CN116774735A - A UAV cluster trajectory planning method and system based on edge computing - Google Patents

Abstract

The invention discloses an unmanned aerial vehicle cluster track planning method and system based on edge calculation, and relates to the field of unmanned aerial vehicle path planning, wherein the method comprises the following steps: constructing a feasible region constraint of the unmanned aerial vehicle according to a kinetic equation of the unmanned aerial vehicle and an obstacle in a three-dimensional space flown by the unmanned aerial vehicle; constructing coupling collision constraint of the unmanned aerial vehicle according to the set distance kept between the unmanned aerial vehicle and other unmanned aerial vehicles in the unmanned aerial vehicle cluster; constructing an energy consumption model of the unmanned aerial vehicle from take-off time to landing time; adding two penalty function items into the energy consumption model to obtain an objective function of the unmanned aerial vehicle, wherein the first penalty function item is the collision penalty between the unmanned aerial vehicle and the obstacle, and the second penalty function item is the collision penalty of the multi-unmanned aerial vehicle collision; and (3) based on the feasible region constraint and the coupling collision constraint, utilizing the edge calculation force of each unmanned aerial vehicle to adopt a penalty function method to solve an objective function through iteration, and obtaining a track planning result of the unmanned aerial vehicle cluster. The invention reduces the energy consumption and improves the solving speed.

Description

A UAV cluster trajectory planning method and system based on edge computing

Technical field

The present invention relates to the technical field of UAV path planning, and in particular to a UAV cluster trajectory planning method and system based on edge computing.

Background technique

UAV trajectory planning refers to calculating the optimal path for the UAV to fly in 3D space given the starting point and target point to achieve the predetermined mission goal. In the application of drones, trajectory planning is very important because it can help drones autonomously complete various tasks, such as patrol, logistics, monitoring, rescue, etc.

In UAV trajectory planning, there are two key issues, trajectory planning of a single UAV and trajectory planning of multiple UAVs.

In terms of single UAV trajectory planning, common UAV trajectory planning algorithms include optimization-based, search-based and learning-based methods. Optimization-based methods usually transform the trajectory planning problem into a mathematical optimization problem and use optimization algorithms to find the optimal solution. These methods can transform the task objective into an optimization objective by defining an appropriate cost function, and then use mathematical optimization algorithms to find the optimal solution. The advantage of these methods is that they can guarantee the global optimal solution, but the calculation amount is large and the efficiency is low. Search-based methods (e.g. , RRT, genetic algorithm, etc.) usually solve the problem by finding the optimal path in the search space. This type of method usually requires defining a heuristic function to guide the search process in order to find the optimal solution faster. The advantage of these methods is that they can find better solutions in a shorter time, but they are not guaranteed to find the global optimal solution. The learning-based method uses machine learning algorithms to learn the trajectory planning strategy of the UAV. These methods usually require a large amount of training data and take a long time to train. Due to the uncertainty of neural networks, practical applications are greatly hindered.

In multi-UAV trajectory planning, methods can be divided into centralized methods and distributed methods according to different calculation and control methods. The centralized approach uses a centralized controller to complete the entire planning calculation. These algorithms aim at global optimization and have high computational complexity. At the same time, they require the UAV to continuously upload its own status and environmental information to the central node and receive control instructions from the central node, which puts forward high requirements on communication bandwidth and is not easy to Scale to large-scale drone swarms. Although the calculation efficiency is low, it can ensure the global optimal solution and cooperative obstacle avoidance between UAVs. The distributed method allows each drone to independently complete its own planning calculations. Each UAV generates its own trajectory based on its own status and environmental information, and coordinates adjacent UAVs through communication to achieve coordinated flight of the UAV group. The distributed method has low computational complexity and can be easily extended to large-scale drone swarms. However, since each drone only plans based on local information, it is difficult to ensure global optimization and high-quality collaborative obstacle avoidance between drones.

