Path planning and platoon control method for multi-unmanned vehicle system with sequential logic tasks

Abstract

The present invention provides a path planning and formation control method for a multi-unmanned vehicle system of sequential logic tasks, including: establishing a signal sequential logic (STL) task specification framework for the multi-unmanned vehicle system; constructing a control barrier function (CBF) corresponding to the STL according to the task specification framework; solving the input control law of the multi-unmanned vehicle system under the constraint of the CBF; completing the path planning and formation tasks of the multi-unmanned vehicle system under the input control law; and cyclically detecting the task completion. The present invention utilizes CBF constraints to solve the input control law of the multi-unmanned vehicle system in real time, has a fast solution speed, is suitable for dynamically solving the input control law, and always has a suitable solution under input restrictions. Due to the robust design of CBF, it can ensure the completion of obstacle avoidance and collision avoidance tasks.

Description

Technical Field

[0001] The present invention relates to a collaborative task path planning and formation control technology for a multi-unmanned system, and in particular, to a path planning and formation control method for a multi-unmanned vehicle system under a sequential logic task. Background Art

[0002] In recent years, multi-unmanned systems have attracted extensive attention from industry and academia. Multi-unmanned systems are composed of multiple intelligent agents, and each agent can complete certain specific tasks by exchanging information and cooperating with adjacent agents. The application scope of multi-unmanned systems is very broad. They can be used in various fields such as mobile robots, ships, aircraft, spacecraft, logistics, medical care and manufacturing to perform tasks such as distributed task allocation, rescue missions, regional patrols, spatial positioning, mapping and biomanufacturing. With the rapid increase of unmanned systems in logistics, medical care and industry, the complexity of the tasks they perform is also increasing, requiring more rigorous and efficient sequential logic control methods. In order to solve sequential logic tasks such as periodic tasks, request-response tasks, and time-limited tasks, it is necessary to introduce formalized sequential logic tasks.

[0003] However, there are many problems in the study of control issues of multi-unmanned systems, especially when faced with complex dual constraints of time and space, that is, sequential logic constraints, it is still a difficult problem for traditional control methods to simultaneously meet the completion and stability of tasks such as formation, arrival, obstacle avoidance, and collision avoidance. For multi-unmanned systems, when there is task coupling between intelligent agents, the safety issue of multi-unmanned systems in collaborative operation is crucial.

[0004] Collaborative control of multiple unmanned systems refers to managing intelligent agents to collaboratively achieve their goals. It is an effective way to solve complex problems. Its target tasks may have spatial constraints in terms of position, posture, etc., and may also have temporal logic constraints such as cycle, sequence, and time limit. The system assigns target tasks to intelligent agents, and each intelligent agent uses multiple inputs (such as subtasks, operation history, information interaction with adjacent intelligent agents, etc.) to achieve control goals. At the same time, the operations performed by each intelligent agent will affect the environment, and then affect the decisions of other intelligent agents. Comprehensive consideration of multiple factors will form the optimal control output. Summary of the invention

[0005] In view of the defects in the prior art, the purpose of the present invention is to provide a path planning and formation control method for a multi-unmanned vehicle system of sequential logic tasks, comprising:

[0006] Establish a signal timing logic (STL) task specification framework for multiple unmanned vehicle systems;

[0007] Constructing a control barrier function (CBF) corresponding to the STL according to the task specification framework;

[0008] Solve the input control law of the multi-unmanned vehicle system under the constraints of the CBF;

[0009] Complete the path planning and formation tasks of the multi-unmanned vehicle system under the input control law;

[0010] Cycle to check task completion.

[0011] Preferably, the signal timing logic (STL) task specification framework for establishing a multi-unmanned vehicle system includes:

[0012] Establish a state space model of a multi-unmanned vehicle system;

[0013] A multi-unmanned vehicle system with M unmanned vehicles is represented by an undirected graph Modeling consists of ε∈v×v represents the association between unmanned vehicles;

[0014] make n represents a single unmanned vehicle i i dimensional state trajectory column vector, m represents a single unmanned vehicle i i -dimensional input column vector;

[0015] make Where n: = n1 + ... + n M Represents the sum of the state dimensions of the multi-unmanned vehicle system;

[0016] The state space model of a single unmanned vehicle is defined as follows:

[0017]

[0018] in

[0019] The first variable x i It is locally Lipschitz continuous and measurable in time t;

[0020] c i (x, t) is used to simulate the given dynamic coupling formed by the connection between unmanned vehicles, or to simulate the unknown dynamics that cannot be accurately measured or the interference noise formed during the movement, as well as the influence of the wind speed in the simulation environment;

[0021] The indefinite term is bounded, that is, ||c i (x,t)||≤C.

[0022] Preferably, the establishing of a signal timing logic (STL) task specification framework for a multi-unmanned vehicle system further comprises: establishing an STL task standardization framework;

[0023] Based on STL formal statements, the sequential logic tasks of the multi-unmanned vehicle system are described, and the STL task specification framework is established:

[0024]

[0025] φ:=G [a，b] ψ|F [a，b] ψ|ψ1U [a，b] ψ2|φ1∧φ2 (2-2)

[0026] Among them, ψ and φ are STL formal statements describing the temporal logic tasks of the multi-unmanned vehicle system, ψ1 and ψ2 are in the same form as ψ, indicating whether the tasks of a single unmanned vehicle or multiple unmanned vehicles in collaboration are satisfied;

[0027] φ1 and φ2 are of the same form as φ. Their respective statements contain three temporal logic operators G, F, U, and the time constraint interval [a, b] for the execution of the corresponding task.

[0028] Combine the operator and the time constraint interval into global, finally, and until operators: G [a,b] , F [a,b] ,U [a,b] ; G [a,b] The ψ operator indicates that the constraint ψ needs to be always satisfied in the time interval [a, b];

[0029] F [a,b] The ψ operator indicates that the constraint ψ must be satisfied at least once in the future time interval [a, b];

[0030] ψ1U [a,b] The ψ2 operator indicates that constraint ψ2 needs to be satisfied at least once in the time interval [a, b] and constraint ψ1 needs to be satisfied continuously before constraint ψ2 is satisfied;

[0031] Predicate logic μ is composed of predicate functions To determine whether the actual state set of the multi-unmanned vehicle system is in the expected state set, the predicate function is only related to the column vector of the system state space Related;

[0032] The predicate μ is defined as:

[0033] Preferably, constructing a control barrier function (CBF) corresponding to the STL according to the task specification framework includes:

[0034] The constraints that need to be satisfied when constructing CBF;

[0035] For a single sequential logic task, define a differentiable function in Variables used to represent the state space trajectory of the system;

[0036] Defining a Collection

[0037]

[0038] If there exists an absolutely continuous function Its state at the initial time of the time interval [t0, t1] is x(t0) = x0, and the initial state x0 is input into the set, so that the function b(x0, t0) ≥ 0, and in the following time, b(x, t) ≥ 0 is always satisfied, that is, Then the control obstacle function b(x,t) is obtained; the state x involves the position state information of a single or multiple unmanned vehicles;

[0039] Smooth estimation of multiple control barrier functions CBFs;

[0040] For multiple unmanned vehicle tasks, there is the following smooth estimation formula:

[0041]

[0042] Where η>0, p represents the number of effective CBFs, when η→∞, the smooth estimate is:

[0043]

[0044]

[0045] make

[0046]

[0047] Satisfy the constraint condition b(x,t)≥0, that is, satisfy the corresponding condition of the candidate CBF within the time constraint interval of the STL task The forward set remains unchanged, then the control obstacle function b corresponding to any subtask embedded in b(x,t) is satisfied l (x,t)≥0 holds;

[0048] Each of the subfunctions b l (x, t) are associated with a temporal logic formula G [al,bl] u l or F [al,bl] u l .

