Vehicle safety control algorithm based on model-free non-zero sum game

Abstract

The invention relates to the technical field of control and information, in particular to a vehicle safety control algorithm based on a model-free non-zero sum game, which comprises the following steps of: constructing a topological structure diagram of network communication between vehicles, and determining a communication relationship between the vehicles; according to the state equation of the # imgabs0 # vehicle following vehicle, the following vehicle communicated with the # imgabs1 # vehicle following vehicle and the leading vehicle, the tracking error of the # imgabs2 # vehicle following vehicle is obtained; defining a target function according to the tracking error and a non-zero sum game theory, and determining a control barrier function of the control strategy network; a control barrier function is substituted into the value function network, a constrained problem is converted into an unconstrained problem, and a model-free vehicle safety control algorithm is obtained; and training the control strategy network and the value function network according to a vehicle safety control algorithm to obtain a control law of the control strategy network, and controlling the # imgabs3 vehicle following vehicle according to the control law.

Description

Technical Field

[0001] The present invention relates to the field of control and information technology, and in particular to a vehicle safety control algorithm based on model-free non-zero-sum game. Background Art

[0002] Vehicle control is a key technology of intelligent transportation systems. The core is to ensure vehicle driving safety, that is, to avoid collisions, while effectively reducing the driving distance between vehicles, so that vehicles can reduce air resistance and fuel consumption during driving, while improving road safety and effectively alleviating traffic congestion. With the advancement of intelligent transportation systems, simply achieving the basic control goals of vehicles is no longer sufficient to meet the needs of modern transportation. The current focus has shifted to optimizing vehicle platform control, including maximizing fuel efficiency, improving passenger comfort, and shortening travel time.

[0003] In recent years, with the advancement of vehicle control technology, technologies including artificial potential field method and model predictive control have been widely used. However, these methods often ignore the impact on other vehicles when solving the problem of safe driving of a single vehicle and cannot achieve global optimization. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the related art. To this end, the present invention provides a vehicle safety control algorithm based on model-free non-zero-sum game, which does not require a specific model of the vehicle, improves the robustness and adaptability of the entire system, and ensures the safety of vehicle-to-vehicle operation.

[0005] The present invention provides a vehicle safety control algorithm based on model-free non-zero-sum game, comprising the following steps: Construct a topology diagram of network communication between vehicles, the topology diagram including a leader vehicle and N The leading vehicle and the following vehicle are determined according to the topological structure diagram. N a connectivity relationship between the following vehicles; According to The state equation of the following vehicle, and The state equations of the leading vehicle and the following vehicle connected to each other are obtained as follows: The tracking error of the following vehicle, =1,2,3……, N ; According to The tracking error of the following vehicle and the non-zero-sum game theory define the objective function and determine the control barrier function of the control strategy network; The control barrier function is introduced into the value function network as an evaluation item, and the constrained problem is transformed into an unconstrained problem, thus obtaining a model-free vehicle safety control algorithm. According to the vehicle safety control algorithm, the control strategy network and the value function network are trained to obtain the control law of the control strategy network, and the control law is used to control the first The following vehicle is controlled.

[0006] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, the leading vehicle and the following vehicle and N The method for establishing a connectivity relationship between the following vehicles comprises the following steps: The topological structure diagram of the inter-vehicle network communication includes a leader vehicle and N The undirected graph formed by the following vehicles is represented as follows: ,in, V To follow the collection of vehicles, , E for N The communication status between the following vehicles, ; The adjacency matrix A is the connectivity between following vehicles, ,when >0 indicates the The following vehicle and i The following vehicles are connected to each other, Indicates i A collection of vehicles with which the following vehicle has communication; The leading vehicle communicates unidirectionally with the following vehicle, O represents the connectivity between the following vehicle and the leading vehicle, .

[0007] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, obtaining the tracking error of the following vehicle includes the following steps: According to i The state equation of the following vehicle, and The state equation of the following vehicle connected to the following vehicle and the state equation of the leading vehicle are obtained. i The tracking error of the following vehicle is given by the following formula: in, For the Following vehicles t The state equation at time, , For the Following vehicles t The location at the moment, For the Following vehicles t The speed of time, For the Following vehicles t The acceleration of time, For the The differential of the state equation of the following vehicle, For the The vehicle follows the control input of the vehicle. B is the first coefficient, C is the second coefficient, is the inertial delay, ; in, For the The tracking error of the following vehicle, For the j Following vehicles t The state equation at time, For leadership vehicles t The state equation at time, ,in Indicates The expected relative distance between the following vehicle and the leading vehicle, , For the The following vehicle and The expected relative distance between the vehicle and the following vehicle.

