driver input and other modules.

In another specific embodiment of the present application, a personalized human-machine collaborative lane keeping robust control system is provided, including a processor and a memory, wherein the memory stores a computer program, and when the computer program is called and executed by the processor, the personalized human-machine collaborative lane keeping robust control method as described above is implemented.

In another specific embodiment of the present application, a computer-readable medium is provided, wherein the computer-readable medium stores a computer program, and when the computer program is called and executed by a computer, the personalized human-machine collaborative lane keeping robust control method as described above is implemented.

The above is only a preferred implementation of the present application. It should be pointed out that a person skilled in the art can make several modifications and improvements without departing from the inventive concept of the present application, and these all fall within the scope of protection of the present application.

Claims (9)

1. A personalized human-machine collaborative lane keeping robust control method, characterized by comprising the following steps: Step 1: Establish a vehicle-road model that takes into account the driver's steering behavior; Step 2: Establish a human-machine collaboration factor model that takes into account the characteristics of the driving group; Step 3: Based on the above vehicle-road model, a T-S fuzzy model of the co-driving LKAS is established, and the steering torque of the intelligent assistance system output by the T-S fuzzy model of the co-driving LKAS is adjusted according to the human-machine cooperation factor obtained by the above human-machine cooperation factor model. 2. The personalized human-machine collaborative lane keeping robust control method according to claim 1 is characterized in that, in step 1, a vehicle-road model considering the driver's steering behavior is established based on a linear time-varying two-degree-of-freedom vehicle extension model, a motor-by-wire steering system model, and a near-far perspective model: In the formula, is the state vector, y _L is the lateral deviation of the vehicle's center of mass at the preview point in the vehicle coordinate system, v _y is the lateral velocity, /> is the yaw angle deviation, /> are the yaw angles at the vehicle center of mass and the preview point, ω _r is the yaw angular velocity, and δ _f is the front wheel steering angle; u is the control input, ω = [ρ, σ] ^T is the disturbance vector, ρ is the road curvature, and σ is the modeling error of the wire control steering system model; y = [θ _near ,θ _far ,δ _f ] ^T is the model output, θ _near is the near viewing angle, and θ _far is the far viewing angle; each coefficient matrix satisfies: Wherein, _vx is the time-varying longitudinal velocity, _Lnear is the single-point preview distance, _Lfar is the far-view point preview distance, _Cf and _Cr are the cornering stiffness of the front and rear wheels respectively, a and b are the distances between the front and rear axles and the vehicle center of mass respectively, _is is the steering gear ratio, _Js is the equivalent moment of inertia of the steering system, and _Bs is the equivalent friction coefficient of the steering system. 3. The personalized human-machine collaborative lane keeping robust control method according to claim 1, characterized in that in step 2, the human-machine collaborative factor model is mathematically described as follows: In the formula, β is the human-machine synergy factor, α is the characteristic parameter of the driving group, which is used to distinguish the degree of intervention of the intelligent assistance system that different driving groups can accept; 2ε is the width of the comfortable driving area, which is related to the driving habits of the driving group; is the comprehensive preview deviation, where/> are the normalized values of the lateral deviation and yaw angle deviation respectively, λ ₁ and λ ₂ are design parameters, and satisfy λ ₁ +λ ₂ =1. 4. The personalized human-machine collaborative lane keeping robust control method according to claim 1 is characterized in that the driving group is divided into an aggressive group, a normal group and a novice group, and it is agreed that the acceptable intervention degrees of the aggressive group, the normal group and the novice group increase in sequence, and the acceptable intervention timing is advanced in sequence. 5. The personalized human-machine collaborative lane keeping robust control method according to claim 1 is characterized in that the vehicle-road model described by equation (1) is described with uncertain parameters: The uncertainty of the cornering stiffness C _f and _Cr is described as follows: In the formula, are all constants, ξ _f and ξ _f are both time-varying parameters, and satisfy ξ _f ≤1, ξ _r ≤1; In this way, formula (1) can be expanded into: In the formula, 6. The personalized human-machine cooperative lane keeping robust control method according to claim 1, characterized in that, in step 3, the T-S fuzzy model of the co-driving LKAS is: Where A _i , B _i , G _i , and C _i represent the coefficient matrices of the state equation of the ith subsystem, and w _i (θ) is the membership function and satisfies θ is the fuzzy antecedent, r represents the number of fuzzy rules, and the control input u(t) satisfies: Where, _Ki is the feedback gain matrix of the i-th subsystem; For the closed-loop control system described by equation (5) and equation (6), a quadratic performance index function is established: In the formula, Q and R are the state weight matrix and control weight coefficient respectively. 7. The personalized human-machine cooperative lane keeping robust control method according to claim 1, characterized in that the antecedent variable v _x ∈ [v _min , v _max ], v _x 、/> Expressed as: In the formula, Therefore, for the T-S fuzzy model shown in equation (5), r = 4, and satisfies: By solving _Ki , the optimal control quantity is obtained. 8. A personalized human-machine collaborative lane keeping robust control system, characterized in that it includes a processor and a memory, the memory stores a computer program, and when the computer program is called and executed by the processor, it implements the personalized human-machine collaborative lane keeping robust control method as described in any one of claims 1-7. 9. A computer-readable medium, characterized in that the computer-readable medium stores a computer program, and when the computer program is called and executed by a computer, it implements the personalized human-machine collaborative lane keeping robust control method as described in any one of claims 1-7.