[0049] Preferably, the construction rules of the CBF include:

[0050] Let p = 1 and consider φ: = G [al,bl] u l and φ: = F [al,bl] u l Two situations;

[0051] The maximum value of the predicate function satisfies At the same time, set the following time nodes:

[0052]

[0053] For F [al,bl] u l , it is only necessary to make the predicate logic u l True if satisfied once;

[0054] For G [al,bl] u l , then it is necessary to make the predicate logic u l Always keep True;

[0055] like Satisfy h l (x(0))≥r;

[0056] Let the control obstacle subfunction have the following function structure b l (x,t):=-γ l (t)+h l (t), the predicate function is obtained by setting the following parameters;

[0057]

[0058]

[0059]

[0060] By selecting the error range γ between the position of the unmanned vehicle and the target point l,∞ , and γ l The coefficient λ of (t) ensures that at any time Satisfy b l (x(t′),t′)≤h l (x(t′))-r;

[0061] In the subsequent control strategy, at any time By satisfying the constraint b l (x(t′),t′)≥0, directly satisfying h l (x(t′))≥r, by choosing and λ, satisfying the robust semantics ρ φ (x,0)≥r;

[0062] For a composite temporal logic operator, the size of the robustness of the total predicate function is calculated according to the following formula:

[0063] ρ μ (x, t): = h(x(t)),

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070] (b) For p>1, consider the function b(x,t): = -ln(exp(-b1(x,t))+exp(-b2(x,t))); by choosing η, r, λ, γ l,0 ,γ l,∞ , satisfying the constraints.

[0071] Preferably, solving the input control law of the multi-unmanned vehicle system under the constraint of CBF includes: establishing a state space model of the multi-unmanned vehicle system;

[0072] For a centralized multi-unmanned vehicle system, there are M intelligent agent systems and a total of K STL semantics φ1...φ k , for each agent The status trajectory of the entire multi-unmanned system It is composed of the state space trajectory column vectors of each agent; Represents the dimension of the entire multi-unmanned system state space;

[0073] For included in v k All agents in the state space model are reconstructed into a dynamic system:

[0074]

[0075] The input of the system Determined by the state trajectory and time of the system; all agents v participating in these constraints k The equivalent state space model parameters of are:

[0076]

[0077]

[0078]

[0079]

[0080] in Produced by the coupling of some or all agents in x.

[0081] Preferably, solving the input control law of the multi-unmanned vehicle system under the constraint of CBF further includes: constructing the constraint conditions of the input control law based on the state space model of the multi-unmanned vehicle system; the constraint conditions include three categories:

[0082] (1) Order Represents the constraint φ k Related control obstacle function, the equivalent state space model of the multi-unmanned vehicle system, and its centralized control obstacle function constraint condition are:

[0083]

[0084] The function is a κ-type Lipschitz function; C is a constant, and

[0085] (2) The control obstacle function corresponding to the obstacle avoidance and collision avoidance tasks always satisfies the corresponding spatial constraints within a certain time interval; the following non-time-varying control obstacle function is defined When the driverless car i∈v k Entering an unsafe area Right now When , the function is expressed as b(x i )<0, when maintained in a safe state, it behaves as And b(x i )≥0, let the set If the forward set needs to be unchanged; that is, in the local control law u i The state space trajectory solution for a single unmanned vehicle To keep the forward direction unchanged, the following time-invariant control barrier function constraints need to be satisfied:

[0086]

[0087] (3) If a temporary team is required, the computational burden is distributed to each unmanned vehicle, and the unmanned vehicle's own computing module solves the input control law. The control obstacle function constraint corresponding to the input control law solved by each unmanned vehicle is:

[0088]

[0089]

[0090]

[0091]

[0092] The weight function

[0093]

[0094] where ||v k || represents the time-varying CBF corresponding to a coupled sequential logic task, b″(x i ,x j ,...,t), and the time-invariant CBF, b″(x i ,x j ,...) The number of unmanned vehicles involved.

[0095] Preferably, solving the input control law of the multi-unmanned vehicle system under the constraint of the CBF further includes: generating the input control law under the constraint based on the CBF and the constraint conditions:

[0096] In the time range N, starting from t = 0, after each time step Δt, the optimization problem is solved in the time interval [t, t + Δt]. The optimization problem is described as a quadratic programming problem under the following centralized CBF constraints:

[0097] arg min u T u

[0098] st

[0099]

[0100] If the computational complexity is to be shared, the input control law is calculated by the small computing module of the unmanned vehicle. Then the quadratic programming problem under the distributed CBF constraint is as follows:

[0101] arg min u i T u i

[0102] st

[0103]

[0104]

[0105]

[0106] The path planning and formation tasks of the multi-unmanned vehicle system are completed by inputting the control law, including: applying the input control law solved at the current moment to the system input at the next moment, and iterating in this way, so that each unmanned vehicle can complete its own arrival, obstacle avoidance, collision avoidance, and formation tasks in a dangerous working environment.

[0107] Preferably, the cyclic detection task completion degree includes:

[0108] Real-time update of CBF:

[0109] Due to the completion of the unmanned vehicle task or the addition of a new task, the equivalent CBF of the multi-unmanned vehicle system is reconstructed at the current moment; the b corresponding to the completed task l (x, t) needs to be removed from the equivalent CBF, and the new task corresponds to b k (x, t) is then added to the equivalent CBF;

[0110] Process unfinished STL tasks:

[0111] If the task is not completed by the specified time, k ,t k +dt0] time period, the CBF of a new arrival or formation task is introduced, and the STL semantics of the task is That is to meet the specified time [t k ,t k +1], if an unmanned vehicle arrives at a certain location area, or if several unmanned vehicles meet certain formation requirements, the corresponding CBF is designed according to the rules;

[0112] If you reach t k+1 =t k +dt0, the task has not been completed yet. Then the CBF constraint duration of the next time period is determined according to the completion of the task. Let L0 be the distance between the unmanned vehicle and the target position at the time of joining the task. The formation is the distance between a single unmanned vehicle and the formation position of another unmanned vehicle. Let L1 be t k+1 The distance of time;

[0113] The required duration of the CBF task at the next moment after redescription is In [t k+1 ,t k+1 +dt1] time period introduces the CBF of the redescribed arrival or formation task, and the STL semantics of the task is where t k+2 =t k+1 +dt1;

[0114] And so on. t k+n+1 =tk+n +dt n , until r≥h(x) is satisfied, that is, the task is completed;

[0115] Before the next trigger task is released, the CBF holding position or the relative position of the unmanned workshop is designed according to the rules;

[0116] Loop to check the completion of tasks until all tasks are completed.

[0117] Compared with the prior art, the present invention has the following beneficial effects:

[0118] 1. The present invention utilizes CBF constraints to solve the input control law of the multi-unmanned vehicle system in real time. The solution speed is fast and it is suitable for dynamically solving the input control law. There is always a suitable solution under the constraint of the input control law. Due to the robust design of CBF, the completion of obstacle avoidance and collision avoidance tasks can be guaranteed.

[0119] 2. The present invention is applicable to emergency scenarios with sudden tasks, such as temporary formation and arrival tasks in complex environments. Even if there are speed restrictions, decisions can be made as quickly as possible, so that all arrival and formation tasks can be safely completed within the specified time or in the shortest possible additional time.