[0008] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, defining the objective function includes the following steps: in, in, For the The cost function of the following vehicle, For the The control barrier function of the following vehicle, For the The control law of the control strategy network of the vehicle following the vehicle, For the The control law of the control strategy network of the following vehicles connected to each other is For the The tracking error of the following vehicle connected to the following vehicle, is a positive definite matrix, which is used to adjust the emphasis of the model-free vehicle safety control algorithm on tracking error.r Function is used to determine the The tracking error of the following vehicle is The following vehicle and The U function is used to limit the upper bound of the control input of the following vehicle. The formula is as follows: in, Indicates The upper bound of the control input of the following vehicle, s is the independent variable of the control law, Indicates The upper bound of the control input of the following vehicle connected to the following vehicle, and is a positive definite matrix, Restriction The upper bound of the control input of the following vehicle, Restrictions and The upper bound of the control input of the following vehicle that is connected to the following vehicle.

[0009] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, determining a control barrier function comprises the following steps: in, for t Moment The following vehicle and The distance control parameters between the following vehicles, is a constant greater than zero, No. The safe distance between the following vehicles, For the Following vehicles t The location at the moment, Indicates The following vehicle and The desired spacing between following vehicles.

[0010] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, constructing a value function includes the following steps: The control barrier function is added as an evaluation item to the value function network, and the value function formula of the value function network is as follows: in, in, is the value function, z is the independent variable related to time.

[0011] According to a vehicle safety control algorithm based on a model-free non-zero-sum game provided by the present invention, training the model-free vehicle safety control algorithm includes the following steps: Select the control law of the initial control strategy network and the initial network weight parameters of the value function network; The Bellman equation error is used to iteratively update the weight parameters of the value function network, thereby iteratively updating the weight parameters and control law of the control strategy network.

[0012] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, the network weight parameters of the value function network and the control law of the control strategy network are iteratively updated using the following formula: in, For the Following vehicle The value function of the iteration, For the Following vehicle The control law of the control strategy network of the next iteration; is the activation function of the value function network, is the activation function of the control strategy network, is the value function network The network weight parameters for the iteration, The control strategy network The network weight parameters for the iteration.

[0013] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, the weight parameters of the value function network are updated iteratively using the Bellman equation error, including the following steps: The error of the Bellman equation is defined as follows: in, is the error of the Bellman equation, For the The difference between the control law of the given control strategy network of the following vehicle and the iteratively generated control law; The Bellman equation is transformed using the regression equation to obtain the following formula: in, in, For the The following vehicle and The interval value function of the following vehicle, For the The following vehicle and The interval control error function between the following vehicles; M The moment The following vehicle and The interval value function and interval control error function of the following vehicle are stored in the matrix, and the formula is as follows: in, is the interval control error function matrix, used to store M The moment The following vehicle and The interval control error function between the following vehicles, is the interval value function matrix, used to store M The moment The following vehicle and The interval value function of the following vehicle; The weight parameter update equation of the value function network is: .

[0014] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, updating the weight parameters of the value function network also includes the following steps: , For the i Following vehicle l The network weight parameters of the value function network of the iteration are l -1 The minimum difference between iterations, then the output loop gets the weight parameters of the value function network , otherwise, let , continue to use the Bellman equation error to iteratively update the weight parameters of the value function network.

[0015] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, the control law of the control strategy network is obtained according to the weight parameters of the value function network.

[0016] The above one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects: According to the vehicle safety control algorithm based on model-free non-zero-sum game in an embodiment of the present invention, by adopting a model-free non-zero-sum game algorithm, no specific vehicle model is required, and each vehicle can reach Nash equilibrium, balance error and energy consumption, and ensure the safety of vehicle operation.

[0017] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 It is one of the flow charts of the vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention.

