CN115214672A

CN115214672A - Automatic driving type human decision-making, planning and controlling method considering workshop interaction

Info

Publication number: CN115214672A
Application number: CN202210978790.5A
Authority: CN
Inventors: 杨洲; 黎攀; 卢祥伟; 蒲雷; 胡海玉; 叶明�
Original assignee: Chongqing Qingyan Institute Of Technology Intelligent Control Technology Co ltd
Current assignee: Chongqing Qingyan Institute Of Technology Intelligent Control Technology Co ltd
Priority date: 2022-08-16
Filing date: 2022-08-16
Publication date: 2022-10-21

Abstract

The invention discloses an automatic driving type human decision-making, planning and controlling method considering workshop interaction behaviors. The method comprises the following steps: establishing a driver style dynamic recognition model, and recognizing the driving style of a week vehicle; designing a decision cost function considering traffic safety, traffic efficiency and driving comfort; establishing a channel changing decision model considering the driving styles of two game parties based on a complete information non-cooperative game theory, and solving the dynamic interaction behavior and decision making between the two game parties by introducing a Stackelberg game; establishing a driving risk field model to complete risk perception of the road environment; a model predictive control algorithm based on a risk field is provided, and an optimal path and a control sequence are solved through an optimizer of a controller, so that path planning and motion control are synchronously realized. The method can process the interactive behaviors in the complex traffic scene, make a reasonable humanoid lane change decision, plan a reasonable collision-free track for the vehicle in real time, and has good application potential.

Description

Automatic driving type human decision-making, planning and controlling method considering workshop interaction

Technical Field

The invention relates to the technical field of automatic driving, in particular to an automatic driving type human decision-making, planning and control method considering workshop interaction.

Background

In a traffic flow environment with mixed future human driving and automatic driving, the lane change decision-making behavior of the automatic driving vehicle is influenced by surrounding vehicles, particularly by uncertainty of driving style of drivers, so that the planning of a lane change track becomes very complicated. How to make a reasonable decision for a vehicle in a complex environment full of uncertainty has been one of the core bottlenecks of an automatic driving decision algorithm. Most of the domestic and overseas research on driving style only considers the driving style of one of two interactive parties. In fact, for two vehicles that interact dynamically during driving, the difference of the driving styles of the two driving parties can have a great influence on the decision-making result. Therefore, when designing an automated driving lane change decision algorithm, the driving styles of interactive vehicles must be considered at the same time to explore the characteristics and rules of the automated driving vehicles with different driving styles in dynamic interaction and decision making.

In addition, if the real-time feedback of the vehicle planned path and motion control is lacked when the decision algorithm is designed, the planned path is easy to lack feasibility. Paths that are not ideal or that are out of range may cause a reduction in vehicle ride safety and ride comfort. Therefore, a method capable of responding to vehicle decisions in real time and performing path planning and motion control is required to be provided, so that the automatically-driven automobile has better flexibility and driving stability.

Disclosure of Invention

In view of the above-mentioned defects in the prior art, the technical problem to be solved by the present invention is to provide an automated driving type human decision-making, planning and control method considering workshop interaction, so as to solve the problems of insufficient rationality of lane change decision-making, path planning and motion control, possible safety risk and the like caused by the fact that an automated driving vehicle fails to sufficiently and quantitatively evaluate the influence of the driving style of surrounding vehicles on the driving behavior of the vehicle. Meanwhile, the invention also explores the characteristics and rules of the automatic driving vehicles with different driving styles in dynamic interaction and decision making, is favorable for realizing the individualized driving of the automatic driving vehicles to different types of drivers, changes the current situation of the human-adapted vehicles into the human-adapted vehicles, and leads the automatic driving vehicles to be capable of achieving the human-like driving.