[0120] 3. The present invention provides an input control law optimization algorithm for distributed control, which can greatly reduce the computational burden in real-time calculations and improve system control stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0121] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0122] Figure 1 This is the initial scenario of the multi-unmanned vehicle cooperation system involved in the embodiment of the present invention;

[0123] Figure 2 is a simulation path diagram under centralized control in an embodiment of the present invention;

[0124] Figure 3 is a numerical distribution diagram of a time-varying control barrier function under centralized control in an embodiment of the present invention;

[0125] Figure 4 is a numerical distribution diagram of a time-invariant control barrier function under centralized control in an embodiment of the present invention;

[0126] Figure 5 1 is a schematic diagram of the completion of a formation task under centralized control in an embodiment of the present invention;

[0127] Figure 6is a simulation path diagram under distributed control in an embodiment of the present invention;

[0128] Figure 7 is a numerical distribution diagram of a time-varying control barrier function under distributed control in an embodiment of the present invention;

[0129] Figure 8 is a numerical distribution diagram of a time-invariant control barrier function under distributed control in an embodiment of the present invention;

[0130] Fig. 9 is a schematic diagram of the completion of a formation task under distributed control in an embodiment of the present invention;

[0131] Fig.10 It is a schematic diagram of the quadratic programming solution strategy for the input control law in an embodiment of the present invention. DETAILED DESCRIPTION

[0132] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several variations and improvements may be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0133] The embodiment of the present invention provides a path planning and formation control method for a multi-unmanned vehicle system under a sequential logic task, comprising:

[0134] S1, establish the signal timing logic (STL) task specification framework for multiple unmanned vehicle systems;

[0135] S2, constructing a control barrier function (CBF) corresponding to the STL according to the task specification framework;

[0136] S3, solve the input control law of the multi-unmanned vehicle system under the constraints of CBF;

[0137] S4, completes the path planning and formation tasks of the multi-unmanned vehicle system under the input control law;

[0138] S5, loop detection task completion.

[0139] This method is aimed at the following multi-unmanned vehicle system path planning and formation control problems and is implemented based on the following principles.

[0140] The path planning and formation control problem of multiple unmanned vehicle systems under sequential logic tasks is:

[0141] Taking task allocation as an example, there are multiple known obstacle points in the known space, and a multi-unmanned system composed of M unmanned vehicles performs regional patrol tasks. After the M unmanned vehicles complete their tasks at their respective initial positions and return to their respective charging ports, some unmanned vehicles temporarily receive tasks to go to new target points to perform tasks. Some complex tasks require unmanned vehicles to form a fast formation in pairs, and vehicles with insufficient power will stay at the charging port until the power is fully charged. Assume that the position information of each unmanned vehicle is known by the control center, and the obstacles are fixed in position and shape. Assume that the M unmanned vehicles are M dynamic points on a two-dimensional plane in a limited working area. If there are relatively few vehicles, centralized control can be adopted, and the control center can centrally calculate the input control law and publish it to each unmanned vehicle in real time. If the scale of vehicles is relatively large, it is only necessary to publish the position information to the corresponding unmanned vehicle separately, and each unmanned vehicle will solve its own control input law separately to help the unmanned vehicle make decisions faster, and the control center is used to publish all task information.

[0142] Furthermore, consider M:=4 autonomous vehicles that can travel in all directions. The state of each autonomous vehicle v1,i∈{1,...,M} is represented as where p i :=[x i,1 ,x i,2 ] T represents the spatial position of the unmanned vehicle, x i,3 Represents the direction of the unmanned vehicle. Combine the states of all unmanned vehicles into a column vector

[0143] For a multi-autonomous vehicle system, each unmanned vehicle must further consider its obstacle avoidance mechanism to ensure that the unmanned vehicles do not come into contact with each other. By using the matrix where f i (x):=[f i,1 (x)f i,2 (x)0] T , for each v1,i∈{1,...,M}, its construction rule is:

[0144]

[0145]

[0146] where k i >0, and f i (x) is Lipschitz continuous, then the dynamic model of each unmanned vehicle is

[0147]

[0148] where u i represents the angular velocity of the wheel, i.e. the input, R i:=0.02 represents the radius of the wheel, the geometric constraint matrix B i The matrix is ​​as follows:

[0149]

[0150] Where L i :=0.2m represents the radius of the unmanned vehicle body.

[0151] g(x):=diag(g1(x1),...,g M (x M ))

[0152]

[0153] The equivalent multi-unmanned vehicle system model can be obtained as shown in formula (6).

[0154] Define the function sat(x) as follows:

[0155]

[0156] In this paper, we consider the boundedness of the disturbance and, without loss of generality, we set c i The absolute values ​​of the upper and lower limits of (x, t) are set to ||c i (x,t)|| ∞ =C=1, such as, let

[0157] c i (x, t): = [sat(x i，1 )sat(x i，2 )sat(x i，3 )] T

[0158] The path planning and formation control method for a multi-unmanned vehicle system of sequential logic tasks provided in an embodiment of the present invention has the following inventive concept:

[0159] The unmanned vehicle constructs the corresponding sub-CBF according to the initial position state information and the corresponding sequential logic tasks, and obtains the equivalent CBF through smoothing estimation of each sub-CBF, and solves the input control law under the common constraints of the equivalent CBF and the dynamic system, so that the CBF always satisfies the situation greater than 0. That is, the satisfaction of each task, and the input control law obtained at each moment is applied to the input at the next moment, so as to control the unmanned vehicle to complete the tasks of arrival, obstacle avoidance, collision avoidance and formation.

[0160] Due to the limitation of input and heading angle, when adding STL tasks that cannot be completed within the specified time, the ideal input obtained exceeds the motor speed and rudder angle limit of the system. The input after applying the limit will result in the actual CBF being non-positive, but it will not affect the solution of the input at the next moment. In other words, if the input control law obtained exceeds the speed and heading angle limit, the input should be constrained to the set value. This constraint may make the task unable to be completed within the specified time, that is, the speed is not fast enough, so the CBF will be negative. But it will not affect the solution of the input at the next moment.

[0161] Signal timing logic: The original purpose of signal timing logic is to specify and monitor the expected behavior of physical systems and impose timing constraints on events. It can standardize properties such as time and space variables, and automatically generate monitors to test the properties on each simulation trajectory. It has been used in many types of continuous and hybrid systems. Signal timing logic provides a formal method to describe optimization problems with complex timing logic constraints. Its robust semantics can use a real value to evaluate the degree to which the control trajectory meets or violates the task requirements. At the same time, its predicate logic can describe the task coupling between subsystems and can handle subtasks such as communication, obstacle avoidance, and collision avoidance.

[0162] The present invention aims at the collaborative task of a multi-unmanned vehicle system with complex sequential logic constraints, forms the control target of the multi-unmanned vehicle system according to the environment and task information, uses STL to establish a task specification framework, forms a formal representation method for the target task and constraints, and describes the physical meaning of each STL subtask by constructing CBFs corresponding to STL subtasks related to arrival, collision avoidance, obstacle avoidance, formation, etc., uses CBFs to form constraint conditions for the state space equations of the multi-unmanned vehicle system, uses a quadratic programming algorithm (Quadratic Programming, QP) to form a feedback control law for the objective function obtained from the input quantity, and iteratively solves the state space equations of the multi-unmanned vehicle system to design path planning and formation control algorithms to achieve intelligent control.

[0163] Based on the above invention concept and design principle, the present invention provides a preferred embodiment to execute S1 and establish a signal timing logic (STL) task specification framework of a multi-unmanned vehicle system, including:

[0164] S11, establish the state space model of multiple unmanned vehicle systems:

[0165] A multi-unmanned vehicle system with M unmanned vehicles is represented by an undirected graph Modeling consists of ε∈v×v represents the relationship between unmanned vehicles. n represents a single unmanned vehicle i i dimensional state trajectory column vector, m represents a single unmanned vehicle i i -dimensional input column vector. Let Where n: = n1 + ... + n M Represents the sum of the state dimensions of the multi-unmanned vehicle system. The state space model of a single unmanned vehicle is defined as follows:

[0166]

[0167] in These are all relatively conventional matrices determined by the system dynamic equations. The first variable x i is locally Lipschitz continuous and measurable in time t. i (x, t) can be used to simulate the given dynamic coupling formed by the connection between unmanned vehicles, and can also be used to simulate unknown dynamics that cannot be accurately measured or interference noise formed during the movement, as well as the impact of the wind speed in the simulation environment. In this embodiment, it is assumed that the size of the indeterminate term is bounded, that is, ||c i (x,t)||≥C, otherwise the state trajectory x of the system cannot be known.

