CN117075473A - Multi-vehicle collaborative decision-making method in man-machine mixed driving environment - Google Patents

Multi-vehicle collaborative decision-making method in man-machine mixed driving environment Download PDF

Info

Publication number
CN117075473A
CN117075473A CN202311030796.0A CN202311030796A CN117075473A CN 117075473 A CN117075473 A CN 117075473A CN 202311030796 A CN202311030796 A CN 202311030796A CN 117075473 A CN117075473 A CN 117075473A
Authority
CN
China
Prior art keywords
cav
vehicle
model
value
interaction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202311030796.0A
Other languages
Chinese (zh)
Inventor
孙剑
杭鹏
崔一鸣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongji University
Original Assignee
Tongji University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tongji University filed Critical Tongji University
Priority to CN202311030796.0A priority Critical patent/CN117075473A/en
Publication of CN117075473A publication Critical patent/CN117075473A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The application relates to a multi-vehicle collaborative decision-making method in a man-machine mixed driving environment, which comprises the steps of acquiring vehicle attributes and vehicle states of adjacent conflict areas, updating a game vehicle list based on a game range, and switching corresponding models according to the types and the numbers of vehicles in the list to carry out strategy solving; and deducing the action strategy of the HV through the CAV and HV interaction model, the CAV collaborative decision model and the man-machine mixed driving interaction and collaborative model coupled with the CAV and HV interaction model respectively, and finally solving the CAV optimal strategy. Compared with the prior art, the method considers the cooperation between CAVs and the interaction between CAVs and HV, adjusts and adapts to the uncertainty in the dynamic interaction process of the human driving vehicle in real time through the optimal parameters, and realizes the efficient and safe passing of vehicles in the mixed driving scene; in addition, the method can also adapt to the personalized driving requirement of CAV.

