CN111791898B - Automatic driving automobile collision avoidance control method based on cooperation type game - Google Patents

Abstract

The invention discloses an automatic driving automobile collision avoidance control method based on a cooperative game, which comprises the following steps of 1, detecting the surrounding environment through vehicle-mounted sensing equipment, judging whether barriers exist around the surrounding environment, and continuing to detect if the barriers do not exist; step 2, combining the position information of the obstacle obtained in the step 1, selecting a series of safe track points by the automatic driving automobile, and fitting the track points by utilizing a Bezier curve so as to generate a collision avoidance track, and step 3, decelerating the automobile to a safe collision avoidance speed; step 4, implementing nonlinear robust trajectory tracking control, and inhibiting the influence of curvature radius change on tracking performance; and 5, after the collision avoidance process is finished, the vehicle is switched to the self-adaptive cruise state. The invention is based on a cooperative game framework, and associates the stability and the tracking performance with the adjustable parameters of the controller respectively, thereby realizing robust track tracking control considering stability.

Description

Automatic driving automobile collision avoidance control method based on cooperation type game

Technical Field

The invention relates to a collision avoidance control method for an automatic driving automobile, in particular to a collision avoidance control method for an automatic driving automobile based on a cooperation type game.

Background

With the rapid development of the automobile technology in the 21 st century, the automatic driving level is continuously improved, and the automatic driving functions of various levels are widely researched. However, as the level of automatic driving increases, the safety of the automatic driving vehicle is one of the important considerations of researchers, especially ensuring that the automatic driving vehicle can realize a collision avoidance function.

Currently, many research methods have been used to develop collision avoidance functions for autonomous vehicles. The reinforcement learning method can master state and environment knowledge by training the automobile when encountering obstacles, thereby realizing the trajectory planning or obstacle avoidance of the automatic driving automobile. Quintic polynomials, bezier curves, and S-curves are widely used to generate feasible collision avoidance trajectories.

Existing research typically utilizes trajectory planning to address collision avoidance issues, however, it remains an issue whether the feasible trajectories generated by such methods can be performed by a vehicle.

Some researches directly utilize a dynamic control method to realize collision avoidance of the vehicle, but related researches still lack consideration on vehicle stability in the collision avoidance process. In addition, various uncertainties in actual scenes have great influence on the dynamic performance of the vehicle, and the uncertainty is not considered in the existing research.

Disclosure of Invention

The invention aims to generate a collision avoidance track by utilizing a Bezier curve, realize efficient track tracking by utilizing front wheel steering (AFS), improve the transverse stability of a vehicle by active rear wheel steering (ARS), realize the cooperative control of the front wheel steering and the rear wheel steering based on a cooperative game theory, improve the transverse stability while tracking collision avoidance and finally realize efficient and safe collision avoidance control.

The invention provides an automatic driving automobile collision avoidance control method based on a cooperative game, which comprises the following steps:

step 1, detecting the surrounding environment through vehicle-mounted sensing equipment, judging whether obstacles exist around, if not, continuing to detect, and if so, entering step 2;

step 2, combining the position information of the obstacle obtained in the step 1, automatically driving the automobile to select a series of safe track points, and fitting the track points by utilizing a Bezier curve so as to generate a collision avoidance track;

step 3, decelerating the vehicle to a safe collision avoidance vehicle speed;

step 4, implementing nonlinear robust trajectory tracking control, and inhibiting the influence of curvature radius change on tracking performance;

and 5, after the collision avoidance process is finished, the vehicle is switched to the self-adaptive cruise state.

Further, in step 1, the vehicle-mounted sensing device selects a vehicle-mounted camera and a radar.

Further, the step 4 includes:

step 4.1, establishing a track tracking dynamic model based on a two-degree-of-freedom vehicle transverse dynamic equation, wherein the dynamic model is simplified as the following form:

wherein the content of the first and second substances,

wherein the content of the first and second substances,

u＝[δ_f δ_r]^T，σ₁＝2(C_f+C_r)；σ₂＝-2(l_fC_f-l_rC_r)；

in the formula:

is the course angle deviation; e.g. of the type_yIs the lateral displacement deviation;

is the lateral velocity deviation;

is yaw angular velocity deviation; v. of_xIs the longitudinal speed of the vehicle; v. of_yIs the vehicle lateral velocity; c. C_RIs the radius of curvature of the path; m is the mass of the automobile; c_fFront wheel cornering stiffness; c_rIs the rear wheel sideDeflection stiffness; l_fThe distance from the center of mass of the automobile to the front axle; l_rThe distance from the mass center of the automobile to the rear axle; i is_zIs horizontal swinging moment of inertia; delta_fIs the angle of rotation of the front wheel, delta_rFor rear wheel steering angle, w_eFor modeling errors due to speed fluctuations