[0168] S12, establish STL task standardization framework:

[0169] Based on STL formal statements, the sequential logic tasks of the multi-unmanned vehicle system are described in a standardized manner, and a mathematical model of the task scenario is established:

[0170]

[0171] φ:=G [a，b] ψ|F [a，b] ψ|ψ1U [a，b] ψ2|φ1∧φ2 (2-2)

[0172] Among them, ψ and φ are STL formal statements describing the temporal logic tasks of the multi-unmanned vehicle system. ψ1 and ψ2 are in the form of ψ in the above formula (2-1). They do not contain time constraints. In the physical sense, they represent whether the task of a single unmanned vehicle or multiple unmanned vehicles is satisfied. They are called Boolean formulas. φ1 and φ2 are in the form of φ in the above formula (2-2). The statement contains three temporal logic operators G, F, U, and the time constraint interval [a, b] for the execution of the corresponding task. It is called a temporal formula. Operators and time constraint intervals are combined into global, finally, and until operators: G [a,b] , F [a,b] ,U [a,b] . G [a,b] The ψ operator indicates that the constraint ψ must always be satisfied within the time interval [a, b]; F[a,b] The ψ operator indicates that the constraint ψ must be satisfied at least once in the future time interval [a,b]; ψ1U [a,b] The ψ2 operator indicates that constraint ψ2 needs to be satisfied at least once in the time interval [a, b] and constraint ψ1 needs to be satisfied continuously before constraint ψ2 is satisfied. STL is based on multiple predicates μ, and each predicate logic μ is defined by a predicate function that represents the actual physical meaning. To determine whether the actual state set of the system is in the expected state set, the predicate function is only related to the column vector of the system state space Related. The predicate μ is defined as follows:

[0173]

[0174] Based on the above embodiment, a preferred embodiment is provided to perform S2, constructing a CBF corresponding to the STL task according to the task framework, including:

[0175] S21, constraints that need to be satisfied when constructing CBF:

[0176] According to the requirements of each unmanned vehicle's arrival, obstacle avoidance, collision avoidance, and formation mission as well as the time constraints required for mission completion, the CBF corresponding to each mission is established.

[0177] For a single sequential logic task, define the following differentiable function in It is used to represent the state space trajectory variables of the system, and defines the following set

[0178]

[0179] This function is related to the system state trajectory and time. If there is an absolutely continuous function as follows Its state at the initial time of the time interval [t0, t1] is x(t0) = x0, and the initial state x0 is input into the above set, so that the function b(x0, t0) ≥ 0, and in the following time, b(x, t) ≥ 0 is always satisfied, that is, it satisfies Then the function b(x, t) can be called a control obstacle function, where the state x can involve the position state information of a single or multiple unmanned vehicles.

[0180] S22, smooth estimation of multiple CBFs:

[0181] If the function b(x, t) can satisfy b(x(t), t) ≥ 0 within the time constraint interval of a certain STL subtask, that is, it satisfies This means that for all t≥0, each unmanned vehicle can meet the task requirements. In order to solve the coupling factors in multiple STL semantics φ, that is, to meet the arrival, obstacle avoidance, collision avoidance, and formation tasks corresponding to all unmanned vehicle subsystems, p CBFs corresponding to the tasks can be considered. The minimum min operator can be used for smooth estimation. For the tasks of multiple unmanned vehicles, in order to meet them simultaneously, the following smooth estimation formula is available:

[0182]

[0183] Where η>0, p represents the number of effective CBFs, when η→∞, the following smooth estimate is obtained:

[0184]

[0185] Obviously, no matter how we choose the size of η, the following inequality holds:

[0186]

[0187] make

[0188]

[0189] It only needs to satisfy the constraint condition b(x,t)≥0, that is, the corresponding condition of the candidate CBF is satisfied within the time constraint interval of the STL task. If the forward set remains unchanged, the control obstacle function b corresponding to any subtask embedded in b(x,t) can be satisfied l (x,t)≥0 holds. Each of the subfunctions b l (x, t) are associated with a temporal logic formula G [al,bl] u l or F [al,bl] u l .

[0190] S23, CBF construction rules:

[0191] Under the above premise, the construction of b(x, t) is divided into step (a) and step (b).

[0192] Step (a): Let p = 1 and consider φ: = G [al,bl] u l and φ: = F [al,bl] u l There are two situations. In order to meet a certain robustness, it is necessary to make the maximum value of the predicate function satisfy At the same time, set the following time nodes:

[0193]

[0194] The above setting reflects the[al,bl] u l , it is only necessary to make the predicate logic u l Satisfies once to be True, and for G [al,bl] u l , then it is necessary to make the predicate logic u l Always holds True. For example, if Obviously, h l (x(0))≥r. Let the control obstacle subfunction have the following function structure b l (x,t):=-γ l (t)+h l (t), the predicate function has a certain robustness by setting the following parameters.

[0195]

[0196]

[0197]

[0198] By choosing the appropriate l,∞ : The error range between the position of the unmanned vehicle and the target point, and λ:γ l The coefficient of (t) can ensure that at any time Satisfy b l (x(t′),t′)≤h l (x(t′))-r.

[0199] Therefore, in the subsequent control strategy, at any time By satisfying the constraint b l (x(t′),t′)≥0, which can directly satisfy h l (x(t′))≥r, by choosing a suitable and λ, can satisfy the robust semantics ρ φ (x,0)≥r.

[0200] For a composite temporal logic operator, if there are multiple robustness requirements, the robustness setting of the total predicate function is strictly calculated according to the following formula:

[0201] ρ μ (x, t): = h(x(t)),

[0202]

[0203]

[0204]

[0205]

[0206]

[0207]

[0208] Step (b): For p>1, a more precise parameter definition step is required. Compared with step (a), b(x(0),0)≥0 is only determined by choosing γ l,0 is difficult to satisfy. Consider the function b(x, t): = -ln(exp(-b1(x, t))+exp(-b2(x, t))). By choosing a suitable γ l,0 , although it can satisfy both b1(x(0),0)≥0 and b2(x(0),0)≥0, the choice of η cannot guarantee that b(x(0),0)≥0 will definitely hold. Therefore, it is necessary to choose a sufficiently large η to improve the accuracy of the smooth estimation of b(x,t). At the same time, it is necessary to choose γ l,∞ , and select a suitable robustness r so that in the time interval [s0,s q ] always exists Satisfies b(x,t)≥0. By choosing appropriate η,r,λ,γ l,0 ,γ l,∞ The above-mentioned purpose can be achieved.

[0209] Based on the above embodiment, a preferred embodiment is provided to execute S3, solving the input control law of the multi-unmanned vehicle system under the constraint of CBF, including:

[0210] S31, establish the state space model of multiple unmanned vehicle systems:

[0211] For a centralized multi-unmanned vehicle system, consider M intelligent agent systems and a total of K STL semantics φ1...φ k , for each agent The status trajectory of the entire multi-unmanned system It is composed of the state space trajectory column vectors of each agent. Represents the dimension of the entire multi-unmanned system state space.

[0212] For included in v k For all agents in the state space model, the state space model can be reconstructed into the following dynamic system:

[0213]

[0214] System Input Determined by the state trajectory and time of the system. All agents v participating in these constraints k The equivalent state-space model parameters are as follows:

[0215]

[0216]

[0217]

[0218]

[0219] in May be part or all It is produced by the coupling of intelligent agents in the network.