Description

Multi-vehicle collaborative decision-making method in man-machine mixed driving environment
Technical Field
The application relates to the field of automatic driving, in particular to a multi-vehicle collaborative decision-making method in a man-machine mixed driving environment.
Background
With the rapid development of autopilot vehicles, human-driving vehicles (HV) and internet autopilot vehicles (Connected and Automated Vehicle, CAV) will share road resources in the form of man-machine traffic streams for a long time in the future. How to ensure safety and improve efficiency in a man-machine mixed driving scene is an important issue of future automatic driving research. The intelligent automatic driving of the bicycle has environmental perception limitation in a complex interaction environment, and interaction safety is difficult to ensure; relatively conservative interaction behavior is often shown, and interaction efficiency is difficult to ensure; optimal decisions are difficult to make when facing complex scenes, and the realistic problems of high equipment cost and the like exist. Along with the improvement of CAV permeability, the driving conflict in the complex environment can be solved in a vehicle-vehicle cooperative mode, and safety and efficiency are improved.
However, at present, vehicle-vehicle coordination related researches are mostly researched under a pure CAV environment or interaction of CAV and HV is researched from a single vehicle intelligent view angle, and comprehensive consideration of interaction of CAV and HV in a mixed driving scene and cooperation between CAV is lacking. Most of the current researches only take HV as a dynamic obstacle or simply express the HV by using a following model, the dynamic interaction characteristics of HV cannot be effectively captured, and the interactivity of CAV and HV cannot be well reflected. In addition, the existing research does not consider individual demand differences among collaborative CAVs, and the obtained model is difficult to meet the individual driving demands of the CAVs.
Disclosure of Invention
The application aims to overcome the defects of the prior art, provides a multi-vehicle collaborative decision-making method in a man-machine mixed driving environment, and is a collaborative decision-making method capable of realizing cooperation between CAV workshops and interaction coupling between CAV and HV. In the specific model method design, CAV personalized driving preference is realized through a profit function, model dynamic adjustment is realized by adopting a twin game mode according to the uncertainty of HV interaction, and then model solving is carried out to obtain an optimal vehicle action decision result.
The aim of the application can be achieved by the following technical scheme:
a multi-vehicle collaborative decision-making method in a man-machine mixed driving environment generally relates to the construction of three decision-making models: a CAV and HV interaction model, a CAV collaborative decision model, a man-machine hybrid interaction and collaboration model; the general reference includes the following steps:
s1, updating vehicle attribute and state information: inputting attribute information of all vehicles in a scene, including vehicle types, kinematic parameters and the like; the vehicle state information includes a vehicle position, a vehicle speed, an acceleration, and the like.
S2, updating a game object list and switching a model: judging vehicles entering the near conflict area and the conflict area according to the vehicle state, and updating interaction and collaboration vehicles contained in the game object list; and determining the actually executed model according to the type and the number of the game objects.
S3, interaction modeling step of CAV and HV: and establishing a model of interaction of CAV and HV.
S4 CAV collaborative decision modeling step: and establishing a model of collaborative decision-making in the CAV vehicle.
S5, man-machine hybrid interaction collaborative modeling step: and (3) coupling the CAV and HV interaction model and the CAV collaborative decision model, carrying out a man-machine mixed driving interaction collaborative model, carrying out model solving, predicting the action of the HV in the game object list, and solving the optimal action strategy of the CAV, namely the acceleration.
S6, updating the vehicle state: the CAV realizes vehicle state update through a kinematic formula according to the output acceleration decision result, and solves the update speed and position; acceleration, speed, position of HV are re-acquired from the environment.
Compared with the prior art, the application has the following advantages:
(1) The application provides the method for simultaneously considering the cooperation between the networked automatic driving vehicles of the mixed driving scene and the interaction between the networked automatic driving vehicles and the human driving vehicles, which is more suitable for the real mixed driving scene and can effectively solve the safety efficiency problem of intelligent automatic driving of the single vehicle.
(2) The application fully considers the uncertainty in the dynamic interaction process of the human driving vehicle through the identification and dynamic adjustment of the weight related parameters of the human driving vehicle income function components.