Step 4.2, designing front wheel steering angle input and rear wheel steering angle input by using robust optimal control according to the dynamic model in the step 4.1:

u＝-R^-1B^TPx+γR^-1B^TPx‖B^TPx‖^∈-2(αx+β)^∈

wherein P is an algebraic Riccati (Riccati) equation

A^TP+PA-2PBR^-1B^TSolution of P + Q ═ 0;

in the formula: r ═ Q ═ I^4×4Is an identity matrix; alpha and beta are selected constants; gamma and epsilon are adjustable coefficients of the controller; the matrix A, B is the matrix defined in step 4.1;

4.3, representing system uncertainty by using a fuzzy set, and designing a corresponding membership function;

step 4.4, determining the value range of the adjustable parameters gamma and epsilon in the step 4.2

γ_min≤γ≤γ_max

∈_min≤∈≤∈_max

Wherein, γ_min、γ_max、∈_min、∈_maxThe known constants respectively represent the minimum value and the maximum value of the two adjustable parameters;

step 4.5, utilizing the Lyapunov function to find the boundary of the system performance V, namely

V＝x^TPx≤η₁(γ,∈,t)+η₂(γ,∈)

Wherein the content of the first and second substances,

p is the matrix P obtained in step 4.2

η₂(γ,∈)＝κΞ

Wherein xi is a second intermediate variable expressed by

In the formula: k is a predetermined constant, t₀The time when the observer starts to observe;

is t₀The system performance at time can be represented by t₀Calculating the system state at the moment;

is a first intermediate variable obtained during the process of scaling to obtain the boundary of V;

respectively corresponding the tracking cost and the stability cost with the stability performance and the accumulated transient performance to obtain corresponding cost functions, and further solving a corresponding game optimization problem;

step 4.6, solving the two-person cooperative game problem consisting of the step 4.4 and the step 4.5, wherein the optimal solution of the two-person cooperative game problem is pareto optimal to obtain the optimal value gamma of the adjustable parameter^*,∈^*(ii) a The set of optimal values gamma^*,∈^*Substituting the control law expression in the step 4.2 to obtain front and rear wheel rotation angles corresponding to pareto optima;

and 4.7, respectively sending the front and rear wheel steering angle input commands in the step 4.5 to the front and rear axle steering mechanisms, and executing corresponding commands by the corresponding steering mechanisms.

Further, in step 4.5, the cost function is as follows:

wherein, J_stabilityCost for stability; j. the design is a square_trackTo track costs; d is the mapping operation of the fuzzy number and the real number.

The invention has the beneficial effects that:

1. the method has the advantages that the characteristic of boundedness of the uncertainty of the actual vehicle parameters is considered, the uncertainty is represented by innovatively using a fuzzy set, and the corresponding robust trajectory tracking control law is designed according to the characteristic, so that the influence of perturbation of the actual vehicle system parameters on the tracking performance can be effectively inhibited;

2. the method takes the characteristics of real-time change of the curvature radius of the generated collision avoidance track into consideration, designs a corresponding robust track tracking control law, and can effectively inhibit the influence of the change of the curvature radius on the tracking performance;

3. the method considers the condition that the default speed of the established dynamic model is constant and the actual speed has certain fluctuation, takes the modeling error generated by speed fluctuation as uncertainty, and can effectively inhibit the influence of the speed fluctuation on the tracking performance;

4. the invention is based on a cooperative game framework, and associates the stability and the tracking performance with the adjustable parameters of the controller respectively, thereby realizing robust track tracking control considering stability.

Drawings

FIG. 1 is a schematic diagram of an autonomous vehicle collision avoidance process;

FIG. 2 is a general flow chart of the method;

FIG. 3 is a flow chart for generating a collision avoidance trajectory;

FIG. 4 is a flow chart of control for non-linear robust path tracking;

FIG. 5 is a diagram illustrating effects of an embodiment;

Detailed Description

The technical solution of the present invention will be described in detail with reference to the accompanying fig. 1-5.