[0020] Figure 2 It is a topological structure diagram of the inter-vehicle network communication of the vehicle safety control algorithm based on the model-free non-zero-sum game provided by the present invention.

[0021] Figure 3 It is a graph showing the change in distance between the third following vehicle and the fourth following vehicle over time in the vehicle safety control algorithm based on the model-free non-zero-sum game provided by the present invention.

[0022] Figure 4 A graph showing changes in the tracking error of a following vehicle over time for a vehicle safety control algorithm based on a model-free non-zero-sum game provided by the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme in the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0024] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0025] The non-zero-sum game method provides a new perspective to solve the problem that the impact on other vehicles is often ignored when solving the problem of safe driving of a single vehicle. In the non-zero-sum game framework, each vehicle is regarded as an independent individual, striving to achieve its own predetermined goals to the maximum extent, and transforming the centralized energy optimization control problem into a non-zero-sum game model. On this basis, in order to ensure driving safety, a control barrier function is introduced. This method transforms the system with full state constraints into an unconstrained multi-vehicle safety coordination system, and ensures the overall stability of the system by establishing a Nash equilibrium.

[0026] According to the above problems, a vehicle safety control algorithm based on model-free non-zero-sum game is proposed. The basic idea is: based on the vehicle topological communication structure, a control barrier function is designed to balance the safety and optimality of the system. A new reinforcement learning method is established for non-zero-sum game and derived into a model-free algorithm to obtain the corresponding control input of each vehicle.

[0027] Figure 1 It is one of the flow charts of the vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention.

[0028] The present invention provides a vehicle safety control algorithm based on model-free non-zero-sum game, such as Figure 1 As shown, the following steps are included: Construct a topology diagram of network communication between vehicles, the topology diagram including a leader vehicle and N The leading vehicle and the following vehicle are determined according to the topological structure diagram. N a connectivity relationship between the following vehicles; According to The state equation of the following vehicle and the The state equations of the leading vehicle and the following vehicle connected to each other are obtained as follows: The tracking error of the following vehicle, =1,2,3……,N ; According to The tracking error of the following vehicle and the non-zero-sum game theory define the objective function and determine the control barrier function of the control strategy (Actor) network; The control barrier function is introduced into the value function (Critic) network as an evaluation item, and the constrained problem is transformed into an unconstrained problem, thus obtaining a model-free vehicle safety control algorithm. According to the vehicle safety control algorithm, the control strategy network and the value function network (Actor-Critic network) are trained to obtain the control law of the control strategy network. The following vehicle is controlled.

[0029] In this embodiment, by adopting a model-free non-zero-sum game algorithm, a specific vehicle model is not required, and each vehicle can reach a Nash equilibrium, balance the error and energy consumption, and ensure the safety of vehicle operation.

[0030] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, the leading vehicle and the following vehicle and N The method for establishing a connectivity relationship between the following vehicles comprises the following steps: The topological structure diagram of the inter-vehicle network communication includes a leader vehicle and N The undirected graph formed by the following vehicles is represented as follows: ,in, V To follow the collection of vehicles, , E for N The communication status between the following vehicles, ; The adjacency matrix A is the connectivity between following vehicles, ,when >0 indicates the The following vehicle and i The following vehicles are connected to each other, Indicates i a set of vehicles with which the following vehicle has communication; The leading vehicle communicates unidirectionally with the following vehicle, and O represents the connectivity between the following vehicle and the leading vehicle. ,when When i The following vehicles can receive the information of the leading vehicle.

[0031] In this embodiment, the communication network topology diagram has low structural requirements, does not require a specific model of the vehicle, and only needs to communicate with local neighboring intelligent agents. It has low communication costs, strong adaptability, and is easy to implement.

[0032] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, obtaining the tracking error of the following vehicle includes the following steps: The tracking error of the vehicle is obtained according to the state equation of the following vehicle, and its formula is as follows: in, For the Following vehicles t The state equation at time, , For the Following vehicles t The location at the moment, For the Following vehicles t The speed of time, For the Following vehicles t The acceleration of time, For the The differential of the state equation of the following vehicle, For the The vehicle follows the control input of the vehicle. B is the first coefficient, C is the second coefficient, is the inertial delay, ; in, For the The tracking error of the following vehicle, For the j Following vehicles t The state equation at time, For leadership vehicles t The state equation at time, ,in Indicates The expected relative distance between the following vehicle and the leading vehicle, , For the The following vehicle and The expected relative distance between the vehicle and the following vehicle.