(3) The cooperative game-based cooperative decision model of the online automatic driving vehicle can also adapt to the actual personalized driving demand of the online automatic driving vehicle.
Drawings
FIG. 1 is a flow chart of the method of the present application
FIG. 2 is a flowchart of the algorithm of the application S3
FIG. 3 is a flowchart of the algorithm of the application S4
FIG. 4 is a schematic diagram of the data flow relationship between key steps in the algorithm of the present application
FIG. 5 is a diagram of an embodiment scenario
Detailed Description
Definition of conflict area, near conflict area:
the conflict area is the area containing conflict points generated by vehicle track intersection, such as the inside of intersections in urban roads and highways;
the application defines the near conflict area as a distance before the vehicle reaches the conflict area, and the application calculates the range by taking the length of the parking line front parking sight as the near conflict area. The present application considers the near-conflict area and vehicles within the conflict area.
The stopping sight distance is a term, and refers to the shortest driving distance required by braking and stopping when a vehicle encounters a front obstacle during driving on the same lane.
The technical scheme of the application is suitable for urban scenes, expressway scenes and the like. The method of the application detects and collects the near conflict area and the vehicle state information in the conflict area through the road side equipment of the network connection, and performs policy calculation and decision information release. Technically feasible.
The technical scheme of the application is further described below with reference to the accompanying drawings and the embodiments.
Examples
A multi-vehicle collaborative decision-making method in a man-machine mixed driving environment generally relates to the construction of three decision-making models: a CAV and HV interaction model, a CAV collaborative decision model, a man-machine hybrid interaction and collaboration model; the general reference includes the following steps:
s1, updating vehicle attribute and state information: inputting attribute information of all vehicles in a scene, including vehicle types, kinematic parameters and the like; the vehicle state information includes a vehicle position, a vehicle speed, an acceleration, and the like.
S2, updating a game object list and switching a model: judging vehicles entering the near conflict area and the conflict area according to the vehicle state, and updating interaction and collaboration vehicles contained in the game object list; and determining the actually executed model according to the type and the number of the game objects.
S3, interaction modeling step of CAV and HV: and establishing a model of interaction of CAV and HV.
S4 CAV collaborative decision modeling step: and establishing a model of collaborative decision-making in the CAV vehicle.
S5, man-machine hybrid interaction collaborative modeling step: and (3) coupling the CAV and HV interaction model and the CAV collaborative decision model, carrying out a man-machine mixed driving interaction collaborative model, carrying out model solving, predicting the action of the HV in the game object list, and solving the optimal action strategy of the CAV, namely the acceleration.
S6, updating the vehicle state: the CAV realizes vehicle state update through a kinematic formula according to the output acceleration decision result, and solves the update speed and position; acceleration, speed, position of HV are re-acquired from the environment.
Further, in S1, the step of updating the vehicle attribute and the state information specifically includes:
acquiring vehicle type information of each vehicle from the environment, namely, the vehicle is CAV or HV; kinematically related parameters such as maximum acceleration, maximum deceleration, maximum speed, etc.
Collecting the current acceleration, speed and position of HV in the updated environment in each time step; and the CAV updates the speed and the position of the vehicle according to the optimal acceleration output by the model and the kinematic formula. The vehicle state information is used as an input to the model.
Further, in S2, the steps of updating the game object list and switching the model are specifically:
all vehicles s= { S in the scene 1 ,S 2 ,…,S k ,…,S K I CAV, denoted m= { M }, is contained therein 1 ,M 2 ,…,M i ,…,M I And J vehicles HV, denoted as N= { N 1 ,N 2 ,…,N j ,…,N J }。
The game scope, i.e., the near conflict area and within the conflict area, is defined. Vehicles entering the game range are added to the game object list; the vehicle exits the game object list after passing through the conflict area. Vehicles in game object listWherein P CAV are contained, expressed as +.> Q vehicle HV, denoted->
Selecting and switching the model according to the types of vehicles and the number of various vehicles in the game object list,
when no CAV vehicle or only one vehicle exists in the game object list, the vehicle can freely pass through without considering interaction cooperation;
when the vehicles in the game object list are all CAV vehicles, only CAV collaborative decisions are carried out, and a CAV collaborative decision model is selected for carrying out vehicle optimal acceleration strategy solving;