As shown in fig. 1 to 4, the embodiment provides a cooperative game-based autonomous automobile collision avoidance control method, including:

step 1, detecting the surrounding environment through vehicle-mounted sensing equipment, judging whether obstacles exist around, if not, continuing to detect, and if so, entering step 2;

in this step, the vehicle-mounted sensing device may be a vehicle-mounted camera, a radar, or the like.

Step 2, combining the position information of the obstacle obtained in the step 1, automatically driving the automobile to select a series of safe track points, and fitting the track points by utilizing a Bezier curve so as to generate a collision avoidance track;

step 3, decelerating the vehicle to a safe collision avoidance vehicle speed;

in the step, collision avoidance with the barrier is ensured through vehicle deceleration, and meanwhile, the stability of the vehicle can be improved in the collision avoidance and steering process;

step 4, implementing nonlinear robust trajectory tracking control, and inhibiting the influence of curvature radius change on tracking performance;

and 4.1, establishing a track tracking dynamics model based on a two-degree-of-freedom vehicle transverse dynamics equation.

Wherein the content of the first and second substances,

u＝[δ_f δ_r]^T，σ₁＝2(C_f+C_r)；σ₂＝-2(l_fC_f-l_rC_r)；

in the formula, the meaning of each parameter is:

is the course angle deviation; e.g. of the type_yIs the lateral displacement deviation;

is the lateral velocity deviation;

is yaw angular velocity deviation; v. of_xIs the longitudinal speed of the vehicle; c. C_yIs the vehicle lateral velocity; c. C_RIs the radius of curvature of the path; m is the mass of the automobile; c_fFront wheel cornering stiffness; c_rIs rear wheel cornering stiffness; l_fThe distance from the center of mass of the automobile to the front axle; l_rThe distance from the mass center of the automobile to the rear axle; i is_zIs horizontal swinging moment of inertia; delta_fIs the angle of rotation of the front wheel, delta_rFor rear wheel steering angle, w_eIs the modeling error due to speed fluctuations.

The above kinetic model is simplified as follows:

wherein the content of the first and second substances,

step 4.2, designing front wheel steering angle input and rear wheel steering angle input by using robust optimal control according to the dynamic model in the step 4.1:

u＝-R^-1B^TPx+γR^-1B^TPx‖B^TPx‖^∈-2(αx+β)^∈

wherein P is an algebraic Riccati (Riccati) equation

A^TP+PA-2PBR^-1B^TP+Q＝0

The solution of (1).

The meaning of each parameter in the above formula is: r ═ Q ═ I^4×4Is an identity matrix; alpha and beta are selected constants; gamma and epsilon are adjustable coefficients of the controller; the matrix A, B is the matrix defined in step 4.1.

And 4.3, representing the uncertainty of the system by using the fuzzy set, and designing a corresponding membership function.

First, the uncertainty from the cornering stiffness of a tire can be defined as:

wherein, is_fBelong to

This fuzzy set;

the rear wheel side cornering stiffness uncertainty may be defined as:

wherein, is_rBelong to

This fuzzy set;

the road curvature uncertainty may be defined as:

wherein, Δ c_RBelong to

This fuzzy set;

the uncertainty that the speed fluctuation produces on the model can be defined as: w is a_eBelong to

This fuzzy set.

The various parameters have the meaning:

nominal yaw stiffness for the front axle;

nominal yaw stiffness of the rear axle;

is the nominal road curvature; delta C_fUncertainty of lateral deflection rigidity of the front axle; delta C_rUncertainty of lateral deflection stiffness of the rear axle; Δ c_RIs the road curvature uncertainty; w is a_eIs equivalent speed fluctuation interference; each S_**Representing a fuzzy set; each mu_**Representing a membership function; each omega_**The value set is expressed, the value of which is determined by a designer, and the value set is generally-30% to + 30% of a nominal value.

Step 4.4, determining the value range of the adjustable parameters gamma and epsilon in the step 4.2

γ_min≤γ≤γ_max

∈_min≤∈≤∈_max

Wherein, γ_min、γ_max、∈_min、∈_maxThe known constants represent the minimum and maximum values of the two adjustable parameters, respectively.