In the current UAV cluster path planning algorithm, there are the following significant problems. One is that it consumes a lot of computing resources. When the scale of the drone cluster is large, the centralized method needs to process a large amount of drones and environmental information, which requires a huge amount of calculation and is difficult to obtain a high-quality solution within a limited time. The second is that UAV communication has high bandwidth requirements. Whether it is a centralized or distributed algorithm, when frequently exchanging information with central nodes or other UAVs, it needs to rely on higher communication bandwidth to ensure the real-time nature of information exchange. The third is that it is difficult to find the optimal path in a complex environment. When the environmental space is complex or there are a large number of obstacles, it is very difficult to find a safe and optimal path for the drone cluster, and it is easy to fall into a local optimum.

Contents of the invention

The purpose of the present invention is to provide a UAV cluster trajectory planning method and system based on edge computing to reduce energy consumption and increase the solution speed.

In order to achieve the above objects, the present invention provides the following solutions:

A UAV cluster trajectory planning method based on edge computing, including:

Construct dynamic equations for UAV motion;

Construct the feasible domain constraints of the drone based on the dynamic equations and the obstacles in the three-dimensional space where the drone flies;

Based on maintaining a set distance between the drone and other drones in the drone cluster, the coupling collision constraints of the drone are constructed;

Discretize the time period between the take-off time and the landing time of the UAV, and construct the energy consumption model of the UAV from the take-off time to the landing time;

The first penalty function term and the second penalty function term are added to the energy consumption model to obtain the objective function of the UAV. The first penalty function term is the collision penalty between the UAV and the obstacle. The third penalty function term is the collision penalty between the UAV and the obstacle. The second penalty function term is the collision penalty for multi-UAV collision;

Based on the feasible region constraint and the coupling collision constraint, the edge computing power of each UAV is used to solve the objective function by iteratively using the penalty function method to obtain the trajectory planning result of the UAV cluster; the UAV Each drone in the cluster shares information through the broadcast and sharing mechanism.

The invention also discloses an edge computing-based UAV cluster trajectory planning system, which includes:

Dynamic equation building module, used to construct the dynamic equation of UAV motion;

A feasible region constraint building module is used to construct the feasible region constraints of the UAV based on the dynamic equations and obstacles in the three-dimensional space where the UAV flies;

The coupling collision constraint building module is used to construct the coupling collision constraints of the UAV based on maintaining a set distance between the UAV and other UAVs in the UAV cluster;

The energy consumption model building module is used to discretize the time period between the take-off time and the landing time of the UAV, and construct the energy consumption model of the UAV from the take-off time to the landing time;

The objective function building module is used to add the first penalty function term and the second penalty function term to the energy consumption model to obtain the objective function of the UAV. The first penalty function term is the distance between the UAV and the obstacle. The collision penalty of , the second penalty function term is the collision penalty of multiple UAV collisions;

A solving module, configured to use the edge computing power of each UAV based on the feasible region constraint and the coupling collision constraint to solve the objective function by iteratively using the penalty function method to obtain the trajectory planning result of the UAV cluster; Each drone in the drone cluster shares information through a broadcast and sharing mechanism.

According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

This invention uses the edge computing power of the UAV to optimize the trajectory, which can reduce energy consumption and increase the solution speed. The first penalty function term and the second penalty function term are added to the objective function, which improves the solution speed and solution success rate.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the drawings needed to be used in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some of the drawings of the present invention. Embodiments, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without exerting creative efforts.

Figure 1 is a schematic flow chart of a UAV cluster trajectory planning method based on edge computing provided by an embodiment of the present invention;

Figure 2 is a schematic diagram of a UAV model provided by an embodiment of the present invention;

Figure 3 is a schematic structural diagram of a UAV cluster trajectory planning system based on edge computing provided by an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

The purpose of the present invention is to provide a UAV cluster trajectory planning method and system based on edge computing to reduce energy consumption and increase the solution speed.

In order to make the above objects, features and advantages of the present invention more obvious and understandable, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

Embodiment 1 provides a UAV cluster trajectory planning method based on edge computing.

As shown in Figure 1, a UAV cluster trajectory planning method based on edge computing includes the following steps.

Step 101: Construct the dynamic equation of drone motion.

In this embodiment, all the drones in the drone cluster are four-rotor drones.

Assume that the center of mass of the drone is located at its geometric center, as shown in Figure 2. In order to accurately simulate the behavior of a quadcopter (UAV), the dynamic equations of UAV motion are given.

;

in, Represents the position coordinates of the UAV, used to describe the translation of the UAV, x, y and z are the x-axis, y-axis and z-axis coordinates respectively. φ represents the roll angle of the UAV rotating around the x-axis, θ represents the pitch angle of the UAV rotating around the y-axis, ψ represents the yaw angle of the UAV rotating around the z-axis, I _xx , I _yy , and I _zz represent respectively The moment of inertia of the UAV around the x-axis, y-axis and z-axis, u ₁ represents the first virtual control quantity, u ₂ represents the second virtual control quantity, u ₃ represents the third virtual control quantity, u ₄ represents the fourth virtual control quantity Quantity, m represents the mass of the drone, g represents the acceleration of gravity,/> ,/> ,/> ,/> ,/> ,/> are the second derivatives of x, y, z, φ, θ, and ψ respectively.

In Figure 2, F ₁ , F ₂ , F ₃ and F ₄ are the pulling forces of the four rotors respectively.

In order to simplify the description, the abstract expression form of the above dynamic equation is:

;

in, Represents the state quantity of the drone, which is a vector composed of x, y, z, φ, θ and ψ,/> for/> The first derivative of Indicates the control amount of the drone,/> is the vector composed of u ₁ , u ₂ , u ₃ and u ₄ ,/> Represents an abstract expression form of dynamic equations.

Step 102: Construct the feasible domain constraints of the UAV based on the dynamic equations and obstacles in the three-dimensional space where the UAV flies.

For obstacles in three-dimensional space, use several spheres Fit their shapes so that these spheres completely surround the obstacles, where/> .

Among them, r represents the radius of the sphere surrounding the drone.

The feasible region constraints are expressed as:

.

in, Represents feasible region constraints,/> Indicates the position coordinates of the drone,/> represents the position coordinates of obstacle p, P represents the number of obstacles, S ^p represents the sphere surrounding obstacle p, and r ^p represents the radius of S ^p .

Although this method reduces the feasible region to a certain extent, it can achieve a balance between the complete feasible region and the amount of calculation.

Step 103: Construct coupling collision constraints of the UAV based on maintaining a set distance between the UAV and other UAVs in the UAV cluster.

In order to ensure safety, each quadcopter drone needs to maintain a certain safe distance (r ^safe ) from other drones within the same time and space range.

The coupling collision constraints are expressed as:

.

in, Represents coupled collision constraints,/> Indicates the position coordinates of the drone,/> Indicates the trajectories of other drones,/> Indicates the position coordinates of other drones,/> Indicates the set distance. The value range of j is the serial number of all other drones in the drone cluster except the current computing drone.

It should be pointed out that for the same space, different drones can pass by at different times. In other words, only when time and space coincide, it means that the trajectory of the drone violates the constraints.

Step 104: Discretize the time period from the take-off time to the landing time of the UAV, and construct an energy consumption model of the UAV from the take-off time to the landing time.

Efficient use of energy in a quadcopter is important to increase its range and endurance, which can lead to more efficient operation and lower costs. The energy consumed by a quadcopter drone can be defined as follows.

;

in, is the take-off time of the drone,/> is the landing time of the drone,/> is the identity matrix.

This embodiment uses the direct point allocation method to discretize the continuous time curve into a finite time series. Continuous time is discretized into sample points, which means that time is discretized as And the state quantity is discretized as =x[0]…x[N].

When optimizing the above discrete problem, the given initial guess (initial value of the trajectory) can be written as , the optimized trajectory is recorded as/> .

Under this discretization method, the optimization problem of UAV can be expressed as follows:

;

The constraints are: .

Among them, input control has an upper limit of control amount and control volume lower limit/> . This problem has realistic feasible domain constraints ( ) and coupled collision constraints (/> ).

Step 105: Add the first penalty function term and the second penalty function term to the energy consumption model to obtain the objective function of the UAV. The first penalty function term is the collision penalty between the UAV and the obstacle, The second penalty function term is the collision penalty for multiple UAV collisions.

The second penalty function term is the collision penalty for a collision between the current drone and other drones in the drone cluster.

It is assumed that the trajectories of other drones are known and determined, that is, the constraints can be regarded as known and determined spaces. On this basis, this embodiment considers convexizing the optimization space.

According to the collision between the UAV and the obstacle, the feasible domain constraint ( ), define the first penalty function term/> .

.

in, Represents the distance between the drone and each obstacle. When the drone does not collide, the penalty term is close to 0. When it collides, the penalty term increases approximately linearly according to the severity of the collision. The final overall penalty value is the sum of the penalty values of the drone to all obstacles. In the above penalty function,/> It is the first penalty factor, which can be dynamically adjusted according to the optimization situation. Compare/> An exact penalty function that takes into account the depth of the drone's collision and ensures that the function is continuously differentiable.

Similarly, the second penalty function term for multi-UAV collision is defined .

;

in, Indicates the distance j between the current calculation drone and other human-machine machines. The value range of j is the serial number of all other drones in the drone cluster except the current calculation drone.

The initial UAV optimization problem can be transformed into the following problem segment trajectory planning, that is, the objective function is expressed as:

;

The constraints of the objective function are: ;

Among them, among them, Represents the energy consumption model,/> () represents the first penalty function term,/> () represents the second penalty function term,/> Represents the first penalty factor,/> Represents the second penalty factor,/> Represents the status quantity of the drone,/> Indicates the control amount of the drone,/> =x[0]…x[N], x[0] represents the state quantity of the initial state, x[N] represents the state quantity of the Nth moment,/> Indicates the status quantity corresponding to the take-off time,/> Represents the state quantity corresponding to the landing time,/> Indicates the upper limit of the control amount,/> Indicates the lower limit of the control amount.

This embodiment uses the penalty function method to solve the objective function. The solution process includes:

Step S1: Given the initial first penalty factor and the second penalty factor/> .

Step S2: Use CasADI and IPOPT to solve the objective function, and check the feasible region constraints and coupled collision constraints.

Step S3: When the constraints are not met (the feasibility constraints are not met or the coupling collision constraints are not met), the penalty factor is expanded (when the feasibility constraints are not met, the first penalty factor is expanded according to the first set step size, and when the feasibility constraints are not met) When the coupling collision constraint is satisfied, the second penalty factor is expanded according to the second set step size). When the constraints are satisfied (both the feasibility constraint and the coupling collision constraint are satisfied), the penalty factor is reduced (the first penalty factor is reduced according to the third set step size). penalty factor and second penalty factor).

Step S4: Loop steps S2 to S3 until the smallest penalty factor is found (the difference between the objective function value of the current iteration and the objective function value of the previous iteration is less than the set threshold). The loop ends and the optimal solution is obtained. The optimal solution is the optimal penalty factor.

While obtaining the optimal penalty factor, the optimized trajectory of the UAV is determined.

Changes in the penalty factor and number of iterations relative to the penalty function during a certain iteration process. for constraints , since the constraint is not violated during the solution process, the penalty factor continues to decrease until it reaches the minimum allowed value. For constraints/> , there are no constraints violated in the first two iterations, so the corresponding penalty factor decreases. However, a constraint violation occurs in the third iteration, and the algorithm appropriately expands the penalty factor until a feasible solution is obtained in the fifth iteration. It is believed that the penalty factor is optimal at this time.

Step 106: Based on the feasible region constraint and the coupling collision constraint, use the edge computing power of each UAV to solve the objective function by iteratively using the penalty function method to obtain the trajectory planning result of the UAV cluster; Each drone in the drone cluster shares information through the broadcast and sharing mechanism.

Among them, in each iteration process in step 106, each UAV uses its edge computing power based on the current trajectory value to solve the objective function corresponding to each UAV using the penalty function method to obtain the optimal penalty factor and optimized trajectory. value. Through multiple iterations until the trajectory value of each UAV is stable, the trajectory planning result of the UAV cluster is obtained. After each UAV obtains the optimized trajectory value in each iteration, the optimized trajectory value is shared with other UAVs through the broadcast and sharing mechanism.

During each iteration, for the nth drone: the initial trajectory guess value based on the current iteration , use the penalty function method to solve the objective function corresponding to the nth UAV, and obtain the optimized trajectory value/> .

Guess the value based on the initial trajectory of the current iteration and the optimized trajectory value obtained by the current iteration/> , determine the initial trajectory guess value of the nth UAV in the next iteration/> .

Through multiple iterations, until the trajectory values of each UAV are stable, the trajectory planning result of the UAV cluster is obtained.

In each iteration, the objective function is used to optimize the UAV whose trajectory is not stable. Each iteration obtains a complete trajectory planning result of the UAV cluster. After multiple iterations, the trajectory value of each UAV is obtained. If both are stable, the final trajectory planning result of the UAV cluster will be obtained.

Based on the initial trajectory guess value of the current iteration and the optimized trajectory value obtained by the current iteration, the initial trajectory guess value of the nth UAV in the next iteration is determined, that is, the initial trajectory guess value update strategy specifically includes:

According to the formula , determine the initial trajectory guess value of the nth UAV in the next iteration;

in, Represents the initial trajectory guess value of the nth UAV in the next iteration,/> Represents the initial trajectory guess value of the nth UAV in the current iteration, /> Indicates the optimized trajectory value of the nth UAV in the current iteration.

The definition of UAV trajectory stability is: .

Among them, tol is a given threshold.

Under the conditions of the initial trajectory guess value update strategy, the solution method of the objective function includes:

Step a1:

For all drones:

Step a1-1: Do not consider space to solve the objective function.

Step a1-2: Check Spatial constraints, if the constraints are met, that is, no collision occurs, the algorithm ends early.

Step a2:

For all drones:

Step a2-1: Use the last trajectory of other drones as space to solve the objective function.

Step a2-2: When the trajectory is stable, the UAV no longer participates in subsequent loop solutions.

Step a3: When there is only the last UAV with an unstable trajectory, the objective function is used to optimize the UAV once, and the final UAV cluster trajectory planning result is obtained.

Each drone in the drone cluster shares information through a broadcast and sharing mechanism, that is, each drone shares the optimal solution after obtaining the optimal solution of the objective function.

After each planning is completed, its planning information is broadcast to surrounding drones. To this end, this embodiment discloses a four-drone information exchange system based on LoRaWAN:

1. The four-UAV information exchange system consists of a UAV cluster and a gateway station. Each drone is equipped with a LoRa communication module, and the gateway station is used to relay information exchange between drones.

2. The drones and gateway stations, and the drones communicate with each other through the LoRa network. The gateway station is responsible for forwarding the information of each drone to realize fully connected communication between drones.

3. Each drone periodically broadcasts its own ID and path information through the LoRa network. Other drones and gateway stations receive this information and grasp the flight status of each drone.

4. The gateway station monitors the communication between each drone. If it detects that a drone has lost contact with the system, it will consider the drone to be faulty and notify other drones of the drone status through the LoRa network to avoid missions. Assigned to the missing drone.

5. After the drone lands, the gateway station collects the flight logs of each drone to evaluate communication quality and system performance.

The four-UAV information exchange system uses the LoRa network to realize simple information exchange among the four UAVs. The UAVs use broadcast communication and gateway station forwarding to achieve full connectivity. The system has small communication volume, is suitable for UAVs to exchange simple instructions and status information, and meets basic collaborative flight requirements. Overall, this system provides a simple and efficient multi-UAV communication solution.

The present invention constructs a UAV cluster trajectory planning method based on edge computing, which can complete information transfer between UAVs and utilize the edge computing power of UAVs for trajectory optimization.

The penalty function method proposed by the present invention can effectively improve the success rate and solution speed of single UAV trajectory solution. This algorithm is designed to achieve effective trajectory solution on a low computing power platform.

The initial value update strategy proposed by the present invention can significantly improve the solving speed. By giving optimized initial guesses, the solver can find the optimal solution more quickly.

The distributed solving algorithm proposed by the present invention does not rely on a central server, allowing the UAV cluster to have better independent intelligence.

Embodiment 2 provides an edge computing-based UAV cluster trajectory planning system.

As shown in Figure 3, a UAV cluster trajectory planning system based on edge computing includes:

The dynamic equation building module 201 is used to construct the dynamic equation of the UAV movement.

The feasible region constraint construction module 202 is used to construct the feasible region constraints of the UAV based on the dynamic equations and obstacles in the three-dimensional space where the UAV flies.

The coupling collision constraint construction module 203 is used to construct coupling collision constraints of the UAV based on maintaining a set distance between the UAV and other UAVs in the UAV cluster.

The energy consumption model building module 204 is used to discretize the time period between the take-off time and the landing time of the UAV, and construct an energy consumption model of the UAV from the take-off time to the landing time.

The objective function building module 205 is used to add the first penalty function term and the second penalty function term to the energy consumption model to obtain the objective function of the drone. The first penalty function term is the relationship between the drone and the obstacle. The second penalty function term is the collision penalty for multiple UAV collisions.

The solving module 206 is configured to use the edge computing power of each UAV based on the feasible region constraint and the coupling collision constraint to solve the objective function by iteratively using the penalty function method to obtain the trajectory planning result of the UAV cluster. ; Each drone in the drone cluster shares information through a broadcast and sharing mechanism.

Each embodiment in this specification is described in a progressive manner. Each embodiment focuses on its differences from other embodiments. The same and similar parts between the various embodiments can be referred to each other. As for the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For relevant details, please refer to the description in the method section.

This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method and the core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the present invention There will be changes in the specific implementation methods and application scope of the ideas. In summary, the contents of this description should not be construed as limitations of the present invention.

Claims (8)

1. An unmanned aerial vehicle cluster track planning method based on edge calculation is characterized by comprising the following steps:

constructing a kinetic equation of unmanned aerial vehicle motion;

constructing a feasible region constraint of the unmanned aerial vehicle according to the dynamics equation and the obstacle in the three-dimensional space of the unmanned aerial vehicle;

according to the method, the unmanned aerial vehicle and other unmanned aerial vehicles in the unmanned aerial vehicle cluster are kept at a set distance, and coupling collision constraint of the unmanned aerial vehicle is built;

discretizing a time period from the take-off time to the landing time of the unmanned aerial vehicle, and constructing an energy consumption model from the take-off time to the landing time of the unmanned aerial vehicle;

adding a first penalty function term and a second penalty function term into the energy consumption model to obtain an objective function of the unmanned aerial vehicle, wherein the first penalty function term is a collision penalty between the unmanned aerial vehicle and an obstacle, and the second penalty function term is a collision penalty for multiple unmanned aerial vehicle collisions;

based on the feasible region constraint and the coupling collision constraint, solving the objective function by using the edge calculation force of each unmanned aerial vehicle through iteration and adopting a penalty function method to obtain a track planning result of the unmanned aerial vehicle cluster; and each unmanned aerial vehicle in the unmanned aerial vehicle cluster performs information sharing through a broadcasting and sharing mechanism.

2. The unmanned aerial vehicle cluster track planning method based on edge calculation according to claim 1, wherein the unmanned aerial vehicle cluster track planning result is obtained by solving the objective function by using a penalty function method through iteration by utilizing the edge calculation force of each unmanned aerial vehicle based on the feasible region constraint and the coupling collision constraint, and specifically comprises the following steps:

during each iteration, for the nth drone:

solving an objective function corresponding to the nth unmanned aerial vehicle by using a penalty function method based on the initial trajectory guess value of the current iteration to obtain an optimized trajectory value;

determining an initial track guess value of the n-th unmanned aerial vehicle in the next iteration according to the initial track guess value of the current iteration and the optimized track value obtained by the current iteration;

and (3) obtaining a track planning result of the unmanned aerial vehicle cluster through multiple iterations until the track values of all the unmanned aerial vehicles are stable.

3. The unmanned aerial vehicle cluster track planning method based on edge calculation according to claim 2, wherein the method is characterized in that the method comprises the following steps of determining an initial track guess value of an nth unmanned aerial vehicle in the next iteration according to the initial track guess value of the current iteration and the optimized track value obtained by the current iteration, and specifically comprises the following steps:

according to the formulaDetermining an initial track guess value of the nth unmanned aerial vehicle in the next iteration;

wherein ,representing the initial trajectory guess value of the nth unmanned aerial vehicle of the next iteration, +.>Representing the initial trajectory guess value of the nth drone of the current iteration, +.>And representing the optimized track value of the nth unmanned aerial vehicle in the current iteration.

4. The unmanned aerial vehicle cluster trajectory planning method based on edge calculation of claim 3, wherein the condition for stabilizing the unmanned aerial vehicle trajectory value is:

；

where tol is a given threshold.

5. The unmanned aerial vehicle cluster trajectory planning method based on edge computation of claim 1, wherein the objective function is expressed as:

；

the constraint conditions of the objective function are as follows:；

wherein ,representing energy expenditure model, +.>() A first penalty function term is represented,/>() Representing a second penalty function term, ">Representing a first penalty factor, ">Representing a second penalty factor, ">Representing the state quantity of the unmanned aerial vehicle, +.>Indicating the control quantity of the unmanned aerial vehicle, +.>=x[0]…x[N]，x[0]State quantity x [ N ] representing initial state]State quantity representing the nth time, +.>State quantity corresponding to take-off time +.>State quantity corresponding to the landing time is represented by +.>Indicating the upper limit of the control amount,/-, and>indicating a lower control amount limit.

6. The unmanned aerial vehicle cluster trajectory planning method based on edge computation of claim 1, wherein the feasible region constraint is expressed as:

；

wherein ,representing feasible region constraints, +.>Representing the position coordinates of the unmanned aerial vehicle, +.>Representing the position coordinates of the obstacle P, P representing the number of obstacles S ^p Representing a sphere surrounding an obstacle p, r ^p Represent S ^p Is set, and the radius of (a) is set.

7. The unmanned aerial vehicle cluster trajectory planning method based on edge computation of claim 1, wherein the coupling collision constraint is expressed as:

；

wherein ,representing coupled crash constraints, +.>Representing the position coordinates of the unmanned aerial vehicle, +.>Representing trajectories of other unmanned aerial vehicles, +.>Representing the position coordinates of other unmanned aerial vehicles, +.>Indicating the set distance.

8. Unmanned aerial vehicle cluster track planning system based on edge calculation, characterized by comprising:

the dynamic equation construction module is used for constructing a dynamic equation of the unmanned plane motion;

the feasible region constraint construction module is used for constructing the feasible region constraint of the unmanned aerial vehicle according to the dynamics equation and the obstacle in the three-dimensional space of the unmanned aerial vehicle;

the coupling collision constraint construction module is used for constructing coupling collision constraint of the unmanned aerial vehicle according to the fact that the unmanned aerial vehicle is kept at a set distance from other unmanned aerial vehicles in the unmanned aerial vehicle cluster;

the energy consumption model construction module is used for discretizing the time period from the take-off time to the landing time of the unmanned aerial vehicle and constructing an energy consumption model from the take-off time to the landing time of the unmanned aerial vehicle;

the objective function construction module is used for adding a first penalty function item and a second penalty function item into the energy consumption model to obtain an objective function of the unmanned aerial vehicle, wherein the first penalty function item is the collision penalty between the unmanned aerial vehicle and an obstacle, and the second penalty function item is the collision penalty of the collision of multiple unmanned aerial vehicles;

the solving module is used for solving the objective function by using the edge computing force of each unmanned aerial vehicle based on the feasible region constraint and the coupling collision constraint through iteration and adopting a penalty function method to obtain a track planning result of the unmanned aerial vehicle cluster; and each unmanned aerial vehicle in the unmanned aerial vehicle cluster performs information sharing through a broadcasting and sharing mechanism.