Intelligent cooperative control method for unmanned aerial vehicle group

Abstract

The invention relates to the field of unmanned aerial vehicle group control, in particular to an intelligent cooperative control method for an unmanned aerial vehicle group. Comprising the steps of receiving and preprocessing environment data, unmanned aerial vehicle state data and task data, and performing data fusion processing on the preprocessed data; performing unmanned aerial vehicle group task allocation by adopting a multi-objective optimization genetic algorithm to generate a task allocation scheme; generating an optimal path based on the task allocation scheme; the unmanned aerial vehicle executes a task according to the optimal path, and the flight path of the unmanned aerial vehicle is adjusted by using a dynamic obstacle avoidance function; a new optimal path is generated according to the current position of the unmanned aerial vehicle, the unmanned aerial vehicle continues to execute the task according to the new optimal path, and the unmanned aerial vehicle returns to the landing point after the task is completed and uploads the collected flight data and environment data. The problems that an existing selection mechanism usually depends too much on a single standard, and most of existing path planning methods only consider a static environment; and the problem that the unmanned aerial vehicle is easy to collide or deviate in the task execution process due to the lack of real-time obstacle avoidance capability is solved.

Description

Technical Field

[0001] The present invention relates to the field of drone swarm control, and in particular to a drone swarm intelligent collaborative control method. Background Art

[0002] With the rapid development of drone technology, drone swarms have been widely used in many fields, such as logistics distribution, agricultural management, inspection and monitoring, and disaster relief. Drone swarm collaborative control technology plays a vital role in achieving efficient collaborative operations of multiple drones. Currently, researchers have developed a variety of methods and algorithms to optimize the task allocation, path planning, and dynamic obstacle avoidance of drone swarms to improve the overall efficiency and safety of drone swarms. Among them, optimization techniques such as genetic algorithms, ant colony algorithms, and particle swarm optimization algorithms are widely used in the task allocation and path planning of drone swarms, and dynamic obstacle avoidance technology based on sensor data has also been widely studied and applied.

[0003] Although existing drone technology has made progress in the intelligent collaborative control of drone swarms, there are still problems in the three key areas of drone swarm task allocation, path planning and dynamic obstacle avoidance: In terms of task allocation, the task allocation method of drone swarms is difficult to take into account the task completion time, power consumption rate and task complexity of the drone swarm at the same time, and has weak adaptability to dynamic changes and environmental uncertainties; in terms of path planning, traditional path planning algorithms perform well in static environments, but have low efficiency and insufficient reliability in dynamic environments; in terms of dynamic obstacle avoidance, it mainly relies on sensor data, has poor adaptability when facing a variety of dynamic obstacles, and is prone to delayed response or excessive avoidance, resulting in a circuitous flight path, increasing the time and energy consumption for task completion, and lacking real-time and robustness. Summary of the invention

[0004] The embodiments of the present invention propose an intelligent collaborative control method for drone swarms to solve the problems that the existing selection mechanism usually relies too much on a single standard, which easily leads to insufficient diversity of the population and affects the search ability of the global optimal solution; most of the existing path planning methods only consider static environments and lack the real-time obstacle avoidance capability for obstacles in dynamic environments; in complex practical application scenarios, drones need to face dynamic factors such as moving obstacles, and traditional static path planning methods are difficult to cope with the above challenges; and the lack of real-time obstacle avoidance capability will cause drones to easily collide or deviate from the path during the execution of the mission, seriously affecting the safety and efficiency of the mission.

[0005] The intelligent collaborative control method of a drone swarm comprises the following steps:

[0006] S 1: Receive and preprocess environmental data, drone status data and task data to obtain preprocessed data, and perform data fusion processing on the preprocessed data to obtain fused data; use a multi-objective optimization genetic algorithm to perform drone group task allocation and generate a task allocation plan;

[0007] S2: Based on the task allocation plan, the optimal path is generated using the optimal path planning algorithm based on the graph neural network; the UAV performs the task according to the optimal path, and the flight path of the UAV is adjusted using the dynamic obstacle avoidance function; the node information in the graph structure is updated according to the current position of the UAV, and a new optimal path is generated using the optimal path planning algorithm based on the graph neural network. The UAV continues to perform the task according to the new optimal path, returns to the landing point after the task is completed, and uploads the collected flight data and environmental data.