Step 4.5, utilizing the Lyapunov function to find the boundary of the system performance V, namely

V＝x^TPx≤η₁(γ,∈,t)+η₂(γ,∈)

Wherein the content of the first and second substances,

p is the matrix P obtained in step 4.2

η₂(γ,∈)＝κΞ

Wherein xi is a second intermediate variable expressed by

The meaning of each parameter in the above two formulas is as follows: k is a predetermined constant, t₀The time when the observer starts to observe;

is t₀The system performance at time can be represented by t₀Calculating the system state at the moment;

is the first intermediate variable obtained during the scaling to obtain the boundary of V.

Respectively corresponding the tracking cost and the stability cost to the stability performance and the accumulated transient performance, and obtaining a corresponding cost function as follows:

wherein, J_stabilityCost for stability; j. the design is a square_trackTo track costs; d is a mapping operation of the fuzzy number and the real number, which is specifically as follows:

where Φ is the value set of ζ, μ_Φ(ζ) is a membership function, and D is a function of η₁(γ,∈,t)、η₂(gamma,. epsilon.) contained fuzzy sets

Mapping to real number domain to obtain the result in real number domainThe function can further solve the corresponding game optimization problem.

Step 4.6, solving the two-person cooperative game problem consisting of the step 4.4 and the step 4.5, wherein the optimal solution of the two-person cooperative game problem is pareto optimal to obtain the optimal value (gamma) of the adjustable parameter^*,∈^*). And (3) substituting the group of values into the control law expression in the step 4.2 to obtain the front wheel corner and the rear wheel corner corresponding to the pareto optimal.

Step 4.7, the input commands of the front wheel steering angle and the rear wheel steering angle in the step 4.5 are respectively sent to the front axle steering mechanism and the rear axle steering mechanism, and the corresponding commands are executed by the corresponding steering mechanisms;

and 5, after the collision avoidance process is finished, the vehicle is switched to the self-adaptive cruise state.

A specific example is provided below:

step 1: detecting surroundings by a vehicle-mounted sensor device, collecting obstacle positioning information (X)_obs，Y_obs)

Step 2: the autonomous vehicle selects a series of safe trajectory points and fits these path points using a Bezier curve to generate a series of collision avoidance points (X)₁，Y₁)，(X₂，Y₂)，…，(X_N，Y_N) And generating a collision avoidance track Y as f (X), wherein f (X) is a snake-shaped collision avoidance track.

And step 3: the vehicle is decelerated to the speed of 60km/h

And 4, step 4: implementing nonlinear robust trajectory tracking control, and realizing the following method:

step 4.1: and giving parameters of a vehicle dynamic model, and establishing a trajectory tracking dynamic model. Example (c): m is 1385 kg; i is_z＝2065kg·m²；l_f＝1.114m；l_r＝1.436m；

Step 4.2: designing a front wheel steering expression and a rear wheel steering expression, wherein the parameters to be determined are determined one by one as follows;

step 4.3: designing a membership function and a value range of uncertainty:

step 4.4: determining an adjustable parameter range gamma epsilon (0, infinity); e [2, + ∞);

step 4.5: obtaining the current system state x, and taking the weight matrix as

Thus, the cost function J is obtained through D mapping operation_stabilityAnd J_track。

Step 4.6: and (4) combining the results of the step 4.4 and the step 4.5, solving to obtain the optimal solution of the two-person cooperative game: (gamma. rays)^*,∈^*)。

Step 4.7, acquiring a current system state x, and taking an undetermined parameter alpha as beta as 0.15; and (gamma)^*,∈^*) And substituting the values into a robust control law expression to obtain front and rear wheel steering angle values, wherein the obtained collision avoidance effect is shown in fig. 5.

And 5: and after the collision avoidance process is finished, the vehicle is switched to the self-adaptive cruise state.

Although the present application has been disclosed in detail with reference to the accompanying drawings, it is to be understood that such description is merely illustrative and not restrictive of the application of the present application. The scope of the present application is defined by the appended claims and may include various modifications, adaptations, and equivalents of the invention without departing from the scope and spirit of the application.