To achieve the above object, the present invention provides an automated driver-like decision-making, planning and control method considering workshop interaction, characterized by comprising the steps of,

s1, acquiring self, surrounding vehicles and environmental information through vehicle-mounted data acquisition equipment of an automatic driving vehicle;

s2, recognizing the driving style of the surrounding vehicle based on the surrounding vehicle driving data collected in the natural driving scene, and embedding the characteristics of different driving styles into the design of a decision cost function and a driving risk field;

s3, designing decision cost functions for the main vehicle and the rear vehicle, wherein the decision cost functions are considering traffic safety, traffic efficiency and driving comfort;

s4, establishing a channel change decision model considering the driving styles and interactive behaviors of the main vehicle and the rear vehicle based on a complete information non-cooperative game theory, and introducing a Stackelberg game to solve the dynamic interactive behavior and decision making between the main vehicle and the rear vehicle;

s5, establishing a driving risk field model to complete risk perception of the road environment;

and S6, planning a collision-free path change path for the main vehicle in real time by combining the risk field model and the model prediction control algorithm, and synchronously realizing the motion control of the main vehicle.

Further, in step S1, the information collected by the vehicle-mounted data collecting device includes the position, speed, heading angle, maximum speed limit of each lane, and the like of the autonomous driving main vehicle and the rear vehicle.

Further, in step S2, the driving style is quantitatively analyzed, and the driving style is divided into an aggressive driving style, a robust driving style, and a conservative driving style. And establishing a driving style dynamic recognition model to recognize the driving style of surrounding vehicles.

And (3) as optimization, the driving styles of the vehicles of the two game parties are considered in the decision, planning and control module, and the characteristics of different driving styles are embedded into the decision cost function in the step (S3) and the design of the driving risk field in the step (S5).

As optimization, in step S3, based on different driving styles, a decision cost function considering traffic safety, traffic efficiency, and driving comfort is designed;

wherein the master decision cost function is as follows:

in the formula

And

respectively represent the driving safety cost, the driving comfort cost and the traffic efficiency cost of the main vehicle.

And

are respectively corresponding weight coefficients;

in step S3, since the rear vehicle does not consider the lane change behavior, the decision cost function of the rear vehicle is two differences from the main vehicle, and first, the lateral driving safety cost of the rear vehicle is equal to the lateral driving safety cost of the main vehicle; secondly, the cost of ride comfort of the rear vehicle is related only to the longitudinal acceleration.

As optimization, in step S4, the two parties of the game are not independent, and they influence each other to make decisions of the other party;

when the main car has the intention of changing lanes, the main car and the rear car start the game, and the specific process is as follows:

the main vehicle firstly calculates the lowest decision-making cost when changing lanes leftwards and rightwards, selects a road with lower decision-making cost to change lanes and stops the game with the rear vehicle on the lane on the other side;

the main vehicle starts a steering lamp and moves transversely in a tentative mode to remind a lane change request of the vehicle behind the target lane; the rear vehicle of the target lane receives the prompt and calculates the longitudinal acceleration when the current decision cost is the lowest; then, the main vehicle calculates the lane change decision and the longitudinal acceleration when the current decision cost is the lowest again;

the game two-party decision module judges whether the lowest decision cost of the two parties reaches optimal balance, namely whether the minimum value of the decision cost functions of the two parties does not change any more or changes little; if the optimal balance is not achieved, the lowest decision cost of the main car and the rear car of the target lane is repeatedly calculated in sequence until both game parties achieve the optimal balance; if the decision cost functions of both game parties are optimally balanced, the game is ended; the optimal decision-making command is transmitted to the planning and control module of the main vehicle and the rear vehicle of the target lane.

Further, in the step S4, a double-layer genetic evolution algorithm is used to solve the double-layer optimization problem of the master-slave game. Moreover, to meet the real-time performance of the decision, the optimal solution of the game is calculated at each moment.

As an optimization, in step S5, a unified risk field model of the real-time traffic environment is established based on the external dimensions, driving style and road environment of the obstacle vehicle, so as to complete risk perception of the road environment, where the driving risk field model is as follows:

in the formula P _oc 、P _r1 、P _r2 Respectively represent the models of the risk field of the obstacle vehicle, the risk field of the road boundary line and the risk field of the road boundary line, and m, n and z respectively represent the number of the obstacle vehicle, the road boundary line and the road boundary line.