[0008] Preferably, the S1 specifically includes:

[0009] In the implementation process of the multi-objective optimization genetic algorithm, the population is initialized, the initial population is generated, and the initial task allocation plan for each individual in the population is randomly generated; the initial task allocation plan is evaluated and optimized by calculating the fitness value.

[0010] Preferably, the S1 specifically includes:

[0011] In the process of allocating tasks of drone swarms using multi-objective optimization genetic algorithm, the improved roulette algorithm is used to select individuals with high fitness values ​​to enter the mating pool.

[0012] Preferably, the S1 specifically includes:

[0013] Based on the selected individuals, single-point crossover operation and random mutation operation are performed to obtain the next generation population; the fitness value of each individual in the next generation population is calculated, and the Pareto front solution is obtained by using fast non-dominated sorting; according to the fitness value, an improved roulette algorithm is used to select individuals with high fitness values ​​in the Pareto front solution to perform single-point crossover operation and random mutation operation to generate a new population; through iterative operation, an optimized Pareto front solution is output, and a task allocation plan is selected from the optimized Pareto front solution.

[0014] Preferably, the S2 specifically includes:

[0015] The optimal path planning algorithm based on graph neural network combines graph neural network, convolutional neural network and fully connected neural network to improve feature extraction and path planning decisions.

[0016] Preferably, the S2 specifically includes:

[0017] In the process of calculating the optimal path by the optimal path planning algorithm based on graph neural network, a graph structure is established for each drone based on the task allocation plan, and the graph neural network is used to extract the initial features of the drone graph structure.

[0018] Preferably, the S2 specifically includes:

[0019] The initial features are input into the convolutional neural network for processing to obtain the features processed by the convolutional neural network; the features processed by the convolutional neural network are input into the fully connected neural network for path planning decision to obtain the optimal path.

[0020] Preferably, the S2 specifically includes:

[0021] When the UAV performs tasks according to the optimal path, it detects surrounding obstacles in real time. When there are obstacles, it uses the dynamic obstacle avoidance function to calculate the obstacle avoidance vector and determine the direction and distance of obstacle avoidance. The flight direction and position of the UAV are adjusted according to the obstacle avoidance vector, and the UAV flies in the calculated direction and moves according to the indicated position.

[0022] Preferably, the S2 specifically includes:

[0023] The dynamic obstacle avoidance function formula is as follows:

[0024]

[0025] in, is the obstacle avoidance vector; α is the obstacle avoidance coefficient; M is the total number of obstacles currently detected; w γ is the dynamic weight of the γth obstacle; d γ is the distance between the current UAV and the γth obstacle; It is the direction vector from the current drone to the γth obstacle, indicating the direction of obstacle avoidance.

[0026] The beneficial effects of the technical solution of the embodiment of the present invention are:

[0027] 1. The multi-objective optimization genetic algorithm improves the efficiency and accuracy of drone swarm task allocation by optimizing the fitness value and selection mechanism, combined with the improved roulette algorithm, enhances the diversity of the population and the global search capability, ensures that each drone can reasonably allocate tasks, and maximizes task completion efficiency and resource utilization.

[0028] 2. Use graph neural networks to extract initial features, and combine convolutional neural networks and fully connected neural networks for processing and path planning decisions, which effectively handles path planning problems in complex environments and improves the globality and accuracy of path planning.

[0029] 3. Combined with real-time sensor data, define dynamic obstacle avoidance functions, flexibly adjust the flight path of the drone according to the real-time position and dynamic changes of obstacles, ensure the accuracy and timeliness of obstacle avoidance, and improve the safety and stability of the drone when performing tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the intelligent collaborative control method for drone swarms described in the present invention. DETAILED DESCRIPTION

[0031] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0032] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0033] The specific scheme of the intelligent collaborative control method of a swarm of unmanned aerial vehicles provided by the present invention is described in detail below with reference to the accompanying drawings.

[0034] Refer to the attached Figure 1 , which shows a flow chart of a method for intelligent collaborative control of a drone swarm provided by an embodiment of the present invention, the method comprising the following steps:

[0035] S1: Receive and preprocess environmental data, drone status data and task data to obtain preprocessed data, and perform data fusion processing on the preprocessed data to obtain fused data; use a multi-objective optimization genetic algorithm to perform drone group task allocation and generate a task allocation plan;

[0036] First, receive environmental data provided by the control center personnel, such as geographic information and obstacle locations; UAV status data, such as the number of UAVs, real-time location coordinates, flight speed, flight altitude, power consumption rate, flight trajectory, and UAV performance parameters; task data, such as the number of tasks, task target location, task complexity, task completion time, and task priority; the types of data provided by the control center personnel are set according to the specific implementation scenario;

[0037] Secondly, the environmental data, drone status data and mission data are preprocessed; the preprocessing includes data cleaning and format conversion to obtain preprocessed data; the preprocessed data is fused using weighted average fusion technology to obtain fused data. The preprocessing and weighted average fusion technology are existing technologies and will not be described in detail here;

[0038] Finally, a multi-objective optimization genetic algorithm is used to perform initial task assignment on the fused data and determine the specific tasks of each UAV; the multi-objective optimization genetic algorithm improves the efficiency and accuracy of task assignment of UAV swarms, enhances the diversity of the population and the global search capability, and ensures that each UAV can be reasonably assigned tasks to maximize the efficiency of task completion and resource utilization by defining and optimizing the fitness value and the selection mechanism based on the improved roulette algorithm.

[0039] In a multi-objective optimization genetic algorithm, a population contains multiple individuals, each of which represents a specific UAV task allocation scheme. The UAV task allocation scheme is a specific strategy for allocating tasks to corresponding UAVs, usually represented as a vector or array. The position of each element in the vector represents the task number, and the value of each element represents the number of the UAV that performs the corresponding task.

[0040] The process of multi-objective optimization genetic algorithm for UAV swarm task allocation is as follows:

[0041] First, the population is initialized. The initial population is generated by random generation and heuristic generation methods. According to the number of drones and the number of tasks, the initial task allocation plan for each individual in the population is randomly generated. The population initialization method is an existing technology, and the specific population initialization method is set according to the specific implementation scenario.

[0042] Then, the initial task allocation plan is evaluated and optimized by calculating the fitness value; the fitness value combines the task completion time, power consumption and task complexity to evaluate the initial task allocation plan of each individual. The fitness value is defined as:

[0043]

[0044] Among them, f is the fitness value of individual i, which is used to evaluate the quality of the initial task allocation plan of individual i; T i is the task completion time of individual i, which is obtained by accumulating the time it takes for individual i to complete all assigned tasks; E i is the power consumption of individual i, which means the total power required by individual i to complete the task. It is obtained by accumulating the power consumption of individual i to complete all assigned tasks. The power consumption can be estimated according to the UAV power consumption model in the prior art, which will not be elaborated here. iis the task complexity of individual i, which is obtained by summing up the task complexity of all tasks assigned to individual i; T max and T min is the maximum and minimum task completion time of all individuals; E max and E min is the maximum and minimum power consumption of all individuals; w1, w2, w3 are the weights of task completion time, power consumption and task complexity, which are used to balance the importance of task completion time, power consumption and task complexity in the fitness value and are set according to the specific application scenario;

[0045] Furthermore, an improved roulette algorithm is used to select individuals with high fitness values ​​to enter the mating pool. The improved roulette algorithm formula is as follows:

[0046]

[0047] Among them, p i is the probability of individual i being selected; f y is the fitness value of individual y; is the total number of individuals in the population; the improved roulette algorithm uses logarithmic operations to avoid numerical overflow problems and ensure the numerical stability of the improved roulette algorithm formula;

[0048] Next, a single-point crossover operation is used to cross-pair the selected individuals to generate new offspring individuals. A random mutation operation is used to randomly mutate the newly generated offspring individuals to increase the diversity of the population, avoid falling into the local optimum, and obtain the next generation population. The fitness value of each individual in the next generation population is recalculated, and a fast non-dominated sort is used to obtain a Pareto front solution, which is a set of non-dominated optimal solutions in a multi-objective optimization problem; further, an improved roulette algorithm is used to select individuals with high fitness values ​​in the Pareto front solution according to the fitness value to perform a single-point crossover operation and a random mutation operation to generate a new population; repeat the above steps, and continuously optimize the population through multiple iterations. The iterative termination condition is to reach the maximum number of iterations or the fitness value change of the new population is less than the preset threshold. The maximum number of iterations and the preset threshold are specifically set according to the specific implementation scenario; fast non-dominated sorting, single-point crossover operation and random mutation operation techniques are all existing technologies and will not be described here.

[0049] The output result of the multi-objective optimization genetic algorithm is a set of optimized Pareto front solutions. Decision makers select the solution that best meets current needs from the optimized Pareto front solutions as the task allocation plan based on actual needs and specific application scenarios.

[0050] S2: Based on the task allocation plan, the optimal path is generated using the optimal path planning algorithm based on the graph neural network; the UAV performs the task according to the optimal path, and the flight path of the UAV is adjusted using the dynamic obstacle avoidance function; the node information in the graph structure is updated according to the current position of the UAV, and a new optimal path is generated using the optimal path planning algorithm based on the graph neural network. The UAV continues to perform the task according to the new optimal path, returns to the landing point after the task is completed, and uploads the collected flight data and environmental data.

[0051] The optimal path planning algorithm based on graph neural network combines three technologies: graph neural network (GNN), convolutional neural network (CNN) and fully connected neural network (FNN). On the basis of the existing technologies of GNN, CNN and FNN, it improves the feature extraction and path planning decision-making parts, introduces the flight speed, power consumption rate, task priority and task distance of the drone, improves the accuracy and adaptability of path planning, and better meets the effect of drone mission requirements.

[0052] The process of calculating the optimal path by the optimal path planning algorithm based on graph neural network is as follows:

[0053] First, refer to the task allocation plan and use expert experience to build a graph structure for each drone. The nodes of the graph represent the position of the drone in the flight environment when performing the task, including the starting point, end point and intermediate waypoints; the edges of the graph represent the flight path, and the weight of the edges takes into account the mission distance, flight speed, power consumption rate, mission priority and obstacle information; among them, the mission distance refers to the total distance between the drone's current starting position and the mission target position; then, the graph neural network (GNN) is used to extract the initial features of a drone graph structure, and the flight speed and power consumption rate of the drone corresponding to the graph structure are introduced to make the extracted initial features more in line with the actual flight situation. The formula for feature extraction by the graph neural network (GNN) is as follows:

[0054]

[0055] in, is the initial feature of node v after the t+1th iteration; σ is the activation function; W is the weight matrix of GNN; e uv is the eigenvector of edge uv; is the initial feature of node u after the tth iteration; node u is the neighbor node of node v; |N(v)| is the set of neighbor nodes of node v; V is the flight speed of the drone corresponding to the graph structure; P is the power consumption rate of the drone corresponding to the graph structure; b is the bias vector of GNN;

[0056] Furthermore, the initial features extracted by GNN are input into CNN for further processing to obtain the features processed by CNN. By introducing the remaining power and task priority, the consideration of power and task importance in the local feature processing process is enhanced, making the path planning more in line with the endurance and mission urgency of the drone. The formula for CNN feature processing is as follows:

[0057]

[0058] Among them, c v Represents the features of node v after being processed by CNN; is the weight matrix of CNN; h v is the initial feature of node v; is the bias vector of CNN; K(v) is h v Convolutional neighborhood collection in CNN; It is the weight matrix of convolutional neighborhood features in CNN; Is a node Initial features of nodes is the neighbor node of node v; F is the remaining power of the drone corresponding to the graph structure; U is the task priority of the drone corresponding to the graph structure; F max is the maximum remaining power of all drones; U max is the maximum task priority of all tasks;

[0059] Finally, the features processed by CNN are input into FNN for path planning decision. FNN integrates global features and local features, and introduces task complexity and task distance, so that path planning can better balance task complexity, flight efficiency and task distance. The formula for path planning decision by FNN is as follows:

[0060]

[0061] Among them, o is the path planning result of the UAV corresponding to the graph structure, that is, the optimal path; R is the weight matrix of the FNN output layer; is the bias vector of FNN; is the feature c of node v after CNN processing v The intermediate feature vector in the lth layer of FNN; R lis the weight matrix of the lth layer of the FNN; L is the number of layers of the fully connected neural network; C is the mission complexity of the UAV corresponding to the graph structure; D is the mission distance of the UAV corresponding to the graph structure, which is obtained by calculating the sum of the path lengths between the starting position of the UAV, each intermediate waypoint and the target position in the graph structure, and using the Euclidean distance formula or the Manhattan distance formula to calculate the distance of each path segment, and summing them up to get the total distance. The Euclidean distance formula or the Manhattan distance formula is an existing technology and will not be described here.

[0062] When the drone performs tasks according to the optimal path, it is also necessary to combine real-time sensor data to define a dynamic obstacle avoidance function. The real-time sensor data includes the distance of the obstacle, the direction vector of the obstacle, the dynamic weight of the obstacle, and the total number of obstacles. According to the real-time sensor data, the dynamic obstacle avoidance function is used to flexibly adjust the flight path of the drone to improve the safety and stability of the drone during mission execution. The formula of the dynamic obstacle avoidance function is as follows:

[0063]

[0064] in, is the obstacle avoidance vector, which helps the drone avoid obstacles safely by providing a new direction and distance; α is the obstacle avoidance coefficient, which is used to adjust the intensity of obstacle avoidance and determines the sensitivity of the drone's position adjustment; M is the total number of obstacles currently detected; w γ is the dynamic weight of the γth obstacle. The dynamic weight of the obstacle is calculated by combining the real-time sensor data with the preset rules. The obstacle avoidance coefficient and the specific preset rules are set according to the specific implementation scenario; d γ is the distance between the current UAV and the γth obstacle, reflecting the magnitude of the potential collision risk; It is the direction vector from the current drone to the γth obstacle, indicating the direction of obstacle avoidance.

[0065] The UAV performs the mission according to the optimal path. During the mission execution, the UAV detects the surrounding obstacles in real time through sensors. If there are obstacles, the dynamic obstacle avoidance function is used to calculate the obstacle avoidance vector and determine the direction and distance of obstacle avoidance. The UAV adjusts the flight direction and position of the UAV according to the calculated obstacle avoidance vector, flies in the calculated direction, and moves according to the indicated position. After the obstacle avoidance is completed, the node information in the graph structure is updated according to the current position of the UAV, and the optimal path planning algorithm based on the graph neural network is used to generate a new optimal path. The UAV continues to perform the mission according to the new optimal path.

[0066] After successfully completing the mission, the drone needs to return to the designated landing point and upload the collected flight data (flight trajectory, flight speed, flight altitude, power consumption rate) and environmental data to the ground control center or cloud server for analysis by the data science team on the ground control center or cloud server to evaluate the drone's mission completion and drone performance.

[0067] In summary, the intelligent collaborative control method of UAV swarms is completed.

[0068] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0069] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0070] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. The intelligent collaborative control method of drone swarm is characterized by: The following steps are involved: S 1: receiving and preprocessing environmental data, drone status data and mission data to obtain preprocessed data, and performing data fusion processing on the preprocessed data to obtain fused data; A multi-objective optimization genetic algorithm is used to perform UAV swarm task allocation and generate a task allocation plan; S2: Based on the task allocation plan, the optimal path is generated using the optimal path planning algorithm based on the graph neural network; the UAV performs the task according to the optimal path, and the flight path of the UAV is adjusted using the dynamic obstacle avoidance function; the node information in the graph structure is updated according to the current position of the UAV, and a new optimal path is generated using the optimal path planning algorithm based on the graph neural network. The UAV continues to perform the task according to the new optimal path, returns to the landing point after the task is completed, and uploads the collected flight data and environmental data.

2. The method for intelligent collaborative control of drone swarms according to claim 1, characterized in that: The S1 specifically includes: In the implementation process of the multi-objective optimization genetic algorithm, the population is initialized, the initial population is generated, and the initial task allocation plan for each individual in the population is randomly generated; the initial task allocation plan is evaluated and optimized by calculating the fitness value.

3. The method for intelligent collaborative control of drone swarms according to claim 2, characterized in that: The S1 specifically includes: In the process of allocating tasks of drone swarms using multi-objective optimization genetic algorithm, the improved roulette algorithm is used to select individuals with high fitness values ​​to enter the mating pool.

4. The method for intelligent collaborative control of drone swarms according to claim 3, characterized in that: The S1 specifically includes: Based on the selected individuals, single-point crossover operation and random mutation operation are performed to obtain the next generation population; the fitness value of each individual in the next generation population is calculated, and the Pareto front solution is obtained by using fast non-dominated sorting; according to the fitness value, an improved roulette algorithm is used to select individuals with high fitness values ​​in the Pareto front solution to perform single-point crossover operation and random mutation operation to generate a new population; through iterative operation, an optimized Pareto front solution is output, and a task allocation plan is selected from the optimized Pareto front solution.

5. The method for intelligent collaborative control of drone swarms according to claim 1, characterized in that: The S2 specifically includes: The optimal path planning algorithm based on graph neural network combines graph neural network, convolutional neural network and fully connected neural network to improve feature extraction and path planning decisions.

6. The method for intelligent collaborative control of drone swarms according to claim 5, characterized in that: The S2 specifically includes: In the process of calculating the optimal path by the optimal path planning algorithm based on graph neural network, a graph structure is established for each drone based on the task allocation plan, and the graph neural network is used to extract the initial features of the drone graph structure.

7. The method for intelligent collaborative control of drone swarms according to claim 6, characterized in that: The S2 specifically includes: The initial features are input into the convolutional neural network for processing to obtain the features processed by the convolutional neural network; the features processed by the convolutional neural network are input into the fully connected neural network for path planning decision to obtain the optimal path.

8. The method for intelligent collaborative control of drone swarms according to claim 7, characterized in that: The S2 specifically includes: When the UAV performs tasks according to the optimal path, it detects surrounding obstacles in real time. When there are obstacles, it uses the dynamic obstacle avoidance function to calculate the obstacle avoidance vector and determine the direction and distance of obstacle avoidance. The flight direction and position of the UAV are adjusted according to the obstacle avoidance vector, and the UAV flies in the calculated direction and moves according to the indicated position.

9. The method for intelligent collaborative control of drone swarms according to claim 8, characterized in that: The S2 specifically includes: The dynamic obstacle avoidance function formula is as follows: in, is the obstacle avoidance vector; α is the obstacle avoidance coefficient; M is the total number of obstacles currently detected; w γ is the dynamic weight of the γth obstacle; d γ is the distance between the current UAV and the γth obstacle; It is the direction vector from the current drone to the γth obstacle, indicating the direction of obstacle avoidance.

Images