Claims (3)

1. An automatic driving automobile collision avoidance control method based on a cooperation type game is characterized by comprising the following steps:

step 1, detecting the surrounding environment through vehicle-mounted sensing equipment, judging whether obstacles exist around, if not, continuing to detect, and if so, entering step 2;

step 2, combining the position information of the obstacle obtained in the step 1, automatically driving the automobile to select a series of safe track points, and fitting the track points by utilizing a Bezier curve so as to generate a collision avoidance track;

step 3, decelerating the vehicle to a safe collision avoidance vehicle speed;

step 4, implementing nonlinear robust trajectory tracking control, and inhibiting the influence of curvature radius change on tracking performance;

step 4.1, establishing a track tracking dynamic model based on a two-degree-of-freedom vehicle transverse dynamic equation, wherein the dynamic model is simplified as the following form:

wherein the content of the first and second substances,

wherein the content of the first and second substances,

u＝[δ_f δ_r]^T，σ₁＝2(C_f+C_r)；σ₂＝-2(l_fC_f-l_rC_r)；

in the formula:

is the course angle deviation; e.g. of the type_yIs the lateral displacement deviation;

is the lateral velocity deviation;

is yaw angular velocity deviation; v. of_xIs the longitudinal speed of the vehicle; v. of_yIs the vehicle lateral velocity; c. C_RIs the radius of curvature of the path; m is the mass of the automobile; c_fFront wheel cornering stiffness; c_rIs rear wheel cornering stiffness; l_fThe distance from the center of mass of the automobile to the front axle; l_rThe distance from the mass center of the automobile to the rear axle; i is_zIs horizontal swinging moment of inertia; delta_fIs the angle of rotation of the front wheel, delta_rFor rear wheel steering angle, w_eFor modeling errors due to speed fluctuations

Step 4.2, designing front wheel steering angle input and rear wheel steering angle input by using robust optimal control according to the dynamic model in the step 4.1:

u＝-R^-1B^TPx+γR^-1B^TPx||B^TPx||^∈-2(αx+β)^∈

wherein P is an algebraic Riccati (Riccati) equation

A^TP+PA-2PBR^-1B^TP+Q＝0

The solution of (1);

in the formula: r ═ Q ═ I^4×4Is an identity matrix; alpha and beta are selected constants; gamma and epsilon are adjustable coefficients of the controller; the matrix A, B is the matrix defined in step 4.1;

4.3, representing system uncertainty by using a fuzzy set, and designing a corresponding membership function;

step 4.4, determining the value range of the adjustable parameters gamma and epsilon in the step 4.2

γ_min≤γ≤γ_max

∈_min≤∈≤∈_max

Wherein, γ_min、γ_max、∈_min、∈_maxThe known constants respectively represent the minimum value and the maximum value of the two adjustable parameters;

step 4.5, utilizing the Lyapunov function to find the boundary of the system performance V, namely

V＝x^TPx≤η₁(γ，∈，t)+η₂(γ，∈)

Wherein the content of the first and second substances,

p is the matrix P obtained in step 4.2

η₂(γ，∈)＝κΞ

Wherein xi is a second intermediate variable expressed by

In the formula: k is a predetermined constant, t₀The time when the observer starts to observe;

is t₀The system performance at time can be represented by t₀Calculating the system state at the moment;

is a first intermediate variable obtained during the process of scaling to obtain the boundary of V;

respectively corresponding the tracking cost and the stability cost with the stability performance and the accumulated transient performance to obtain corresponding cost functions, and further solving a corresponding game optimization problem;

step 4.6, solving the two-person cooperative game problem consisting of the step 4.4 and the step 4.5, wherein the optimal solution of the two-person cooperative game problem is pareto optimal to obtain the optimal value gamma of the adjustable parameter^*，∈^*(ii) a The set of optimal values gamma^*，∈^*Substituting the control law expression in the step 4.2 to obtain front and rear wheel rotation angles corresponding to pareto optima;

step 4.7, the input commands of the front wheel steering angle and the rear wheel steering angle in the step 4.5 are respectively sent to the front axle steering mechanism and the rear axle steering mechanism, and the corresponding commands are executed by the corresponding steering mechanisms; and 5, after the collision avoidance process is finished, the vehicle is switched to the self-adaptive cruise state.

2. The cooperative game-based automatic driving automobile collision avoidance control method according to claim 1, characterized in that:

in step 1, the vehicle-mounted sensing device selects a vehicle-mounted camera and a radar.

3. The cooperative game-based automatic driving automobile collision avoidance control method according to claim 1, wherein in step 4.5, the cost function is as follows:

wherein, J_stabilityCost for stability; j. the design is a square_trackTo track costs; d is the mapping operation of the fuzzy number and the real number.

Images