As an optimization, in step S6, a control strategy for coupling path planning and motion control is proposed, and the specific process is as follows:

designing a target function of the controller by utilizing field intensity distribution, transverse distance deviation, course angle deviation and controlled variable increment of a driving risk field based on a vehicle prediction model, and taking controlled variable constraint and dynamic constraint in a control process as constraint conditions; through a typical optimization solving algorithm, the position of the field intensity of the lowest risk field obtained by solving is used as a planned path at the next moment, and the solved optimal front wheel corner enables the vehicle to reach the planned path, so that path planning and motion control are synchronously realized; and carrying out a new round of optimization solution at the next moment, and repeating the steps until the main lane changing is finished.

The invention has the beneficial effects that: the method is based on the Stackelberg game, carries out decision algorithm modeling, then carries out path planning and motion control in real time according to a decision result, and feeds back the vehicle state and the path parameters to the decision module, thereby designing a closed-loop decision framework considering vehicle interaction and driving style. Meanwhile, the driving styles of the vehicles of the two interactive parties are considered, the characteristics of different driving styles are embedded into the design of a decision cost function and a driving risk field model, and the influence of different driving styles on decision and path planning is researched. The decision model can process the interactive behavior in the complex traffic scene to make a reasonable man-like lane change decision, and the model prediction control controller based on the driving risk field can plan a reasonable collision-free track for the main vehicle in real time, so that the decision model has good application potential.

Drawings

Fig. 1 is a schematic diagram of a human-like decision, planning and control of an autonomous driving vehicle according to an embodiment of the present invention.

Fig. 2 is a schematic diagram of a gaming scenario provided in an embodiment of the present invention.

Fig. 3 is a flow chart of the lane change decision provided in the embodiment of the present invention.

Fig. 4 is a schematic view of a driving risk yard according to an embodiment of the present invention.

Fig. 5 is a flow chart of path planning and motion control provided by the embodiment of the present invention.

Fig. 6 is a schematic diagram of simulation of decision making, planning and control in a game scenario provided by an embodiment of the present invention.

Detailed Description

The present invention will be further described with reference to the accompanying drawings and examples, wherein the terms "upper", "lower", "left", "right", "inner", "outer", and the like, as used herein, refer to an orientation or positional relationship indicated in the drawings, which is for convenience and simplicity of description, and does not indicate or imply that the referenced devices or components must be in a particular orientation, constructed and operated in a particular manner, and thus should not be construed as limiting the present invention. The terms "first," "second," "third," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

The embodiment of the invention provides an automatic driving person decision-making, planning and controlling method considering workshop interaction; specifically, as shown in fig. 1, a schematic diagram of automated vehicle-like human decision making, planning and control; the method comprises the following steps:

the vehicle-mounted data acquisition equipment in the step S1 comprises a positioning module, a communication module, a speed sensor and a radar. The collected information comprises the position, speed and course angle of the main vehicle and the rear vehicle, the maximum speed limit of each lane and the like.

And S2, recognizing the driving style of the surrounding vehicle based on the surrounding vehicle driving data collected in the natural driving scene. Embedding the characteristics of different driving styles into the design of a decision cost function and a driving risk field;

the specific steps of the driver style recognition in the step S2 include: collecting vehicle driving data in a natural driving scene, cleaning the data, removing abnormal values, extracting characteristics and reducing dimensions of the data to obtain high-quality automatic driving road test data; clustering the road test data subjected to dimension reduction by adopting a random forest classification algorithm, and defining the driving style of a driver according to a clustering result; and (3) integrating characteristic variables of a driver and characteristic differences of the driving style of the driver, establishing a driving style recognition model based on an XGB algorithm, training and testing the driving style recognition model, and finishing effective recognition of the driving style of the vehicle.