when the vehicles in the game object list contain CAV and HV at the same time and only one CAV exists, only a CAV and HV interaction decision is needed, and a CAV and HV interaction model is selected for carrying out CAV optimal acceleration strategy solving;
when the vehicles in the game object list contain CAV and HV simultaneously and contain a plurality of CAV, the CAV collaborative decision model and the CAV and HV interactive decision model are required to be coupled to realize man-machine hybrid interactive collaboration.
S3, interaction modeling step of CAV and HV: and establishing a model of interaction of CAV and HV. The interaction modeling of HV and a plurality of CAVs can be realized by taking the CAVs as a whole. On the basis of solving the HV actions through model prediction, the overall benefit of the CAV is optimized.
Further, in S3, the interaction model in the step of modeling CAV and HV interaction is specifically:
based on game theory, a non-cooperative model is constructed, and CAV and HV are respectively divided into a leader and a follower.
Design of the CAV benefit function f taking into consideration the requirements of safety, efficiency and comfort in the running process of the vehicle p And a benefit function f of HV q . The two gain function forms are consistent, and the individual gain value of the vehicle is the weighted summation of the three.
In the method, in the process of the application,is the safety index of CAV, +.>Is an efficiency index of CAV, < >>Is a CAV comfort index.The weight values of the three indexes are respectively. Likewise, a +>Safety, efficiency, comfort index of HV, respectively,>are the respective weight values. The specific index calculation is shown in the following formula.
Wherein DeltaL p Is the distance between the CAV vehicle p and the conflict point of the HV vehicle q, v p (k+1) is the speed of the CAV vehicle p at time k+1, a p (k)、a p (k+1) is the acceleration of the CAV vehicle p at the k-th and k+1-th times, respectively. ΔL q V is the distance between the q distance of the HV vehicle and the conflict point of the CAV vehicle p q (k+1) is the speed of the HV vehicle q at time k+1, a q (k)、a q (k+1) is the acceleration of the HV vehicle q at the kth and k+1 times, respectively.
On the basis of the weight value of the previous moment, the safety and efficiency weight value of the HV is updated in real time according to the similarity degree of the predicted acceleration of the CAV to the actual acceleration of the CAV. In order to reduce the complexity of model parameter updating, the weight value corresponding to the safety and efficiency index is converted into the form of a function of the same parameter theta. The following formula is shown:
where A, B is a constant.
Establishing an interaction model, and predicting the possibility of HV under the condition that the CAV makes a certain acceleration action decision by itself according to the profit function of HVOptimum acceleration to be takenOn the basis, self-income is calculated, and the self-income is calculated in CAV self-acceleration strategy spaceOptimal solution for maximizing self-income by searching in->The following formula is shown:
s.t.
in the method, in the process of the application,CAV object list M for participating in gaming g Is +.> Policy space for HV at this time->Middle->Income of corresponding HV, ++>To enable->A maximum optimal acceleration strategy; />To get +.>HV adopts optimal acceleration strategy +.>The corresponding CAV overall benefit, +.>A CAV optimal acceleration strategy to maximize the benefit. At the same time (I)>Is->Minimum and maximum of>Is->Minimum and maximum values of (2);is a safety constraint, wherein->For the time when the CAV vehicle p arrives at the conflict point according to the current state, the +.>The time for the HV vehicle q to reach the conflict point is set according to the current state. Optimal acceleration strategy will be achieved at CAV +.>HV optimal acceleration strategy determined on the basis of the model>As a model in->The prediction of HV is described as +.>For use in subsequent twinning games.
In order to achieve more accurate capturing of the HV motion real-time interaction uncertainty, updating of HV weight related parameters theta is achieved in a twin game mode.
At the initial moment, weight values corresponding to three CAV indexes are obtained based on natural driving data calibration And CAV fixed parameter value +.>And the corresponding HV initial weight value related parameter is theta 0 . At theta 0 R (r is set to be odd number) values are discretized again in the range of the definition domain to form an initial theta value space theta 0 。
In which the value space Θ 0 The specific value taking method of each theta value is as follows: at the initial moment, in order to take values as comprehensively as possible, the range of a value taking space is determined as followsDividing the minimum value in the value space>Optimal parameter of initial moment->Maximum value->In addition, at [0, θ 0 ]、/>Respectively equidistantly taking (r-3)/2 values, adding them into the value space, and forming value space theta with r number of elements 0 。