[0033] In this embodiment, the safety of the operation between vehicles can be ensured by calculating the tracking error and relative distance of the following vehicle.

[0034] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, defining the objective function includes the following steps: in, in, in, For the The cost function of the following vehicle, For the The control barrier function of the following vehicle, For the The control law of the control strategy network of the vehicle following the vehicle, For the The control law of the control strategy network of the following vehicles connected to each other is For the The tracking error of the following vehicle connected to the following vehicle, is a positive definite matrix, which is used to adjust the emphasis of the model-free vehicle safety control algorithm on tracking error. r Function is used to determine the The tracking error of the following vehicle is The following vehicle and The U function is used to limit the upper bound of the control input of the following vehicle. The formula is as follows: in, Indicates The upper bound of the control input of the following vehicle, s is the independent variable of the control law, Indicates The upper bound of the control input of the following vehicle connected to the following vehicle, and is a positive definite matrix, Restriction The upper bound of the control input of the following vehicle, Restrictions and The upper bound of the control input of the following vehicle that is connected to the following vehicle.

[0035] In this embodiment, the robustness and adaptability of the entire system are improved by adopting a model-free non-zero-sum game algorithm.

[0036] According to some embodiments of the present invention, a non-zero-sum game is a cooperative game in which the sum of gains or losses of each party in the game is not zero.

[0037] According to some embodiments of the present invention, a strategy combination reaches a Nash equilibrium when the equilibrium strategy of each player is to maximize his or her expected return, and at the same time, all other players also follow such a strategy.

[0038] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, determining a control barrier function comprises the following steps: in, for t Moment The following vehicle and The distance control parameters between the following vehicles, is a constant greater than zero, No. The safe distance between the following vehicles, For the Following vehicles t The location at the moment, Indicates The following vehicle and The desired spacing between following vehicles.

[0039] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, constructing a value function includes the following steps: The control barrier function is added as an evaluation item to the value function network, and the value function formula of the value function network is as follows: in, in, is the value function, z is the independent variable related to time.

[0040] In this embodiment, the security and optimality of the system are balanced by designing a control barrier function.

[0041] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, Training a model-free vehicle safety control algorithm consists of the following steps: Select the control law of the initial control strategy network and the initial network weight parameters of the value function network; The Bellman equation error is used to iteratively update the weight parameters of the value function network, thereby iteratively updating the weight parameters and control law of the control strategy network.

[0042] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, the network weight parameters of the value function network and the control law of the control strategy network are iteratively updated using the following formula: in, For the Following vehicle The value function of the iteration, For the Following vehicle The control law of the control strategy network of the next iteration; is the activation function of the value function network, is the activation function of the control strategy network, is the value function network The network weight parameters for the iteration, The control strategy network The network weight parameters for the iteration.

[0043] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, the weight parameters of the value function network are updated iteratively using the Bellman equation error, including the following steps: The error of the Bellman equation is defined as follows: in, is the error of the Bellman equation, For the The difference between the control law of the given control strategy network of the following vehicle and the iteratively generated control law; The Bellman equation is transformed using the regression equation to obtain the following formula: in, Will M The moment The following vehicle and The interval value function and interval control error function of the following vehicle are stored in the matrix, and the formula is as follows: in, is the interval control error function matrix, used to store M The moment The following vehicle and The interval control error function between the following vehicles, is the interval value function matrix, used to store M The moment The following vehicle and The interval value function of the following vehicle; The weight parameter update equation of the value function network is: According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, updating the weight parameters of the value function network also includes the following steps: , For the i Following vehicle l The network weight parameters of the value function network of the iteration are l -1 The minimum difference between iterations, then the output loop gets the weight parameters of the value function network , otherwise, let , continue to use the Bellman equation error to iteratively update the weight parameters of the value function network.

[0044] According to some preferred embodiments of the present invention, =0.1.

[0045] According to a vehicle safety control algorithm based on model-free non-zero-sum game provided by the present invention, the control law of the control strategy network is obtained according to the weight parameters of the value function network.