[0220] S32, construct the constraints of the input control law, including three categories:

[0221] First, order Indicates that according to steps (a) and (b), φ is constrained k Related control obstacle function, for the equivalent dynamic system of the multi-unmanned vehicle system in the above formula, its centralized control obstacle function constraint condition is:

[0222]

[0223] The function is a κ-type Lipschitz function. For the constant C, since the disturbance function related to the state trajectory and time of all unmanned vehicles formed by the disturbance of the system and the internal coupling is bounded, we have the following inequality:

[0224] Second, the control obstacle function corresponding to the obstacle avoidance and collision avoidance tasks needs to always meet the corresponding spatial constraints within a certain time interval. The following non-time-varying control obstacle function is defined When the driverless car i∈v k Entering an unsafe area Right now When , the function is expressed as b(x i )<0, when maintained in a safe state, it behaves as And b(x i )≥0, let the set If the forward set needs to be unchanged, that is, in the local control law u i The state space trajectory solution for a single unmanned vehicle To keep the forward direction unchanged, the following time-invariant control barrier function constraints need to be satisfied:

[0225]

[0226] Third, if the environment is relatively complex, with relatively dangerous obstacles and involves triggering events, such as temporarily joining a formation task, more efficient online computing efficiency is required. In this case, the computing burden can be distributed to each unmanned vehicle, and the unmanned vehicle's own computing module can solve the input control law. The control obstacle function constraints corresponding to each unmanned vehicle solving the input control law are as follows:

[0227]

[0228]

[0229]

[0230]

[0231] The weight function

[0232]

[0233] In this embodiment and other embodiments, the weight function D i The settings are arbitrary, as long as the conditions are met

[0234]

[0235] where ||v k || represents the time-varying CBF corresponding to a coupled sequential logic task, b″(x i ,x j ,...,t), and the time-invariant CBF, b″(x i ,x j ,...) The number of unmanned vehicles involved. If an unmanned vehicle does not have any coupling tasks, there is no need to consider equations (9-3) and (9-4). It only needs to consider the equivalent time-varying CBF corresponding to its own STL task: b related to the arrival task i (x i ,t) and time-invariant CBF: b related to the obstacle avoidance task i (x i ) constitutes the CBF constraints, such as equations (9-1) and (9-2).

[0236] S33, generate the input control law under the constraint:

[0237] With the above CBF and the constraints of the input solution, the optimization problem is solved within the time range N, starting from t = 0, and after each time step Δt, within the time interval [t, t + Δt]. The optimization problem is described as a quadratic programming problem under the following centralized CBF constraints:

[0238] arg min u T u

[0239] st

[0240]

[0241] If we want to share the computational complexity, the input control law can be calculated by the small computing module of the unmanned vehicle. Then a distributed quadratic programming problem under CBF constraints is as follows:

[0242] arg min u i T u i

[0243] st

[0244]

[0245]

[0246]

[0247] Based on the above embodiments, a preferred embodiment is provided to execute S4 to complete the path planning and formation tasks of the multi-unmanned vehicle system under the input control law: the input control law solved at the current moment is applied to the system input at the next moment, and it is iterated in this way, so that each unmanned vehicle can complete its arrival, obstacle avoidance, collision avoidance, and formation tasks in a dangerous working environment.

[0248] Based on the above embodiment, a preferred embodiment is provided to perform S5, cyclically detecting task completion, including:

[0249] S51, real-time update of CBF:

[0250] Due to the completion of the unmanned vehicle task or the addition of a new task, it is necessary to reconstruct the equivalent CBF of the multi-unmanned vehicle system at the current moment. The b corresponding to the completed task l (x, t) needs to be removed from the equivalent CBF, and the new task corresponds to b k (x,t) then needs to be added to the equivalent CBF.

[0251] S52, process the unfinished STL tasks:

[0252] And due to the speed and heading angle limitations of the unmanned vehicle in actual situations, if the task is not completed within the specified time, it is necessary to re-describe the corresponding STL task and reconstruct the CBF for it. The specific method is as follows:

[0253] In [t k ,tk +dt0] time period, the CBF of a new arrival or formation task is introduced, and the STL semantics of the task is That is to meet the specified time [t k ,t k +1], if an unmanned vehicle reaches a certain location area, or several unmanned vehicles meet certain formation requirements, then the corresponding CBF is designed according to steps (a) and (b). k+1 =t k +dt0 If the task is not completed, the CBF constraint duration of the next time period is determined based on the completion of the task. Let L0 be the distance between the unmanned vehicle and the target position at the time of joining the task, and the formation is the distance between a single unmanned vehicle and the formation position of another unmanned vehicle, such as the distance between one unmanned vehicle and the left side of another unmanned vehicle. Let L1 be t k+1 The distance between the moments. The required duration of the CBF task at the next moment after redescription is In [t k+1 ,t k+1 +dt1] time period introduces the CBF of the redescribed arrival or formation task, and the STL semantics of the task is where t k+2 =t k+1 +dt1.

[0254] And so on. t k+n+1 =t k+n +dt n , until r≥h(x) is satisfied, that is, the task is completed. Before the next task is triggered, design the CBF to maintain the position or the relative position of the unmanned workshop according to steps (a) and (b). Loop to check the task completion status until all tasks are completed.

[0255] The path planning and formation control method of a multi-unmanned vehicle system under a sequential logic task provided by an embodiment of the present invention can ensure the robustness and stability of the system and the handling of emergencies under complex sequential logic constraints.

[0256] The technical solution provided by the embodiment of the present invention is further described below with reference to the accompanying drawings and through simulation examples.

[0257] S100, establish a state space model of a multi-unmanned vehicle system; see Figure 1 ,The multi-unmanned vehicle system consists of 4 unmanned vehicles. Assume that the initial positions of the unmanned vehicles are [-1,0], [0,0], [1,0], [0,5] and the initial heading angles are all 0°. The scene includes three obstacles and a wall. Unsafe_set1, Unsafe_set2, Unsafe_set3, U wall, respectively Figure 1 The area surrounded by blue, orange, and yellow lines and the purple vertical line represent the charging area of ​​each unmanned vehicle. charge1 ,U charge2 ,U charge3 ,U charge4 , , are [10,-1], [10,0], [10,1], [11,0] respectively. The initial task is for the unmanned vehicles to return to their respective charging ports from their respective starting points within 15 seconds. Between 15 seconds and 25 seconds, unmanned vehicle No. 1 needs to go to the target point [10,5] and needs to be accompanied by unmanned vehicle No. 2. Both need to form a team urgently. The team requires that unmanned vehicle No. 2 is located one meter west of unmanned vehicle No. 1; unmanned vehicle No. 3 needs to go to the target point [10,-10] to perform the task; unmanned vehicle No. 4 needs to stay at the [11,0] charging station to charge due to insufficient power. Their STL semantics are expressed as follows. p i Represents the two-dimensional position information of unmanned vehicle i, which is a two-dimensional column vector, x i，1 , x i，2 , x i，3 represents the x, y axis position and heading angle of the unmanned vehicle i, and

[0258] S200, normalize the above sequential logic tasks:

[0259] φ1:=F [0，15] (||p1-[10 -1] T ||≤0.1)∧G [0，25] (||p1-[5 0] T ||≥3)∧F [15，25] (||p1-[10 5] T ||≤0.1)∧G [0，25] (||p1-[10 2.5] T ||≥1)∧G [0，25] (||p1-[10 -5] T ||≥2)∧G [0，25] (x 1，1 ≤17)

[0260] φ2:=F [0，15] (||p2-[10 0] T ||≤0.1)∧G [0，25] (||p2-[5 0] T ||≥3)∧G [0，25] (||p2-[10 2.5] T ||≥1)∧G [0，25] (||p2-[10 -5] T ||≥2)∧F[15，16] (||p1-p2+[4 0] T ||≤0.5)∧G [16，25] (||p1-p2+[4 0] T ||≤0.5)∧G [0，25] (x 2，1 ≤17)