Twin game solving: at each subsequent time step, the value space theta is based on the last moment k Each theta value in the model is solved on HV through a CAV and HV interaction model to obtain a series of HV optimal acceleration strategies corresponding to the theta values
Fitness calculation of optimal parametersAnd (3) determining: calculating Θ k HV optimal acceleration strategy corresponding to each theta value in the systemIs +.>Is shown in the following formula.
The larger the alpha value, i.e. the smaller the difference between the calculated value and the actual value, the higher the fitness. The corresponding fitness alpha in all theta values is highest and is used as the optimal parameter at the next moment
Value list theta k+1 Updating: according to the optimal parameters at the next momentAnd the corresponding fitness alpha * The value number r at the next time is checked. In this embodiment, the rule is formulated as shown in table 1.
TABLE 1 correspondence between the number of twin values and fitness
When the adaptability reaches a certain level, the method is optimalAnd (5) performing model calculation, and not performing twin game. Until the fitness exceeds a level that does not require twinning.
Setting the length of the value interval asMinimum value of value limit->Maximum value->The calculation method is shown in the following formula.
If calculatedMake->Correspondingly->
If calculatedMake->Correspondingly->
Finally, the value space theta of the value of the next moment theta k+1 At the position ofThe nearby discrete values are obtained as shown in the following formula.
S4 CAV collaborative decision modeling step: and establishing a model of collaborative decision-making in the CAV vehicle. The method can be used as a vehicle collaborative decision model in a pure CAV scene, and can also be used as an inner layer model of man-machine hybrid driving interaction collaborative modeling to realize CAV individual acceleration solution.
Further, in S4, the collaborative decision model in the CAV collaborative decision modeling step is specifically:
and (3) designing a characteristic function of the CAV by considering the requirements of safety, efficiency and comfort in the running process of the vehicle, wherein the individual benefit value of the vehicle is the weighted summation of the three. And the calculation mode of each index is consistent with the CAV and HV interaction model.
Wherein p' is M in the game object list g CAV vehicles other than CAV vehicle p, i.e.The security inside the CAV is calculated. Weight value +.>Can be according to the personalized preference of CAVAnd (5) adjusting.
According to the basic principle of the cooperative game, a characteristic function, an optimization target and constraint conditions are designed, benefits are distributed according to the shape principle, and the optimal action of the vehicle is planned. Shapley is calculated as follows:
in the psi- p Is the Shapley value of CAV vehicle p. s is M g Representing a collection of possible participant-cooperative gaming objects during the computing process. v(s) is the benefit generated by the set of participants for s. S and M g I represents s, M respectively g The number of vehicles involved.
v(s) is the benefit generated by the participant set s, calculated by:
and (5) taking the maximum total profit of the CAV vehicle as a target, and establishing a model. The following formula is shown:
s.t.
in the method, in the process of the application,is a safety constraint between CAVs.
Further, in S5, the interactive decision model in the man-machine hybrid interactive decision modeling step is specifically:
meanwhile, interaction of CAV and HV is considered, cooperation of the CAV is achieved, and a double-layer optimization model is built. The interaction of the outer layer of CAV as a whole with HV is shown in the following formula:
s.t.
in the method, in the process of the application,the overall benefit of CAV is the sum of the benefits of CAV contained in the game object list. Safety restraint examinationSafety between CAV and HV is a concern. The rest of the amounts have the same meaning as the CAV and HV interaction models.
The inner layer is CAV internal cooperation, and the optimal acceleration strategy of the CAV vehicle is obtained by solving according to a CAV cooperation decision modelHowever, in the model, all CAV and HV except for the safety index in the game object list are considered in calculation.
Further, in S6, after an appropriate model is selected according to the game object list to solve the optimal acceleration strategy, the vehicle state is updated.
The state of CAV is expressed and updated according to a kinematic formula:
in the method, in the process of the application,to participate in game CAV vehicle speed, acceleration, position.
The HV real state is used for carrying out information collection again from the environment to obtain the actual position of the HV real stateSpeed->Acceleration->
An example scenario is shown in fig. 5.
The above description is only illustrative of the preferred embodiments of the application and is not intended to limit the scope of the application in any way. Any alterations or modifications of the application, which are obvious to those skilled in the art based on the teachings disclosed above, are intended to be equally effective embodiments, and are intended to be within the scope of the appended claims.