S3, designing decision cost functions for the main vehicle and the rear vehicle, wherein the decision cost functions are used for considering traffic safety, traffic efficiency and driving comfort;

in the step S3, a total main vehicle decision cost function is calculated according to the traffic safety cost, the traffic efficiency cost, the driving comfort cost, and the corresponding weight coefficients as follows:

in the formula

And

And

are the corresponding weight coefficients.

The driving safety cost function includes two parts, longitudinal and lateral safety costs, defined as follows:

in the formula

And

representing the longitudinal and lateral security costs, respectively. Gamma represents the main lane change decision result, and gamma belongs to the { -1,0,1} to represent { lane change to the left, lane keeping and lane change to the right }.

Relative distance and relative speed between the main vehicle and the vehicle ahead of the current lane are defined as follows:

in the formula

Indicating the longitudinal velocity of the host vehicle and the vehicle behind the target lane,

and

is the position coordinates of the rear vehicle and the main vehicle in the target lane.

Are weighting factors for relative velocity and relative distance. Iv is the safety factor of the liquid crystal,

is a minimum value to avoid a denominator of 0.κ ∈ {1,2,3}, indicating lane 1, lane 2, lane 3}.

Relative distance and relative speed of the host vehicle and the rear vehicle of the target lane are defined as follows:

in the formula

Representing the longitudinal speed of the vehicle behind the target lane,

is the position coordinates of the vehicle behind the target lane.

Are weighting factors for relative velocity and relative distance.

In the formula a _x And a _y Respectively the longitudinal and transverse speeds of the host vehicle during travel,

are the corresponding weight coefficients. Gamma represents the main lane change decision result, and gamma belongs to the { -1,0,1} to represent { lane change to the left, lane keeping and lane change to the right }.

In the formula

Represents the highest speed of the vehicle in the lane,

representing the longitudinal speeds of the front vehicle and the main vehicle in the current lane, deltas representing the relative distance between the main vehicle and the rear vehicle in the target lane, d _m Which represents the minimum safe distance between the mobile station and the mobile station,

the decision cost function of the rear vehicle of the target lane has similar structure and expression with the main vehicle, and has two differences. Firstly, the rear vehicle of the target lane only considers acceleration and deceleration and does not consider lane change behavior, so that the transverse driving safety cost of the rear vehicle of the target lane is equal to that of the main vehicle, namely:

in addition, the driving comfort cost of the vehicle behind the target lane is only related to the longitudinal acceleration, namely:

s4, establishing a lane change decision model considering the driving styles and interactive behaviors of the main vehicle and the rear vehicle based on a complete information non-cooperative game theory, and introducing a Stackelberg game to solve the dynamic interactive behaviors and decision making between the main vehicle and the rear vehicle;

in step S4, fig. 2 is a schematic diagram of a game scene provided by the embodiment of the present invention, as shown in the figure, the main vehicle and the front vehicle 2 both travel in the middle lane, the front vehicle 2 travels slowly, the main vehicle must make a decision to decelerate and follow the front vehicle 2 or change lanes. If lane changing is considered, the details of the target lane must be considered to determine which lane change is more cost effective and the risk value is more favorable for the host vehicle to travel.

Fig. 3 is a flow chart of lane change decision provided by the embodiment of the present invention, as shown in the figure, when the main car has a lane change intention, the main car and the rear car start to play games, and the specific sub-steps are as follows:

and S41, calculating the lowest decision cost function value when the main vehicle switches the lane leftwards and rightwards by combining a series of conditions such as driving styles, relative positions, relative speeds and the like of both sides of the game.

And S42, the main vehicle selects a road with a lower decision cost function to change the lane, and stops the game with the rear vehicle on the other lane.

S43, the main vehicle starts a steering lamp and moves transversely tentatively to remind the vehicle of changing the lane behind the target lane.

And S44, calculating the longitudinal acceleration when the decision cost of the rear vehicle of the target lane is the lowest at the current moment by combining a series of conditions such as driving styles, relative positions and relative speeds of both game parties.

And S45, calculating the lane change decision and the longitudinal acceleration when the decision cost of the main vehicle is the lowest at the current moment again by the main vehicle.