[0261] φ3:=F [0，15] (||p3-[10 1] T ||≤0.1)∧G [0，25] (||p3-[5 0] T ||≥3)∧F [15，25] (||p3-[10 -10] T ||≤0.1)∧G [0，25] (||p3-[10 2.5] T ||≥1)∧G [0，25] (||p3-[10 -5] T ||≥2)∧G [0，25] (x 3，1 ≤17)

[0262] φ4:=F [0，15] (||p4-[11 0] T ||≤0.1)∧G [0，25] (||p4-[5 0] T ||≥3)∧G [0，25] (||p4-[10 2.5] T ||≥1)∧G [0，25] (||p4-[10 -5] T ||≥2)∧G [0，25] (x 4，1 ≤17)∧G [15，25] (||p4-[11 0] T ||≤0.1)

[0263] This complex sequential logic task indicates that unmanned vehicles 1, 2, 3, and 4 need to arrive at their respective charging stations within the time interval [0, 15]. [0，15] (||p1-[10 -1] T ||≤0.5) means that unmanned vehicle 1 needs to reach the target point [10, -1] within the time constraint, and the radius of the target point is 0.5; among them, unmanned vehicle 4 needs to stay at [11 0] T Stay and charge in an area with a central radius of 0.5; the temporal logic formula F representing the formation task [15，16] (||p1-p2+[4 0] T ||≤0.5)∧G [16，25](||p1-p2+[4 0] T ||≤0.5) means that unmanned vehicle 1 and unmanned vehicle 2 must maintain a distance of 4 meters in the east-west direction within the time interval [15, 16], and the difference between unmanned vehicle 2 and the target position shall not exceed 0.5 meters, and this formation shall be maintained afterwards. In the time interval [15, 25], unmanned vehicles 1 and 2 need to go to the new target point to complete the task. When completing the collaborative task, the unmanned vehicles must not touch each other at all times. Figure 1 The three obstacles and the wall in the temporal logic formula G [0，25] (||p1-[5 0] T ||≥3) means that the obstacle with the center at [5, 0] and a radius of 3 must not be touched within the total STL task time [0, 25]. [0，25] (x 2，1 ≤17) means that the unmanned vehicle 2 can never touch the wall x: =17.

[0264] S300, CBF design for sequential logic tasks

[0265] The CBF design method corresponding to each STL subtask is as mentioned in steps (a) and (b), and its specific form is:

[0266] For example, F [0，15] (||p1-[10 -1] T ||≤0.5), its CBF is designed as follows:

[0267] b1(x, t) = (1.1 + || [10, -1] T -[-1, 0] T ||)-t*(1+||[10,-1] T -[-1, 0] T ||)-||p1-[10,-1] T || (12)

[0268] When t=0,b1(x,0)=1.1,When t=15,b(x,15)=0.1-||p1-[10,-1] T ||, if u1 is used as the constraint of CBF T u1 is the objective function to solve the input control law. By satisfying b1(x, t) ≥ 0, the arrival mission of the No. 1 unmanned vehicle can be completed. The CBF design of the other arrival missions is similar.

[0269] S400, solving the input control law of the multi-unmanned vehicle system under the constraints of the CBF:

[0270] The equivalent CBF of the multi-unmanned vehicle system is:

[0271]

[0272] For the obstacle avoidance STL task G [0，25] (||p1-[5 0] T ||≥3), its time-invariant CBF design is as follows:

[0273] b1(x)=||p1-[5,0] T ||-3 (13)

[0274] Similarly, under the time-invariant CBF constraint, b1(x)≥0 can be satisfied within the time [0, 25], that is, obstacles are avoided. The same is true for the other CBF designs for obstacle avoidance.

[0275] The equivalent time-invariant CBF of the multi-unmanned vehicle system is:

[0276]

[0277] A more special design of the present invention is that, for the time node t=15, the CBF corresponding to the arrival task of the unmanned vehicle is reconstructed, and the obstacle avoidance-related non-time-varying CBF does not need to be changed.

[0278] For formation mission F [15，16] (||p2-p1+[4 0] T ||≤0.5)∧G [16，25] (||p2-p1+[4 0] T ||≤0.5), unmanned vehicle 1 and unmanned vehicle 2 need to meet the formation requirements within 1 second and in this posture, unmanned vehicle 2 follows unmanned vehicle 1 to the new target point. Due to the limitation of system input, -1000<u(i)<1000, the formation action may not be completed within 1 second.

[0279] Furthermore, the present invention provides an embodiment of redefining the STL task and reconstructing its CBF, including:

[0280] In the time period [15, 15+dt0], the CBF constraint duration of the next time period is determined according to the completion status of the formation, where

[0281] dt0=1, b(x,t)=γ(t)-||p1-p2+[4,0] T ||,

[0282] b(x,15)=γ(15)-||p1-p2+[4,0] T ||,γ(15)=||p i -p j +[4, 0] T ||

[0283] b(x,16)=γ(16)-||p1-p2+[4,0] T ||, γ(16)=0.5

[0284] γ(t) is usually designed as a linear function.

[0285] If the formation task is not completed, calculate the next CBF constraint duration dt1

[0286] Let L0 = || p1 - p2 + [4, 0] T || t＝15 , L1=||p1-p2+[4,0] T || t＝16 ,

[0287] At time t=16, the following CBF is redesigned:

[0288] b(x,16)=γ(16)-||p1-p2+[4,0] T ||,γ(16)=||p1-p2+[4,0] T ||

[0289] b(x,16+dt1)=γ(16+dt1)-||p1-p2+[4,0] T ||,γ(16+dt1)=0.5

[0290] Similarly, if the formation requirements are not met, k =16+dt1 to calculate the next CBF constraint duration:

[0291] L0=||p1-p2+[4,0] T || t＝15 ,

[0292] And so on, until L is satisfied. n =||p1-p2+[4,0] T ||≤0.5, the lower limit dt of the constraint duration needs to be met in the embodiment of the present invention i ≥0.5*dt0

[0293] Until time t K Meet the formation requirements L n =||p1-p2+[4,0] T ||≤0.5, set CBF to maintain this formation:

[0294] b(x,t)=γ(t)-||p1-p2+[4,0] T || (14)

[0295] where γ(t)∈(0,0.5], Similarly, γ(t) is designed as a linear function.

[0296] S500, solving the input control law of the multi-unmanned vehicle system under the constraint of the CBF, and completing the path planning and formation tasks of the multi-unmanned vehicle system based on the input control law;

[0297] In the embodiment of the present invention, after the equivalent CBF, including b′(x, t) and b″(x), is designed, within a time length of N=25 seconds, a quadratic programming operation of solving the input control law is performed every step length Δt=0.02, i.e., the quadratic programming strategy:

[0298] arg min u T u

[0299] st

[0300]

[0301] To share the computational complexity among each unmanned vehicle, assuming that each unmanned vehicle contains an independent computing module and the communication delay between unmanned vehicles is negligible, the following distributed quadratic programming strategy in method (3) can be adopted:

[0302] arg min u i T u i

[0303] st

[0304]

[0305]

[0306]

[0307] And apply each set of solved inputs to the corresponding unmanned vehicle state space model.

[0308] The embodiment of the present invention adopts a centralized control solution system input control law, and therefore only includes one equivalent time-varying CBF, b′(x, t) and one equivalent time-invariant CBF: b″(x).

[0309] Apply the u calculated at time t to the equivalent state space model of the multi-unmanned vehicle system, that is, the formula:

[0310]

[0311] The 4th-5th order Runge-Kutta algorithm is used to solve the above differential equations to obtain the position state of each unmanned vehicle at time t+Δt, and the process is repeated. In the process, it is necessary to repeatedly detect whether a trigger event is encountered, such as reaching a new target point, emergency formation task, etc., so as to choose whether it is necessary to reconstruct the equivalent CBF of the multi-unmanned system.

[0312] In the embodiments of the present invention, it is assumed that the construction time of the CBF can be ignored.

[0313] For the overall simulation results of the multi-unmanned vehicle system, please refer to Figure 2 .

[0314] The numerical distribution of the control obstacle function joins the formation task at t = 15. Due to the existence of speed limit, heading angle limit and disturbance, when approaching t = 15 and t = 25, although some unmanned vehicles have reached the target point, the equivalent time-varying CBF, b′(x, t) < 0 will appear, that is, the unmanned vehicles cannot be strictly restricted to a small range of the target point. Please refer to Figure 3 .

[0315] Since no obstacles are touched during the embodiment of the present invention, the non-time-varying equivalent CBF: b″(x)≥0 is established everywhere. Figure 4 .

[0316] For the temporary formation task, since it is impossible to complete the formation requirement within the specified 1 second, you can only apply for additional time to complete the formation task. The completion rate of the formation task over time, that is, the distance between the position point 4 meters west of the unmanned vehicle 2 and the unmanned vehicle 1 at each moment, is expressed by the following formula: dis(t) = || p1-p2+[4,0] T || t . See Figure 5 The formation mission is completed at 16.20s. The embodiment of the present invention also uses distributed control to solve the input control law, including multiple b′(x, t) and b″(x).

[0317] For example, for the time-varying CBF of unmanned vehicle 1: b′1(x1, t), it is expressed as the arrival task: F [0，15] (||p1-[10 -1] T ||≤0.1)

[0318] The corresponding function is as shown in formula (12).

[0319] For the non-time-varying CBF of unmanned vehicle 1: b″1(x), it represents all obstacle avoidance-related tasks of unmanned vehicle 1:

[0320] G [0，25] (||p1-[5 0] T||≥3)∧G [0，25] (||p1-[10 2.5] T ||≥1)∧G [0，25] (||p1-[10 -5] T ||≥2)∧G [0，25] (x 1，1 ≤17)

[0321] The corresponding smoothing estimates of multiple functions in the form of formula (13) are specifically in the form of:

[0322]

[0323] The b corresponding to a single G operator sequential logic task l (x) is in the form of formula (13). The CBF design rules of other unmanned vehicles are similar.

[0324] For formation mission F [15，16] (||p2-p1+[4 0] T ||≤0.5)∧G [16，25] (||p2-p1+[4 0] T ||≤0.5), and its coupled time-varying CBF is as follows: b′5(x1, x2, t) = γ(t)-||p1-p2+[4, 0] T ||

[0325] When the time domain and step size remain unchanged, the trigger conditions are set the same as those of the centralized control, and the input control law is solved according to the distributed control obstacle function constraint (11).

[0326] For simulation results of distributed control, please refer to Figure 6 In the simulation experiment, the solution rate of distributed control is much higher than that of centralized control, but the design of CBF is more complicated.

[0327] For the numerical distribution of time-varying CBF, see Figure 7 Due to the limitation of obstacles and speed, the CBF of the formation task will not satisfy the condition of b′5(x1, x2, t) ≥ 0 everywhere. The input control law adopts the optimal solution within the speed range, so it will be less than 0.

[0328] For the numerical distribution of time-invariant CBF, see Figure 8 .

[0329] For the completion of the formation mission, please refer to Fig. 9 The formation completed its mission in 19.72 seconds.

[0330] For the selection of optimal solution strategy for input control law, please refer to Fig.10 .

[0331] The above embodiment of the present invention provides a method for path planning and formation control of a multi-unmanned vehicle system under a sequential logic task, comprising the following steps: establishing a signal sequential logic (STL) task specification framework for the multi-unmanned vehicle system; constructing a control barrier function (CBF) corresponding to the STL task according to the task framework; solving the input control law of the multi-unmanned vehicle system under the constraint of the CBF; completing the path planning and formation tasks of the multi-unmanned vehicle system under the input control law; and cyclically detecting the task completion degree. The present invention is aimed at the path planning and formation control problems of a multi-unmanned vehicle system under sequential logic. Considering the disturbances caused by the operating environment of the unmanned vehicle system, such as wind speed and electromagnetic interference, it is necessary to add a certain robust value to the constraint of the CBF when solving the input control law. The present invention relates to the triggering of tasks, and it is necessary to monitor the position information of each unmanned vehicle subsystem in real time. It belongs to a method for solving input online, and it is necessary to update the CBF form used for solving the input control law according to the triggering conditions.

[0332] The above are only preferred embodiments of the invention and are not intended to limit the scope of the invention. That is, any equivalent changes and modifications made according to the content of the patent application scope of the present invention should be within the technical scope of the present invention.

[0333] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various modifications or variations within the scope of the claims, which does not affect the essence of the present invention. The above preferred features can be used in any combination without conflicting with each other.

Claims

1. A path planning and formation control method for a multi-unmanned vehicle system with sequential logic tasks, characterized in that: include: Establish a signal timing logic STL task specification framework for multiple unmanned vehicle systems; Constructing the control barrier function CBF corresponding to the STL according to the task specification framework; Solve the input control law of the multi-unmanned vehicle system under the constraints of the CBF; Complete the path planning and formation tasks of the multi-unmanned vehicle system under the input control law; Circularly detecting the completion of the path planning and formation tasks; Wherein, solving the input control law of the multi-unmanned vehicle system under the constraints of the CBF includes: Establish a state space model of multiple unmanned vehicle systems; For a centralized multi-unmanned vehicle system, there are M intelligent agent systems and a total of K STL semantic φ1...φ k , for each agent The status trajectory of the entire multi-unmanned system It is composed of the state space trajectory column vectors of each agent; Represents the dimension of the entire multi-unmanned system state space; For included in v k All agents in the state space model are reconstructed into a dynamic system: Among them, the input of the system Determined by the state trajectory and time of the system; all agents v participating in these constraints k The equivalent state space model parameters are: where c k (x,t) consists of some or all It is generated by the coupling of intelligent agents in The method further includes: constructing constraints for input control laws based on the state space model of the multi-unmanned vehicle system; the constraints include three categories: (1) Order Represents the constraint φ k Related control obstacle function, the equivalent state space model of the multi-unmanned vehicle system, and its centralized control obstacle function constraint conditions are: The function is a κ-type Lipschitz function; C is a constant, and (2) The control obstacle function corresponding to the obstacle avoidance and collision avoidance tasks always satisfies the corresponding spatial constraints within a certain time interval; the following non-time-varying control obstacle function is defined When the unmanned vehicle i∈v k Entering an unsafe area Right now When , the function is expressed as b(x i )<0, when maintaining in a safe state, it behaves as And b(x i )≥0, let the set If the forward set needs to be constant; that is, in the local control law u i State space trajectory solution for a single unmanned vehicle To keep the forward direction unchanged, the following time-invariant control barrier function constraints need to be satisfied: (3) If a temporary team is required, the computational burden is distributed to each unmanned vehicle, and the computing module of the unmanned vehicle solves the input control law. The control obstacle function constraint corresponding to the input control law solved by each unmanned vehicle is: The weight function where ||v k || represents the time-varying CBF corresponding to a coupled sequential logic task, b″(x i ,x j ,...,t), and the time-invariant CBF, b″(x i ,x j ,...) the number of unmanned vehicles involved; The method further includes: generating an input control law under the constraints based on the CBF and the constraints: In the time range N, starting from t = 0, every time step Δt, the optimization problem is solved in the time interval [t, t + Δt]. The optimization problem is described as a quadratic programming problem under centralized CBF constraints: If we want to share the computational complexity, the input control law can be calculated by the small computing module of the unmanned vehicle. Then the quadratic programming problem under the distributed CBF constraint is as follows:

2. The path planning and formation control method for a multi-unmanned vehicle system with sequential logic tasks according to claim 1 is characterized in that: The signal timing logic STL task specification framework for establishing a multi-unmanned vehicle system includes: Establish a state space model of multiple unmanned vehicle systems; The multi-autonomous vehicle system consists of an undirected graph Composition, of which ε∈v×v represents the association between unmanned vehicles, and M represents the number of individual unmanned vehicles; make n represents a single unmanned vehicle i i dimensional state trajectory column vector, m represents a single unmanned vehicle i i -dimensional input column vector; make Represents multiple n-dimensional state trajectory vectors of unmanned vehicles, where n:=n1+...+n M Represents the sum of the state dimensions of the multi-unmanned vehicle system; The state space model of a single unmanned vehicle is defined as follows: in The first variable x i It is locally Lipschitz continuous and measurable in time t; ||c i (x, t)||≤C, used to simulate the given dynamic coupling formed by the connection between unmanned vehicles, or to simulate unknown and imprecise dynamics or interference noise generated during movement, as well as the impact of wind speed in the simulation environment; The state space model of the single unmanned vehicle is combined to obtain the state space model of the multi-unmanned vehicle system.

3. The path planning and formation control method for a multi-unmanned vehicle system with sequential logic tasks according to claim 2 is characterized in that: The said establishing a signal timing logic STL task specification framework for the multi-unmanned vehicle system also includes: establishing an STL task standardization framework; Based on STL formal statements, the temporal logic tasks of the multi-unmanned vehicle system are described, and an STL task standardization framework is established: φ:=G [a,b] ψ|F [a,b] ψ|ψ1U [a,b] ψ2|φ1∧φ2 (2-2) Among them, ψ and φ are STL formal statements describing the temporal logic tasks of the multi-unmanned vehicle system. ψ1 and ψ2 are the same as ψ, indicating whether the tasks of a single unmanned vehicle or multiple unmanned vehicles in collaboration are satisfied. φ1 and φ2 are of the same form as φ. Their respective statements contain three temporal logic operators G, F, U, and the time constraint interval [a, b] for the execution of the corresponding task. Combine the operator and the time constraint interval into global, finally, until operators: G [a,b] , F [a,b] ,U [a,b] ; G [a,b] The ψ operator indicates that the constraint ψ must always be satisfied within the time interval [a, b]; F [a,b] The ψ operator indicates that the constraint ψ must be satisfied at least once in the future time interval [a, b]; ψ1U [a,b] The ψ2 operator indicates that constraint ψ2 needs to be satisfied at least once in the time interval [a, b] and constraint ψ1 needs to be satisfied continuously before constraint ψ2 is satisfied; Predicate logic μ is composed of predicate functions To determine whether the actual trajectory of the multi-unmanned vehicle system is within the time and space range that meets the requirements of the STL task, the predicate function is only related to the system state space column vector Related; The predicate logic μ is defined as:

4. The path planning and formation control method for a multi-unmanned vehicle system with sequential logic tasks according to claim 1 is characterized in that: The step of constructing the control barrier function CBF corresponding to the STL according to the task specification framework includes: The constraints that need to be satisfied when constructing CBF; For a single sequential logic task, define a differentiable function in Variables used to represent the state space trajectory of the system; Defining a Collection If there exists an absolutely continuous function Its state at the initial moment of the time interval [t0, t1] is x(t0) = x0, and the initial state x0 is input into the set, so that the function b(x0, t0) ≥ 0, and b(x, t) ≥ 0 is always satisfied in the following time, that is, Then the control obstacle function b(x,t) is obtained; where the state x involves the position state information of a single or multiple unmanned vehicles; Smooth estimation of multiple control barrier functions (CBFs); For multiple unmanned vehicle tasks, there is the following smooth estimation formula: Where η>0, p represents the number of effective CBFs, when η→∞, the smooth estimate is: make Satisfy the constraint condition b(x,t)≥0, that is, satisfy the CBF corresponding to the time constraint interval of the STL task If the forward set remains unchanged, then the control obstacle function b corresponding to any subtask embedded in b(x,t) is satisfied l (x,t)≥0 holds; Each subfunction b l (x, t) is associated with a temporal logic formula G [al,bl] u l or F [al,bl] u l .

5. The path planning and formation control method for a multi-unmanned vehicle system with sequential logic tasks according to claim 4 is characterized in that: The construction rules of the CBF include: (a) Let p = 1 and consider φ: = G [al,bl] u l and φ: =F [al,bl] u l Two situations; The maximum value of the predicate function satisfies At the same time, set the following time nodes: For F [al,bl] u l , it is only necessary to make the predicate logic u l True if satisfied once; For G [al,bl] u l , then it is necessary to make the predicate logic u l Always keep True; like Satisfy h l (x(0))≥r; Let the control obstacle sub-function have the following function structure b l (x,t):=-γ l (t)+h l (t), the predicate function is obtained by setting the following parameters; By selecting the error range γ between the position of the unmanned vehicle and the target point l,∞ and γ l The coefficient λ of (t) ensures that at any time Satisfy b l (x(t′),t′)≤h l (x(t′))-r; In the subsequent control strategy, that is, at any time By satisfying the constraint b l (x(t′),t′)≥0, directly satisfying h l (x(t′))≥r, by choosing and λ, satisfying robust semantics ρ φ (x,0)≥r; For a composite temporal logic operator, the robustness of the total predicate function is calculated according to the following formula: (b) For p>1, consider the function b(x,t): = -ln(exp(-b1(x,t))+exp(-b2(x,t))); by choosing η, r, λ, γ l,0 ,γ l,∞ , satisfying the constraints.

6. The path planning and formation control method for a multi-unmanned vehicle system with sequential logic tasks according to claim 1 is characterized in that: The path planning and formation tasks of the multi-unmanned vehicle system are completed under the input control law, including: applying the input control law solved at the current moment to the system input at the next moment, and iterating in this way, so that each unmanned vehicle can complete its own arrival, obstacle avoidance, collision avoidance, and formation tasks in a dangerous operating environment.

7. The path planning and formation control method for a multi-unmanned vehicle system with sequential logic tasks according to claim 5 is characterized in that: The completion of the path planning and formation tasks is cyclically detected, including: Real-time update of CBF: Due to the completion of the unmanned vehicle task or the addition of a new task, the equivalent CBF of the multi-unmanned vehicle system is reconstructed at the current moment; the b corresponding to the completed task l (x, t) needs to be removed from the equivalent CBF, and the new task corresponds to b k (x, t) is then added to the equivalent CBF; Process unfinished STL tasks: If the task is not completed by the specified time, k ,t k +dt0] time period, the CBF of the new arrival or formation task is introduced, and the STL semantics of the task is F [tk,tk+1] (r≥h(x)), that is, to meet the requirements of k ,t k +1] arrives at a certain location area, or several unmanned vehicles meet certain formation requirements, then the corresponding CBF is designed according to the constructed rules; In t k+1 =t k If the task is not completed at time +dt0, the CBF constraint duration of the next time period is determined based on the completion of the task. Let L0 be the distance between the unmanned vehicle and the target position at the time of joining the task, and the formation is the distance between a single unmanned vehicle and the formation position of another unmanned vehicle. Let L1 be the distance between t k+1 The distance of time; The required duration of the CBF task at the next moment after redescription is In [t k+1 ,t k+1 +dt1] time period, the CBF of the redescribed arrival or formation task is introduced, and the STL semantics of the task is F [tk+1,tk+2] (r≥h(x)), where t k+2 =t k+1 +dt1; And so on. t k+n+1 =t k+n +dt n , until r≥h(x) is satisfied, that is, the task is completed; Loop to check the completion of tasks until all tasks are completed.

Images