Claims (9)

1. A multi-vehicle collaborative decision-making method in a man-machine mixed driving environment is characterized by generally comprising the steps of constructing three decision-making models: a CAV and HV interaction model, a CAV collaborative decision model, a man-machine hybrid interaction and collaboration model; the general reference includes the following steps:
s1, updating vehicle attribute and state information: inputting attribute information of all vehicles in a scene, wherein the attribute information comprises vehicle types and kinematic parameters; vehicle state information including vehicle position, vehicle speed, acceleration;
s2, updating a game object list and switching a model: judging vehicles entering the near conflict area and the conflict area according to the vehicle state, and updating interaction and collaboration vehicles contained in the game object list; determining an actually executed model according to the types and the quantity of the game objects;
s3, interaction modeling step of CAV and HV: establishing a model of interaction between CAV and HV;
s4 CAV collaborative decision modeling step: establishing a model of collaborative decision in the CAV vehicle;
s5, man-machine hybrid interaction collaborative modeling step: coupling the CAV, the HV interaction model and the CAV collaborative decision model, carrying out a man-machine mixed driving interaction collaborative model, carrying out model solving, predicting the action of the HV in the game object list, and solving the optimal action strategy of the CAV, namely the acceleration;
s6, updating the vehicle state: the CAV realizes vehicle state update through a kinematic formula according to the output acceleration decision result, and solves the update speed and position; acceleration, speed, position of HV are re-acquired from the environment.
2. The method according to claim 1, wherein in the step S1, the vehicle attribute and status information updating step specifically includes:
acquiring vehicle type information of each vehicle from the environment, namely, the vehicle is CAV or HV; the kinematics related parameters comprise maximum acceleration, maximum deceleration and maximum speed;
collecting the current acceleration, speed and position of HV in the updated environment in each time step; the CAV updates the speed and the position of the vehicle according to the optimal acceleration output by the model and the kinematic formula; the vehicle state information is used as an input to the model.
3. The method of claim 1, wherein in the step S2, the game object list updating and model switching steps are as follows:
all vehicles s= { S in the scene 1 ，S 2 ，…，S k ，…，S K I CAV, denoted m= { M }, is contained therein 1 ，M 2 ，…，M i ，…，M I And J vehicles HV, denoted as N= { N 1 ，N 2 ，…，N j ，…，N J }；
Defining a game range, namely a near conflict area and a conflict area; vehicles entering the game range are added to the game object list; the vehicle exits the game object list after passing through the conflict area; vehicles in game object listWherein P CAV are contained, expressed as +.> Q vehicle HV, denoted->
Selecting and switching the model according to the types of vehicles and the number of various vehicles in the game object list,
when no CAV vehicle or only one vehicle exists in the game object list, the vehicle can freely pass through without considering interaction cooperation;
when the vehicles in the game object list are all CAV vehicles, only CAV collaborative decisions are carried out, and a CAV collaborative decision model is selected for carrying out vehicle optimal acceleration strategy solving;
when the vehicles in the game object list contain CAV and HV at the same time and only one CAV exists, only a CAV and HV interaction decision is needed, and a CAV and HV interaction model is selected for carrying out CAV optimal acceleration strategy solving;
when the vehicles in the game object list contain CAV and HV simultaneously and contain a plurality of CAV, the CAV collaborative decision model and the CAV and HV interactive decision model are required to be coupled to realize collaborative decision in the man-machine mixed driving scene.
4. The method according to claim 1, wherein in step S3,
the interaction model in the CAV and HV interaction modeling step is specifically as follows:
based on game theory, constructing a non-cooperative model, and dividing CAV and HV into a leader and a follower respectively;
design of the CAV benefit function f taking into consideration the requirements of safety, efficiency and comfort in the running process of the vehicle p And a benefit function f of HV q The method comprises the steps of carrying out a first treatment on the surface of the The two gain function forms are consistent, and the individual gain value of the vehicle is the weighted summation of the three;
in the method, in the process of the application,is the safety index of CAV, +.>Is an efficiency index of CAV, < >>Is a CAV comfort index;the weight values of the three indexes are respectively; />Safety, efficiency, comfort index of HV, respectively,>is the weight value of each; the specific index calculation is shown as the following formula:
wherein DeltaL p Is the distance between the CAV vehicle p and the conflict point of the HV vehicle q, v p (k+1) is the speed of the CAV vehicle p at time k+1, a p (k)、a p (k+1) is the acceleration of the CAV vehicle p at the kth and k+1 times, respectively; ΔL q V is the distance between the q distance of the HV vehicle and the conflict point of the CAV vehicle p q (k+1) is the speed of the HV vehicle q at time k+1, a q (k)、a q (k+1) is the acceleration of the HV vehicle q at the kth and k+1 times, respectively;
on the basis of the weight value of the HV at the previous moment, the safety and efficiency weight value of the HV is updated in real time according to the similarity degree of the predicted acceleration of the CAV to the actual acceleration of the CAV; in order to reduce the complexity of model parameter updating, the weight value corresponding to the safety and efficiency index is converted into a function form of the same parameter theta; the following formula is shown:
wherein A, B is a constant;
establishing an interaction model, and predicting the optimal acceleration which the HV may take under the condition that the CAV makes a certain acceleration action decision by itself according to the profit function of the HVOn the basis, self-income is calculated, and in CAV self-acceleration strategy space +.>Optimal solution for maximizing self-income by searching in->The following formula is shown:
s.t.