And S46, judging whether the lowest decision cost function values of the game parties reach optimal balance by the decision module of the game parties, namely whether the minimum value of the decision cost functions of the game parties does not change any more or changes little, and if not, circulating the substeps S44 and S45. Until both gaming parties reach optimal equilibrium. And if the decision cost functions of both parties of the game are optimally balanced, the game is ended. The optimal decision instructions are transmitted to the respective planning and control modules by the main vehicle and the rear vehicle of the target lane.

In this embodiment, both vehicles, interacting dynamically at decision time, try to minimize decision cost. The difference of the driving styles of the two parties has great influence on the decision result. When the main vehicle and the rear vehicle of the target lane with different driving styles are in interaction to prepare for lane changing, if the driving style of the rear vehicle of the target lane is conservative, the safety of the main vehicle can be worried about, the main vehicle is allowed to change lanes by deceleration, and the main vehicle is prevented from changing lanes by acceleration of an aggressive vehicle with high probability, so that the driving space of the main vehicle is ensured. Similarly, the host vehicles with different driving styles interact with the vehicles with the same driving style, and the decision thereof can be greatly different. The optimal decision of the main vehicle and the rear vehicle of the target lane is shown in the following table:

TABLE 1

It should be noted that for the above-mentioned two-layer optimization problem of the master-slave game, the precision of the strategy solution is important, and the present invention adopts a two-layer genetic evolution algorithm to solve. And, to meet the real-time nature of the decision, the optimal solution for the game is computed at each time.

in step S5, a risk field model of an obstacle vehicle, a risk field model of a road boundary line, and a risk field model of a comprehensive road and all obstacle vehicles are established, so that a mathematical model of a uniformly described risk field of a traffic environment can be obtained. The method comprises the following specific steps:

for the host coordinate points (X, Y), a risk field model of surrounding obstacle vehicles is defined as:

P _oc (X,Y)＝A _oc ·e ^φ (8)

in the formula A _oc Represents the maximum value of the risk field of the obstacle vehicle, (X) _oc ,Y _oc ) Is the barycentric coordinate of the obstacle vehicle, epsilon is the coefficient for adjusting the shape of the crest of the risk field of the obstacle vehicle, L _oc ，W _oc Respectively representing the length and width of the vehicle, k _x ，k _y Adjustment coefficients for the length and width of the obstacle vehicle, respectively, and ρ represents a high-order coefficient.

The risk field model of the road is represented as:

P _r1 ＝A _r1 ·exp((-d _r1 +d _c +W _oc /2)·h ₁ ) (9)

in the formula, P _r1 ,P _r2 Risk field values in (X, Y) coordinates for road boundaries and lane boundaries, d _r1 ,d _r2 Is the closest distance of the vehicle to the road boundary and lane boundary, d _c Is a safety threshold, σ _line Is the risk field adjustment coefficient of the road boundary line. Epsilon and h ₁ Is the risk field adjustment factor for the lane boundary.

The traffic environment unified description risk field mathematical model can be obtained by integrating the road boundary line, the lane boundary line and the risk field models of all the obstacle vehicles:

where m, n, z represent the number of obstacle vehicles, road boundaries and lane boundaries, respectively.

Fig. 4 provided by the embodiment of the present invention is a contour diagram of an example scene containing obstacle vehicles with different driving styles and corresponding driving risk fields. The risk level of each position of the road can be seen from the graph; on the basis, the safety degree of each position of the road is calculated efficiently by the main vehicle in a complex traffic scene, a high-risk area is avoided, and a low-risk area is selected as much as possible to drive, so that behavior decision and path planning are performed safely and efficiently, and the traffic efficiency is improved.

And S6, planning a collision-free lane change path for the main vehicle in real time by combining the risk field model and the model prediction control algorithm, and synchronously realizing the motion control of the main vehicle.

In step S6, fig. 4 is a flow chart of path planning and motion control provided in the embodiment of the present invention, and as shown in the figure, based on a vehicle kinematics model, a risk level corresponding to a position of a host vehicle is combined with a cost function of model prediction control, so that path planning and motion control can be synchronously implemented. The specific substeps are as follows:

s61, establishing an automobile two-degree-of-freedom kinematic model, and performing linear discretization treatment on the automobile two-degree-of-freedom kinematic model to obtain the following model:

θ＝α+β (13)

β＝arctan(l _r /(l _r +l _f ).tanδ _f ) (14)

wherein the system state vector is

The controlled variable is u = delta _f Wherein δ _f ,

β,θ,α _x ,v _x Respectively, the vehicle front wheel rotation angle, the yaw angle, the slip angle, the course angle, the longitudinal acceleration and the longitudinal speed. (X, Y) is the vehicle coordinate position. Δ t is the discrete state time step, and χ (k + 1) represents the system state of the discrete system at step k + 1. l _r ,l _f The distances of the centroid to the front and rear axes, respectively.

S62, designing a target function of the controller by utilizing field intensity distribution, transverse distance deviation, course angle deviation and controlled variable increment of the traffic risk field, and taking controlled variable constraint and dynamic constraint in the control process as constraint conditions. The objective function and constraint conditions are as follows:

in the formula of U _PF (i) The field strength of (X (i | k), Y (i | k)) representing the predicted position of the host vehicle within the prediction time domain can be solved by equation (11) in step S5, and Δ Y (k) represents the predicted position of the host vehicle and the lane center line Y _ref The error in the lateral distance of (a),

indicating the predicted position heading angle of the host vehicle andwith reference to the error of the heading angle, Δ u (i + k) represents the control increment to be minimized in the control time domain. Q ₁ ～Q ₄ Is the corresponding weight matrix of the cost function. Np and Nc respectively represent a prediction time domain and a control time domain of model predictive control, and Np is more than or equal to Nc.

And S63, solving the position of the field intensity of the lowest risk field as a path at the moment of k +1 through a typical optimization solving algorithm. Meanwhile, an optimal control increment sequence delta u of the vehicle at the moment k is solved ^* (k) Further, the actual controlled variable u (k) at the time k is obtained.

u(k|k)＝u(k-1|k-1)+Δu ^* (k|k) (16)

And S64, updating the state vector of the vehicle based on the formula (16), and performing a new round of optimization solution at the next time step k +1 to reciprocate until the main lane change is finished.

The embodiment of the invention designs a typical expressway lane change scene to verify the feasibility and the effectiveness of the proposed human-like decision, planning and control algorithm, and all the scenes and the algorithms are realized based on an MATLAB program. The lane change success of the main vehicle under the multi-vehicle game condition is verified, and the self-adaptability of the algorithm is also verified. The specific process is as follows:

according to the gaming scenario diagram of fig. 2, the dynamic gaming of the rear vehicle 1 of the target lane and the rear vehicle 2 of the target lane is considered simultaneously when the main vehicle has lane change intention. The embodiment of the invention designs four conditions, and correspondingly sets the driving styles of the main vehicle, the rear vehicle 1 of the target lane and the rear vehicle 2 of the target lane as follows: (robust, aggressive), (aggressive, robust), (robust ), (aggressive, conservative, robust). In order to simplify the simulation scenario, the embodiment assumes that the game with the rear vehicle 2 in the target lane is ended after the main vehicle makes a decision to change lanes to the left, the rear vehicle 2 in the target lane becomes to run at a constant speed, the game with the rear vehicle 1 in the target lane is completed until the lane change is completed, and the same assumption is followed for the lane change to the right.

Fig. 6 provided in the embodiments of the present invention shows a detailed test result. It can be seen that when the target lane rear vehicle 1 and the target lane rear vehicle 2 are both of aggressive driving style, the target lane rear vehicles 1 and 2 both adopt a larger acceleration to prevent the main vehicle from changing lanes, so as to ensure that the space is not occupied by the main vehicle, and the main vehicle also selects lane keeping to decelerate and follow the front vehicle for ensuring the driving safety. If the rear vehicle 1 of the target lane is aggressive and the rear vehicle 2 of the target lane is stable, the rear vehicle 1 of the target lane is more aggressive because of paying more attention to the running efficiency and preventing the main vehicle from changing lanes by adopting larger acceleration, so that the decision-making cost of the game of the main vehicle and the rear vehicle 2 of the target lane is lower than that of the game of the main vehicle and the rear vehicle 1 of the target lane, and the lane change to the right is safer and more comfortable. The method verifies that the decision, planning and control algorithm provided by the invention can make reasonable lane change decision and smoothly complete the lane change process, and effectively simulates the decision and interactive behavior of human drivers in real traffic scenes.

The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations can be devised by those skilled in the art in light of the above teachings. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.

Claims

1. An automatic driver-like decision-making, planning and controlling method considering workshop interaction is characterized by comprising the following steps,

2. A method for automated driver-like decision-making, planning and control taking into account vehicle-to-vehicle interactions as claimed in claim 1, wherein: and (4) simultaneously considering the driving styles of the vehicles of the game parties in a decision-making, planning and control module, and embedding the characteristics of different driving styles into the decision-making cost function in the step (S3) and the design of the driving risk field in the step (S5).

3. A method for automated driver-like decision, planning and control taking into account plant interactions, as claimed in claim 1, wherein: in the step S3, based on different driving styles, a decision cost function considering traffic safety, traffic efficiency and driving comfort is designed;

wherein the master decision cost function is as follows:

in the formula

And

respectively represent the driving safety cost, the driving comfort cost and the passing efficiency cost of the main vehicle.

And

are respectively corresponding weight coefficients;

4. A method for automated driver-like decision, planning and control taking into account plant interactions, as claimed in claim 1, wherein:

in the step S4, the two game parties are not independent and influence the decision making of the other party;

the main vehicle firstly calculates the lowest decision-making cost when changing lanes leftwards and rightwards, selects a road with lower decision-making cost to change lanes and stops the game with a rear vehicle on the lane on the other side;

the main vehicle starts a steering lamp and moves transversely tentatively to prompt a lane change request of the rear vehicle of the target lane; the rear vehicle of the target lane receives the prompt and calculates the longitudinal acceleration when the current decision cost is the lowest; then, the main vehicle calculates the lane change decision and the longitudinal acceleration when the current decision cost is the lowest again;

the game two-party decision module judges whether the lowest decision cost of the two parties reaches optimal balance, namely whether the minimum value of decision cost functions of the two parties does not change any more or changes little; if the optimal balance is not achieved, the lowest decision cost of the main vehicle and the rear vehicle of the target lane is repeatedly calculated in sequence until both game parties achieve the optimal balance; if the decision cost functions of both game parties are optimally balanced, the game is ended; the optimal decision-making command is transmitted to respective planning and control modules by the main vehicle and the rear vehicle of the target lane.

5. A method for automated driver-like decision, planning and control taking into account plant interactions, as claimed in claim 1, wherein: in step S5, a unified risk field model of the real-time traffic environment is established based on the overall dimension, the driving style and the road environment of the obstacle vehicle, so as to complete the risk perception of the road environment, wherein the driving risk field model comprises the following steps:

6. A method for automated driver-like decision, planning and control taking into account plant interactions, as claimed in claim 1, wherein: in step S6, a control strategy for coupled path planning and motion control is proposed, which specifically includes the following processes:

designing a target function of the controller by utilizing field intensity distribution, transverse distance deviation, course angle deviation and controlled variable increment of a driving risk field based on a vehicle prediction model, and taking controlled variable constraint and dynamic constraint in a control process as constraint conditions;

through a typical optimization solving algorithm, the position of the field intensity of the lowest risk field obtained by solving is used as a planned path at the next moment, and the solved optimal front wheel corner enables the vehicle to reach the planned path, so that path planning and motion control are synchronously realized; and carrying out a new round of optimization solution at the next moment, and repeating the steps until the main lane changing is finished.