[0046] The technical solution of the present invention is further explained below in conjunction with a specific embodiment. It should be noted that this specific embodiment is only for those skilled in the art to better understand the technical solution of the present invention, and should not be regarded as an unreasonable limitation on the protection scope of the present invention.

[0047] Embodiment 1 Figure 2 It is a topological structure diagram of the inter-vehicle network communication of the vehicle safety control algorithm based on the model-free non-zero-sum game provided by the present invention. Figure 3It is a graph showing the change in distance between the third following vehicle and the fourth following vehicle over time in the vehicle safety control algorithm based on the model-free non-zero-sum game provided by the present invention. Figure 4 A graph showing changes in the tracking error of a following vehicle over time for a vehicle safety control algorithm based on a model-free non-zero-sum game provided by the present invention.

[0048] In the simulation experiment, it is assumed that the topological structure of the vehicle-to-vehicle network communication is as follows: Figure 2 For each vehicle, (identity matrix), , , , the control input of each vehicle satisfies , safe following distance , the ideal vehicle spacing is 7. Vehicle following vehicle, defining tracking error , that is, the error of position, velocity and acceleration, defined According to the topological structure diagram of the inter-vehicle network communication, the activation function of the critic network of each vehicle can be set. , , , , , its initial weight parameter Each item is 1. Similarly, you can set the activation function of the Actor network of each car. , , , , , its initial weight parameter Each item is 1.

[0049] like Figure 3 As shown in the figure, without considering the control barrier function (Without CBF), the distance between the third and fourth following vehicles may be as small as the risky region. However, after adding the control barrier function (With CBF), the distance between the third and fourth following vehicles can strictly ensure that the vehicle distance is in the safe region, which shows that the model can improve the robustness and adaptability of the entire system.

[0050] like Figure 4 As shown in the figure, after learning is completed, the tracking errors of the first following vehicle, the second following vehicle, the third following vehicle, the fourth following vehicle and the fifth following vehicle gradually tend to 0, and the closer the distance to the leading vehicle is, the easier it is to stabilize.

[0051] Compared with the prior art, a vehicle safety control algorithm based on a model-free non-zero-sum game in an embodiment of the present invention has the following technical effects: (1) No specific vehicle model is required, which improves the robustness and adaptability of the entire system.

[0052] (2) The network communication topology has low structural requirements, strong applicability and is easy to implement.

[0053] (3) Each following vehicle only needs to communicate with its local neighboring following vehicles, which reduces the communication cost and ensures the safety of inter-vehicle operation.

[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vehicle safety control algorithm based on model-free non-zero-sum game, characterized in that: The steps include: Construct a topology diagram of network communication between vehicles, the topology diagram including a leader vehicle and N The leading vehicle and the following vehicle are determined according to the topological structure diagram. N a connectivity relationship between the following vehicles; According to The state equation of the following vehicle and the The state equations of the leading vehicle and the following vehicle connected to the following vehicle are obtained by The tracking error of the following vehicle, =1,2,3……, N ; According to The tracking error of the following vehicle and the non-zero-sum game theory define the objective function and determine the control barrier function of the control strategy network; The control barrier function is introduced into the value function network as an evaluation item, the constrained problem is transformed into an unconstrained problem, and a model-free vehicle safety control algorithm is obtained; According to the vehicle safety control algorithm, the control strategy network and the value function network are trained to obtain the control law of the control strategy network, and the control law is used to control the first The following vehicle is controlled.

2. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 1 is characterized in that: Determine the leading vehicle and the following vehicle and N The method for establishing a connectivity relationship between the following vehicles comprises the following steps: The topological structure diagram of the inter-vehicle network communication includes a leader vehicle and N The undirected graph formed by the following vehicles is represented as follows: ,in, V To follow the collection of vehicles, , E for N The communication status between the following vehicles, ; The adjacency matrix A is the connectivity between following vehicles, ,when >0 indicates the The following vehicle and i The following vehicles are connected to each other, Indicates i a set of vehicles with which the following vehicle has communication; The leading vehicle communicates unidirectionally with the following vehicle, O represents the connectivity between the following vehicle and the leading vehicle, .

3. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 2 is characterized in that: Obtaining the tracking error of the following vehicle includes the following steps: According to i The state equation of the following vehicle, and The state equation of the following vehicle connected to the following vehicle and the state equation of the leading vehicle are obtained. i The tracking error of the following vehicle is given by the following formula: in, For the Following vehicles t The state equation at time, , For the Following vehicles t The location at the moment, For the Following vehicles t The speed of time, For the Following vehicles t The acceleration of time, For the The differential of the state equation of the following vehicle, For the The vehicle follows the control input of the vehicle. B is the first coefficient, C is the second coefficient, is the inertial delay, ; in, For the The tracking error of the following vehicle, For the j Following vehicles t The state equation at time, For leadership vehicles t The state equation at time, ,in Indicates The expected relative distance between the following vehicle and the leading vehicle, , For the The following vehicle and The expected relative distance between the vehicle and the following vehicle.

4. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 3 is characterized in that: Defining the objective function involves the following steps: in, in, For the The cost function of the following vehicle, For the The control barrier function of the following vehicle, For the The control law of the control strategy network of the vehicle following the vehicle, For the The control law of the control strategy network of the following vehicles connected to each other is For the The tracking error of the following vehicle connected to the following vehicle, is a positive definite matrix, which is used to adjust the emphasis of the model-free vehicle safety control algorithm on tracking error. r Function is used to determine the The tracking error of the following vehicle is The following vehicle and The U function is used to limit the upper bound of the control input of the following vehicle. The formula is as follows: in, Indicates The upper bound of the control input of the following vehicle, s is the independent variable of the control law, Indicates The upper bound of the control input of the following vehicle connected to the following vehicle, and is a positive definite matrix, Restriction The upper bound of the control input of the following vehicle, Restrictions and The upper bound of the control input of the following vehicle that is connected to the following vehicle.

5. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 4 is characterized in that: Determining the control barrier function includes the following steps: in, for t Moment The following vehicle and The distance control parameters between the following vehicles, is a constant greater than zero, No. The safe distance between the following vehicles, For the Following vehicles t The location at the moment, Indicates The following vehicle and The desired spacing between following vehicles.

6. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 5 is characterized in that: Building a value function involves the following steps: The control barrier function is added as an evaluation item to the value function network, and the value function formula of the value function network is as follows: in, in, is the value function, z is the independent variable related to time.

7. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 6 is characterized in that: Training a model-free vehicle safety control algorithm consists of the following steps: Select the control law of the initial control strategy network and the initial network weight parameters of the value function network; The Bellman equation error is used to iteratively update the weight parameters of the value function network, thereby iteratively updating the weight parameters and control law of the control strategy network.

8. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 7 is characterized in that: The network weight parameters of the value function network and the control law of the control strategy network are iteratively updated using the following formula: in, For the Following vehicle The value function of the iteration, For the Following vehicle The control law of the control strategy network of the next iteration; is the activation function of the value function network, is the activation function of the control strategy network, is the value function network The network weight parameters for the iteration, The control strategy network The network weight parameters for the iteration.

9. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 8, characterized in that: Using the Bellman equation error to iteratively update the weight parameters of the value function network includes the following steps: The error of the Bellman equation is defined as follows: in, is the error of the Bellman equation, For the The difference between the control law of the given control strategy network of the following vehicle and the iteratively generated control law; The Bellman equation is transformed using the regression equation to obtain the following formula: in, in, For the The following vehicle and The interval value function of the following vehicle, For the The following vehicle and The interval control error function between the following vehicles; M The moment The following vehicle and The interval value function and interval control error function of the following vehicle are stored in the matrix, and the formula is as follows: in, is the interval control error function matrix, used to store M The moment The following vehicle and The interval control error function between the following vehicles, is the interval value function matrix, used to store M The moment The following vehicle and The interval value function of the following vehicle; The weight parameter update equation of the value function network is: 。 10. The vehicle safety control algorithm based on model-free non-zero-sum game according to claim 9, characterized in that: Updating the weight parameters of the value function network also includes the following steps: , For the i Following vehicle l The network weight parameters of the value function network of the iteration are l -1 The minimum difference between iterations, then the output loop gets the weight parameters of the value function network , otherwise, let , continue to use the Bellman equation error to iteratively update the weight parameters of the value function network.

Images