in the method, in the process of the application,CAV object list M for participating in gaming g Is +.>Policy space for HV at this time->Middle->The benefit of the corresponding HV,to enable->A maximum optimal acceleration strategy; />To get +.>HV adopts optimal acceleration strategy +.>The corresponding CAV overall benefit, +.>A CAV optimal acceleration strategy to maximize the benefit; at the same time (I)>Is->Minimum and maximum of>Is->Minimum and maximum values of (2);is a safety constraint, wherein->For the time when the CAV vehicle p arrives at the conflict point according to the current state, the +.>The time for the HV vehicle q to reach the conflict point according to the current state; optimal acceleration strategy will be achieved at CAV +.>HV optimal acceleration strategy determined on the basis of the model>As a model in->The prediction of HV is described as +.>For use in subsequent twinning games.
5. The method of claim 4, wherein,
updating the HV weight related parameter theta by adopting a twin game mode;
at the initial moment, weight values corresponding to three CAV indexes are obtained based on natural driving data calibration CAV fixed parameter value +.>The corresponding HV initial weight value related parameter is theta 0 The method comprises the steps of carrying out a first treatment on the surface of the At theta 0 R values are discretized again in the range of the definition domain to form an initial theta value space theta 0 R is set to an odd number;
in which the value space Θ 0 The value method of each theta value is as follows: at the initial moment, determining the range of the value space asDividing the minimum value in the value space>Optimal parameter of initial moment->Maximum value->In addition, at [0, θ 0 ]、/>Respectively equidistantly taking (r-3)/2 values, adding them into the value space, and forming value space theta with r number of elements 0 ；
Twin game solving: at each subsequent time step, the value space theta is based on the last moment k Each theta value in the model is solved on HV through a CAV and HV interaction model to obtain a series of HV optimal acceleration strategies corresponding to the theta values
Fitness calculation of optimal parametersAnd (3) determining: calculating Θ k HV optimal acceleration strategy corresponding to each theta value in the systemIs +.>Is shown in the following formula:
alpha is defined as fitness, the larger the alpha value, i.e. the smaller the difference between the calculated value and the true value, the higher the fitness; the corresponding fitness alpha in all theta values is highest and is used as the optimal parameter at the next moment
Value list theta k+1 Updating: according to the optimal parameters at the next momentAnd the corresponding fitness alpha * Confirming the value number r of the next moment; when the adaptability reaches a certain level, the optimal +.>Performing model calculation, and not performing twin game; until the fitness exceeds the level without twinning;
setting the length of the value interval asMinimum value of value limit->Maximum value->The calculation method is as follows:
if calculatedMake->Correspondingly->
If calculatedMake->Correspondingly->
Finally, the value space theta of the value of the next moment theta k+1 At the position ofThe nearby discrete values are obtained as shown in the following formula:
6. the method of claim 5, wherein the number of twinning values corresponds to the fitness as shown in table 1:
TABLE 1 correspondence between the number of twin values and fitness
7. The method according to claim 1, wherein in the step S4, the collaborative decision model in the CAV collaborative decision modeling step is specifically:
designing a characteristic function of CAV (computer aided design) by considering the requirements of safety, efficiency and comfort in the running process of the vehicle, wherein the individual profit value of the vehicle is the weighted summation of the three; the calculation mode of each index is consistent with the CAV and HV interaction model;
wherein p is ′ For M in a game object list g CAV vehicles other than CAV vehicle p, i.e.Calculating the security inside the CAV; weight value +.>According to the personality of CAVThe chemical preference is adjusted;
according to the basic principle of the cooperative game, designing a characteristic function, an optimization target and constraint conditions, distributing benefits according to the shape principle, and planning the optimal action of the vehicle; shapley is calculated as follows:
in the psi- p Is the shape of CAV vehicle p; s is M g Representing a collection of possible participant-cooperative gaming objects during the computing process; v(s) is the benefit generated by the set of participants for s; s and M g I represents s, M respectively g The number of vehicles involved;
v(s) is the benefit generated by the participant set s, calculated by:
establishing a model by taking the maximum total income of the CAV vehicle as a target; the following formula is shown:
s.t.
in the method, in the process of the application,is a safety constraint between CAVs.
8. The method of claim 1, wherein in the step S5, the interactive decision model in the man-machine hybrid interactive decision modeling step is specifically:
meanwhile, interaction between CAV and HV is considered, and cooperation inside the CAV is considered to construct a double-layer optimization model; the interaction of the outer layer of CAV as a whole with HV is shown in the following formula:
s.t.
in the method, in the process of the application,the overall income of the CAV is the sum of the income of the CAV contained in the game object list; safety constraints consider the safety between CAV and HV; the rest of the amounts have the same meaning as CAV and HV interaction models;
the inner layer is CAV internal cooperation, and the optimal acceleration strategy of the CAV vehicle is obtained by solving according to a CAV cooperation decision modelHowever, in the model, all CAV and HV except for the safety index in the game object list are considered in calculation.
9. The method according to claim 1, wherein in the step S6, after selecting an appropriate model according to the game object list to solve the optimal acceleration strategy, the vehicle state is updated;
the state of CAV is expressed and updated according to a kinematic formula:
in the method, in the process of the application,the speed, acceleration and position of the CAV vehicle are participated in the game;
the HV real state is used for carrying out information collection again from the environment to obtain the actual position of the HV real stateSpeed->Acceleration->
CN202311030796.0A 2023-08-15 2023-08-15 Multi-vehicle collaborative decision-making method in man-machine mixed driving environment Pending CN117075473A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311030796.0A CN117075473A (en) 2023-08-15 2023-08-15 Multi-vehicle collaborative decision-making method in man-machine mixed driving environment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311030796.0A CN117075473A (en) 2023-08-15 2023-08-15 Multi-vehicle collaborative decision-making method in man-machine mixed driving environment

Publications (1)

Publication Number Publication Date
CN117075473A true CN117075473A (en) 2023-11-17

Family

ID=88701650

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311030796.0A Pending CN117075473A (en) 2023-08-15 2023-08-15 Multi-vehicle collaborative decision-making method in man-machine mixed driving environment

Country Status (1)

Country Link
CN (1) CN117075473A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117302207A (en) * 2023-11-29 2023-12-29 华东交通大学 Intelligent automobile driving safety early warning system suitable for mixed driving environment

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117302207A (en) * 2023-11-29 2023-12-29 华东交通大学 Intelligent automobile driving safety early warning system suitable for mixed driving environment
CN117302207B (en) * 2023-11-29 2024-02-13 华东交通大学 Intelligent automobile driving safety early warning system suitable for mixed driving environment

Similar Documents

Publication Publication Date Title
CN111696370B (en) Traffic light control method based on heuristic deep Q network
CN108595823B (en) Autonomous main vehicle lane changing strategy calculation method combining driving style and game theory
Ma et al. Trajectory planning for connected and automated vehicles at isolated signalized intersections under mixed traffic environment
CN112907967B (en) Intelligent vehicle lane change decision-making method based on incomplete information game
CN114495527B (en) Internet-connected intersection vehicle road collaborative optimization method and system in mixed traffic environment
CN111845773B (en) Automatic driving vehicle micro-decision-making method based on reinforcement learning
CN109726804B (en) Intelligent vehicle driving behavior personification decision-making method based on driving prediction field and BP neural network
CN111081065A (en) Intelligent vehicle collaborative lane change decision model under road section mixed traveling condition
CN112700642B (en) Method for improving traffic passing efficiency by using intelligent internet vehicle
WO2023186024A1 (en) Self-adaptive parking lot changeable entrance and exit control method
CN117075473A (en) Multi-vehicle collaborative decision-making method in man-machine mixed driving environment
Ahmad et al. Applications of evolutionary game theory in urban road transport network: A state of the art review
CN114463997A (en) Lantern-free intersection vehicle cooperative control method and system
CN113867354A (en) Regional traffic flow guiding method for intelligent cooperation of automatic driving of multiple vehicles
Liao et al. A game theory based ramp merging strategy for connected and automated vehicles in the mixed traffic: A unity-sumo integrated platform
Hou et al. Large-scale vehicle platooning: Advances and challenges in scheduling and planning techniques
Li et al. Enhancing cooperation of vehicle merging control in heavy traffic using communication-based soft actor-critic algorithm
Lu et al. A game-theoretic approach on conflict resolution of autonomous vehicles at unsignalized intersections
CN115100866B (en) Vehicle-road cooperative automatic driving decision-making method based on layered reinforcement learning
CN117227755A (en) Automatic driving decision method and system based on reinforcement learning under complex traffic scene
CN114926823B (en) WGCN-based vehicle driving behavior prediction method
Lovellette et al. Lane and speed allocation mechanism for autonomous vehicle agents on a multi-lane highway
Liu et al. Behavioral Decision-Making Approach for Vehicle Platoon Control: Two Noncooperative Game Models
Dabiri et al. Personalised optimal speed advice to cyclists approaching an intersection with uncertain green time
Chen et al. Platoon separation strategy optimization method based on deep cognition of a driver’s behavior at signalized intersections

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination