Unmanned aerial vehicle cluster obstacle avoidance reconstruction method and device

Abstract

The invention relates to an unmanned aerial vehicle cluster obstacle avoidance reconstruction method and device. The method comprises the following steps: monitoring obstacles and unmanned aerial vehicles in a cluster in real time, and constructing an environment space model; based on the environment space model, taking the shortest obstacle avoidance path as a target, and constructing a path planning target function; acquiring a flight path according to the path planning objective function; and when the environment space model changes, updating and reconstructing the flight path by using a gradient descent method. By adopting the method, the path planning can be updated in real time when a new obstacle or environment change is detected, and the unmanned aerial vehicle cluster can be ensured to flexibly cope with the change of a complex environment.

Description

Technical Field

[0001] The present application relates to the field of drone technology, and in particular to a method and device for drone cluster obstacle avoidance reconstruction. Background Art

[0002] With the development of drone swarm technology, large-scale drone swarms are being used more and more widely in military, logistics, disaster relief and other fields.

[0003] However, when performing tasks in complex environments, drone swarms still face multiple challenges in obstacle avoidance, such as obstacle perception, communication delay, and path reconstruction. In existing technologies, most obstacle avoidance methods have problems such as slow response speed and path planning limitations, which make it difficult to meet the actual obstacle avoidance needs of large-scale drone swarms. Summary of the invention

[0004] Based on this, it is necessary to provide a drone cluster obstacle avoidance and reconstruction method and device that can improve the obstacle avoidance and path reconstruction capabilities in order to address the above technical problems.

[0005] In a first aspect, the present application provides a method for reconfiguring a drone cluster by avoiding obstacles. The method comprises:

[0006] Monitor obstacles and drones in the cluster in real time and build an environmental space model;

[0007] Based on the environmental space model, the path planning objective function is constructed with the shortest obstacle avoidance path as the goal; the flight path is obtained according to the path planning objective function;

[0008] When the environmental space model changes, the flight path is updated and reconstructed using the gradient descent method.

[0009] In one embodiment, real-time monitoring of obstacles includes:

[0010] Obtain environmental perception data based on sensor data collected in real time by several sensors onboard the drone;

[0011] Multi-scale feature fusion technology is used to fuse the environmental perception data and obtain obstacle monitoring results.

[0012] In one embodiment, real-time monitoring of each drone in the cluster includes:

[0013] Based on the data sharing strategy, data is shared among the drones in the cluster; the shared data includes environmental perception data and the status information of each drone in the cluster;

[0014] Among them, the data sharing strategy is obtained by solving the data allocation objective function. The data allocation objective function is constructed with the goal of minimizing the total communication delay between drones in the cluster and the constraint that the data transmitted between drones in the cluster is not greater than the maximum communication capacity of a single drone.

[0015] In one embodiment, updating and reconstructing the flight path using the gradient descent method includes:

[0016] Based on the path planning objective function and the current environmental space model, the gradient of the path planning objective function is obtained, and the current flight path is updated and reconstructed in real time according to the gradient of the path planning objective function.

[0017] In one embodiment, the method further comprises:

[0018] For each UAV in the cluster, the consistency control law is obtained based on the adjacency matrix, the expected control input, and the state difference with the adjacent UAVs. The elements of the adjacency matrix represent the connection weights with the adjacent UAVs. The adjacency matrix is ​​optimized with the goal of maximizing the convergence speed of collaborative control and the constraint that the communication overhead is not greater than the upper limit of the total communication bandwidth.

[0019] For all UAVs in the cluster, a control law function is constructed with the goal of minimizing the weighted sum of task execution error and control cost. The Hamilton-Jacobi-Bellman equation is used to solve the control law function and obtain the execution control law.

[0020] The coordinated control and execution of UAV clusters are achieved according to the consistency control law and execution control law.

[0021] In one embodiment, the method further comprises:

[0022] According to the location distance between the UAV and the task point and the task allocation matrix, a task cost function is constructed; the task allocation matrix represents the corresponding relationship between the UAV and the task point;

[0023] Based on the task cost function, the task allocation optimization function is constructed with the goal of minimizing the total task cost of each drone in the cluster and the constraint that each task point corresponds to only one drone. The task allocation matrix is ​​optimized according to the task allocation optimization function.

[0024] In a second aspect, the present application also provides a drone cluster obstacle avoidance and reconstruction device. The device includes:

[0025] Environmental perception module, used for real-time monitoring of obstacles;

[0026] Data sharing module, used for real-time monitoring of each drone in the cluster;

[0027] The data processing module is used to build an environmental space model based on the environmental perception module and the data sharing module; based on the environmental space model, a path planning objective function is built with the shortest obstacle avoidance path as the goal; the flight path is obtained according to the path planning objective function; when the environmental space model changes, the flight path is updated and reconstructed using the gradient descent method.

[0028] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps in the above-mentioned drone cluster obstacle avoidance reconstruction method when executing the computer program.

[0029] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps in the above-mentioned drone cluster obstacle avoidance reconstruction method when executed by a processor.

[0030] In a fifth aspect, the present application also provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned drone cluster obstacle avoidance reconstruction method are implemented.

[0031] The above-mentioned drone cluster obstacle avoidance reconstruction method and device proposes a dynamic path planning and reconstruction algorithm based on the gradient descent method, which can update the path planning in real time when new obstacles or environmental changes are detected, ensuring that the drone cluster can flexibly respond to changes in complex environments. Compared with traditional static path planning methods, the present invention significantly improves the flexibility and robustness of path planning, reduces collision risks, and improves the reliability of task completion. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flowchart of a method for reconfiguring a drone cluster obstacle avoidance in one embodiment;

[0033] Figure 2 It is a structural block diagram of a drone cluster obstacle avoidance reconstruction device in one embodiment. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0035] The application of existing communication technologies (such as 4G or Wi-Fi) in drone clusters faces the problems of large communication delays and limited bandwidth, especially when the cluster is large or the environment is complex. This communication bottleneck leads to inefficient data sharing and state synchronization, which affects the real-time decision-making ability of the drone cluster and increases the difficulty of obstacle avoidance and path planning. Therefore, in one embodiment, a drone cluster obstacle avoidance reconstruction method is provided, which utilizes the ultra-high rate and ultra-low latency characteristics of 6G communication technology to achieve rapid data sharing and real-time communication within the drone cluster. Subsequent embodiments can all be based on 6G communication technology.

[0036] In one embodiment, Figure 1 As shown, the drone cluster obstacle avoidance reconstruction method includes the following steps:

[0037] Step 102: Monitor obstacles and each drone in the cluster in real time and build an environmental space model.

[0038] In this embodiment, the environmental space model is a model that describes the spatial layout, obstacles, and distribution of drones and their relationships in the environment where the drone cluster is located. Relying on the sensors and communication equipment carried by the drones, the data collected and transmitted by each drone in real time is summarized to construct the environmental space model for the subsequent formulation of obstacle avoidance strategies and reconstruction of obstacle avoidance routes.

[0039] Step 104, based on the environment space model, with the shortest obstacle avoidance path as the goal, construct a path planning objective function; and obtain the flight path according to the path planning objective function.

[0040] The obstacle avoidance path corresponds to each drone, with the current position of the corresponding drone as the starting point and the corresponding mission point as the target point, and obstacles must be avoided. In order to optimize the obstacle avoidance path, a path planning objective function is constructed, which takes the shortest obstacle avoidance path as the goal. The path planning objective function is solved to obtain the optimized drone flight path. Commonly used algorithms for solving path planning objective functions include ant colony algorithm, genetic algorithm, etc.

[0041] Step 106: When the environment space model changes, the flight path is updated and reconstructed using the gradient descent method.

[0042] When an obstacle suddenly appears in the space near the drone cluster, the environmental space model will change, and the flight path must be dynamically planned in real time, that is, path reconstruction. The path reconstruction process needs to consider multiple factors, including the location of the obstacle, the flight formation within the cluster, and the task priority, to ensure that the drone cluster can avoid the obstacle in the shortest time and continue to perform the task.

[0043] This embodiment adopts a path reconstruction algorithm based on the gradient descent method. By calculating the gradient of the path planning objective function, an update plan for the flight path can be obtained. The above algorithm can quickly respond to changes and update the path planning in real time when new obstacles or environmental changes are detected, ensuring that the drone cluster can flexibly respond to changes in complex environments.

[0044] In one embodiment, the real-time monitoring of obstacles in step 102 includes: obtaining environmental perception data based on sensor data collected in real time by a number of sensors carried on the drone; and performing feature fusion on the environmental perception data using multi-scale feature fusion technology to obtain obstacle monitoring results.

[0045] Traditional environmental perception systems often rely on a single type of sensor (such as lidar or camera), resulting in insufficient perception accuracy and robustness. In complex dynamic environments, the data from a single sensor may be affected by factors such as noise and occlusion, resulting in reduced obstacle detection accuracy and increased risk of collisions among drone swarms.

[0046] In this embodiment, the drone cluster perceives the surrounding environment in real time through the multi-sensor array it carries, including vision, radar, lidar and other sensors, and combines the high-precision positioning data provided by the 6G network to achieve high-precision detection and identification of obstacles.

[0047] Specifically, suppose each drone U i (i=1,2,…,N) is equipped with multiple sensors, including lidar, visual sensor and radar sensor. Each sensor S j (j=1,2,…,M) The environmental data obtained at time t can be expressed as a matrix D i,j (t), where each element d of the matrix i,j,k (t) represents the measurement value of the environment by the sensor at position k.

[0048] Define each drone U i The comprehensive environment perception matrix is ​​E i (t), which is obtained by weighted fusion of the data matrices of all sensors:

[0049]

[0050] Among them, w j Indicates sensor S j The weight of , the weight can be determined by the following optimization model:

[0051]

[0052] Among them, ∥·∥ F represents the F-norm of the matrix, Indicates drone U i The real environment perception matrix at time t (can be obtained through historical data or high-precision measurement methods). This optimization problem can be solved by gradient descent or other optimization algorithms to obtain the optimal sensor weights. By optimizing the sensor data fusion weights, the accuracy of environmental perception is further improved, which can effectively cope with changing environmental conditions and ensure the safe flight of drone clusters.

[0053] Based on the environmental perception data, that is, the above environmental perception matrix E i (t), UAV U i The following obstacle detection algorithm can be used to identify obstacles in the environment. i (t) is thresholded and the threshold τ is defined as:

[0054]

[0055] Among them, B i (t) is the binary environment matrix, which is used to represent the location of obstacles.

[0056] In order to further improve the detection accuracy, an obstacle detection method based on multi-scale feature fusion is adopted. First, the matrix E i (t) Decomposed into feature representations F at multiple scales i,s (t), where s represents the scale factor:

[0057] F i,s (t) = f s (E i (t))

[0058] Here, f s (·) is a function representing scale transformation, which can be Gaussian blur, Laplace pyramid, etc. Then, the weighted sum of features of different scales is performed to obtain the final obstacle detection matrix O i (t):

[0059]

[0060] Among them, α s is the scale weight coefficient, which can also be optimized by a method similar to sensor weight optimization. i (t) is further binarized to obtain the final obstacle detection result:

[0061]

[0062] Here η is the detection threshold, which is set through experiments or experience.

[0063] In one embodiment, the real-time monitoring of each drone in the cluster in step 102 includes: based on a data sharing strategy, sharing data among the drones in the cluster; the shared data includes environmental perception data and status information of each drone in the cluster;

[0064] Among them, the data sharing strategy is obtained by solving the data allocation objective function. The data allocation objective function is constructed with the goal of minimizing the total communication delay between drones in the cluster and the constraint that the data transmitted between drones in the cluster is not greater than the maximum communication capacity of a single drone.

[0065] In one embodiment, by utilizing the high-speed, low-latency characteristics of 6G communication technology, each drone in a drone cluster can share environmental perception data and flight status information in real time, forming a global environmental perception and flight status map, ensuring information synchronization between drones in the cluster, and generating an environmental space model to achieve real-time monitoring of drones and obstacles in the cluster.

[0066] In one embodiment, it is assumed that the drone cluster consists of N drones, each drone U i (i=1,2,…,N) shares data with other drones through 6G communication technology. At each time t, drone U i The environmental perception data E that needs to be perceived i (t) and its own state information S i (t) Transmit to other UAVs. The latency and bandwidth of data communication directly affect the real-time and integrity of shared data.

[0067] Let C ij (t) represents the time t UAV U i UAV j The communication channel capacity when sending data (unit: bps), then the data transmission delay T ij (t) can be expressed as:

[0068]

[0069] Among them, Data Size ij (t) represents the drone U i At time t, the UAV U j The size of the data transferred (unit: bit).

[0070] The high bandwidth and low latency characteristics supported by 6G communication technology can significantly reduce T ij (t) value, so that data can be quickly shared within the cluster. However, in order to maximize communication efficiency, the capacity of the communication channel C ij(t) It needs to be optimized based on the distance between drones, interference factors, and the allocation strategy of the 6G network.

[0071] Channel capacity C ij (t) The distance from the drone d ij (t) is related and can usually be expressed as:

[0072]

[0073] Where, B is the channel bandwidth (unit: Hz); P i For UAV i Transmitting power (unit: W); G i and G j UAV i and U j The antenna gain; N0 is the background noise power spectral density (unit: W / Hz); I ij (t) is the interference power in the channel (unit: W); α is the path loss exponent, usually between 2 and 4.

[0074] According to the above formula, it can be seen that by adjusting the transmission power P i , optimize antenna gain G i and G j , and minimize interference I ij (t), can improve the channel capacity C ij (t), thereby reducing the communication delay T ij (t).

[0075] In one embodiment, in order to improve the efficiency of data sharing, a data sharing strategy based on distributed optimal allocation is designed.

[0076] The total amount of shared data D(t) is defined as:

[0077]

[0078] The goal is to minimize the total communication delay between all drones in the cluster:

[0079]

[0080] The constraints are:

[0081]

[0082] Among them, D max (t) represents the maximum communication capacity of a single UAV at time t.

[0083] This optimization problem can be solved by the Lagrange multiplier method to obtain the optimal data sharing strategy:

[0084]

[0085] Among them, λ is the Lagrange multiplier, which is obtained by The partial derivative of is solved to obtain the optimal data sharing allocation plan. Finally, each drone transmits data according to the plan to ensure the highest data communication efficiency of the entire drone cluster.

[0086] In one embodiment, updating and reconstructing the flight path using the gradient descent method in step 106 includes: obtaining the gradient of the path planning objective function based on the path planning objective function and the current environmental space model, and updating and reconstructing the current flight path in real time according to the gradient of the path planning objective function.

[0087] Most existing path planning methods are based on static environment assumptions and lack adaptability to dynamic environments. Once the environment changes (such as the appearance of new obstacles or changes in the target position), it is difficult for the drone cluster to adjust the path in time, which may cause the path planning to fail or reduce the execution efficiency. In addition, although some dynamic planning algorithms can cope with environmental changes, their computational complexity is high and it is difficult to apply them in real time in large-scale clusters. In response to the above problems, this embodiment proposes a path reconstruction solution based on the gradient descent method.

[0088] After detecting an obstacle, the drones in the cluster dynamically reconfigure their paths in real time based on shared environmental information, through distributed computing and the low-latency transmission capabilities of the 6G network. The path reconstruction process takes into account multiple factors, including the location of the obstacle, the flight formation within the cluster, and the mission priority, to ensure that the drone cluster can avoid obstacles in the shortest possible time and continue to perform its mission.

[0089] Step 106 is performed based on the path planning objective function defined in step 104. Therefore, the path planning objective function is first described here.

[0090] At time t, suppose the drone U i Need to start from Move to target point The goal of path planning is to find an optimal path P i (t) = {p i (t0),p i (t1),…,p i (t f )}, so that the path length is the shortest and avoids obstacles. Considering the dynamic environment, the path P i (t) needs to be updated in real time to cope with new obstacles or path changes of other drones that may appear at any time. The path planning objective function Ji(t) is defined as:

[0091]

[0092] in, represents the velocity component of the path, that is, the differential of the path length, C i (p i (t)) represents the cost function of the obstacle on the path, which is usually inversely proportional to the distance of the obstacle; λ is the weight factor used to balance the path length and obstacle avoidance cost.

[0093] Based on the above path planning objective function J i (t), when the UAV detects a new obstacle or receives path change information from other UAVs, that is, when the environment space model changes, it immediately adjusts the current path P i (t)Reconstruction.

[0094] First, the path planning objective function J i (t) is expanded by first-order Taylor to obtain an approximate representation of the objective function:

[0095]

[0096] in, Denotes the objective function J i (t) About path P i The gradient of (t), Δp i (t) represents a small adjustment of the path. The optimal update direction of the path can be obtained. In order to solve the gradient We first i Two components of (t)—the path length term and obstacle avoidance cost C i (p i (t)), respectively, and obtain the corresponding gradient.

[0097] The gradient of the path length term is:

[0098]

[0099] The gradient of the obstacle avoidance cost term is:

[0100]

[0101] Among them, q(t) is the position of the obstacle, and the obstacle avoidance cost function C i (p i (t)) is expressed as

[0102] Combining the gradients of the above two parts, we can get the objective function J i The total gradient of (t) is:

[0103]

[0104] At each time t, UAV U i The path will be updated according to the gradient:

[0105]

[0106] Among them, γ is the learning rate, which controls the step size of the path update.

[0107] Since multiple drones in a drone cluster may need to plan paths at the same time, it is necessary to consider the path coordination problem between drones. In order to avoid path conflicts, a path coordination optimization method based on a consensus algorithm is proposed in one embodiment. The path of each drone is defined as P i (t), then the path set of the drone cluster is P(t) = {P1(t), P2(t), …, P N (t)}. The goal of path collaborative optimization is to minimize the objective function J of each UAV i (t) and the cost function C of the path conflict within the cluster conflict (t):

[0108]

[0109] Among them, the path conflict cost function C conflict (t) can be defined as:

[0110]

[0111] By optimizing the above path planning objective function, both path conflicts and total path costs can be reduced.

[0112] Most existing path planning methods are based on static environment assumptions and lack adaptability to dynamic environments. Once the environment changes (such as the appearance of new obstacles or changes in the target position), it is difficult for the drone cluster to adjust the path in time, which may cause the path planning to fail or reduce the execution efficiency. In addition, although some dynamic planning algorithms can cope with environmental changes, their computational complexity is high and it is difficult to apply them in real time in large-scale clusters. The dynamic path planning and reconstruction algorithm based on the gradient descent method proposed in this embodiment can update the path planning in real time when new obstacles or environmental changes are detected, ensuring that the drone cluster can flexibly respond to changes in complex environments.

[0113] In one embodiment, the method further includes: for each drone in the cluster, obtaining a consistency control law based on the adjacency matrix, the expected control input, and the state difference with the adjacent drones; wherein the elements of the adjacency matrix represent the connection weights with the adjacent drones, and the adjacency matrix is ​​optimized with the goal of maximizing the convergence speed of collaborative control and with the constraint that the communication overhead is not greater than the upper limit of the total communication bandwidth; for all drones in the cluster, constructing a control law function with the goal of minimizing the weighted sum of task execution error and control cost, solving the control law function using the Hamilton-Jacobi-Bellman equation, and obtaining the execution control law; and realizing the collaborative control and execution of the drone cluster according to the consistency control law and the execution control law.

[0114] In a large-scale UAV swarm, assuming that each UAV i The state vector is x i (t), the control input is u i (t). The dynamic model of the UAV can be expressed as:

[0115]

[0116] In order to achieve collaborative control, all drones in the cluster need to use a consistency algorithm to ensure that their states and control inputs are consistent. The consistency error is defined as:

[0117] e ij (t) = x i (t)-x j (t)

[0118] The goal of cooperative control is to minimize the consistency error so that:

[0119]

[0120] To achieve this goal, the following consistency control law is adopted:

[0121]

[0122] in, Indicates that the drone U i A collection of directly connected neighboring drones; a ij is an element of the adjacency matrix, representing the number of drones U i and U j The connection weight between them; It's a drone i The expected control input in task execution.

[0123] In order to improve the efficiency of collaborative control, the adjacency matrix A = [a ijThe choice of ] needs to be optimized. The optimization goal is to maximize the convergence speed of cooperative control while ensuring the minimum communication overhead. Define the graph Laplace matrix L as:

[0124] L=DA

[0125] Where D is the degree matrix, which is a diagonal matrix with diagonal elements

[0126] The convergence speed of cooperative control is related to the second smallest eigenvalue λ2(L) of the graph Laplacian matrix L, which is called the algebraic connectivity of the graph. In order to improve the convergence speed, the optimization goal is to maximize λ2(L):

[0127]

[0128] At the same time, considering the communication overhead, the following constraints are added:

[0129]

[0130] Among them, B max is the upper limit of the total communication bandwidth. This optimization problem can be solved by semidefinite programming (SDP) to obtain the optimal adjacency matrix A * , thereby achieving efficient collaborative control.

[0131] Under the premise of ensuring state consistency, in order to realize the task execution of the drone cluster, this embodiment designs the execution control law based on the optimal control theory. The task goal of the drone cluster can be defined as a global performance indicator function J global This function is usually a weighted sum of the task error and the control cost:

[0132]

[0133] in, Indicates drone U i The target state of ; ρ is the weight factor used to balance the state error and control cost.

[0134] In order to minimize J global , the optimal control law can be obtained by solving the Hamilton-Jacobi-Bellman (HJB) equation

[0135]

[0136] Among them, V i (x i (t)) is the value function. By iteratively solving the HJB equation, the analytical solution or numerical solution of the optimal control law can be obtained.

[0137] In one embodiment, the method also includes: constructing a task cost function based on the position distance between the drone and the task point and the task allocation matrix; the task allocation matrix represents the corresponding relationship between the drone and the task point; based on the task cost function, with the goal of minimizing the total task cost of each drone in the cluster and with the constraint that each task point corresponds to only one drone, constructing a task allocation optimization function; optimizing the task allocation matrix according to the task allocation optimization function.

[0138] In the actual execution process of drone clusters, task allocation is also an important factor affecting the overall performance. The task cost function of each drone is defined as:

[0139]

[0140] Among them, r k (t) represents the position of task point k; φ ik is the element of the task allocation matrix, representing the UAV U i Whether responsible for task point k.

[0141] To optimize task allocation, the goal is to minimize the total task cost of all drones in the cluster:

[0142]

[0143] The constraint is that each mission point can only be handled by one drone:

[0144]

[0145] This optimization problem can be solved by the Hungarian algorithm or other task allocation algorithms to obtain the optimal task allocation matrix Φ * , thereby ensuring that the efficiency of the drone cluster is maximized during the execution process.

[0146] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0147] 1. Significantly improved real-time performance: By utilizing the ultra-high speed and ultra-low latency characteristics of 6G communication technology, the present invention realizes rapid data sharing and real-time communication within the drone cluster. Compared with traditional communication technology, the present invention can significantly reduce data transmission delays, ensuring that in complex dynamic environments, each drone in the cluster can share environmental information and path planning data in real time, thereby greatly improving the real-time performance of obstacle avoidance decisions and path reconstruction.

[0148] 2. High-precision environmental perception through multi-sensor data fusion: This invention achieves accurate detection and identification of obstacles in complex environments by integrating data from multiple sensors such as vision, radar, and lidar, combined with the high-precision positioning capability of 6G communication technology. By optimizing the sensor data fusion weight, the accuracy of environmental perception is further improved, which can effectively respond to changing environmental conditions and ensure the safe flight of drone clusters.

[0149] 3. Flexibility and robustness of dynamic path planning: This invention proposes a dynamic path planning and reconstruction algorithm based on the gradient descent method, which can update the path planning in real time when new obstacles or environmental changes are detected, ensuring that the drone cluster can flexibly respond to changes in complex environments. Compared with traditional static path planning methods, this invention significantly improves the flexibility and robustness of path planning, reduces the risk of collision, and improves the reliability of task completion.

[0150] 4. Efficiency of collaborative control: Through the collaborative control strategy combining the consistency algorithm and the optimization control, the present invention ensures that the drone cluster can collaborate efficiently during the execution process, avoid path conflicts, and achieve optimal allocation and execution of tasks. Compared with the traditional centralized control method, the distributed collaborative control strategy of the present invention greatly improves the coordination of the cluster and the overall task execution efficiency, and is particularly suitable for complex task scenarios of large-scale drone clusters.

[0151] 5 Global optimization of path planning and task execution: The present invention adopts Laplace matrix optimization and Hamilton-Jacobi-Bellman equation in graph theory to ensure the global optimality of path planning and task execution of drone clusters. By optimizing the communication network structure and control law design, the present invention can achieve global optimization of path planning and task execution in large-scale drone clusters, significantly improving the task completion speed and accuracy of the cluster.

[0152] 6. Wide range of applicable scenarios: The method of the present invention has wide applicability and can provide efficient and stable drone cluster control and execution capabilities in a variety of complex mission scenarios. It is particularly suitable for scenarios that require rapid response and efficient collaboration, such as military operations, emergency rescue, logistics distribution, etc. The strategy of the present invention provides solid technical support for the application of drone clusters in a changing environment, and has important practical value and broad market prospects.

[0153] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0154] Based on the same inventive concept, the embodiment of the present application also provides a drone cluster obstacle avoidance reconstruction device for implementing the drone cluster obstacle avoidance reconstruction method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more drone cluster obstacle avoidance reconstruction device embodiments provided below can refer to the limitations of the drone cluster obstacle avoidance reconstruction method above, and will not be repeated here.

[0155] In one embodiment, Figure 2 As shown, a drone cluster obstacle avoidance and reconstruction device is provided, including: an environment perception module 202, a data sharing module 204 and a data processing module 206, wherein:

[0156] The environment perception module 202 is used to monitor obstacles in real time;

[0157] The data sharing module 204 is used to monitor each drone in the cluster in real time;

[0158] The data processing module 206 is used to construct an environmental space model based on the environmental perception module and the data sharing module; based on the environmental space model, a path planning objective function is constructed with the shortest obstacle avoidance path as the goal; the flight path is obtained according to the path planning objective function; when the environmental space model changes, the flight path is updated and reconstructed using the gradient descent method.

[0159] Each module in the above-mentioned drone cluster obstacle avoidance reconstruction device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0160] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in all the above method embodiments when executing the computer program.

[0161] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in all the above method embodiments are implemented.

[0162] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in all the above method embodiments when executed by a processor.

[0163] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0164] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited thereto. The processors involved in each embodiment provided in this application may be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, data processing logic devices based on quantum computing, etc., but are not limited thereto.

[0165] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0166] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A drone cluster obstacle avoidance reconstruction method, characterized in that: The method comprises: Monitor obstacles and drones in the cluster in real time and build an environmental space model; Based on the environmental space model, a path planning objective function is constructed with the shortest obstacle avoidance path as the goal; and a flight path is obtained according to the path planning objective function; When the environmental space model changes, the flight path is updated and reconstructed using the gradient descent method.

2. The method according to claim 1, characterized in that: Real-time monitoring of obstacles includes: Obtain environmental perception data based on sensor data collected in real time by several sensors onboard the drone; The multi-scale feature fusion technology is used to perform feature fusion on the environmental perception data to obtain obstacle monitoring results.

3. The method according to claim 2, characterized in that Real-time monitoring of each drone in the cluster includes: Based on the data sharing strategy, data is shared among the drones in the cluster; the shared data includes the environmental perception data and the status information of each drone in the cluster; Among them, the data sharing strategy is obtained by solving the data allocation objective function, and the data allocation objective function is constructed with the goal of minimizing the total communication delay between the drones in the cluster and the constraint that the data transmitted between the drones in the cluster is not greater than the maximum communication capacity of a single drone.

4. The method according to claim 1, characterized in that: The updating and reconstruction of the flight path by using the gradient descent method comprises: Based on the path planning objective function and the current environmental space model, the gradient of the path planning objective function is obtained, and the current flight path is updated and reconstructed in real time according to the gradient of the path planning objective function.

5. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: For each UAV in the cluster, a consistent control law is obtained based on the adjacency matrix, the expected control input, and the state difference with the adjacent UAVs; wherein the elements of the adjacency matrix represent the connection weights with the adjacent UAVs, and the adjacency matrix is ​​optimized with the goal of maximizing the convergence speed of collaborative control and the constraint that the communication overhead is not greater than the upper limit of the total communication bandwidth; For all UAVs in the cluster, a control law function is constructed with the goal of minimizing the weighted sum of task execution error and control cost, and the control law function is solved using the Hamilton-Jacobi-Bellman equation to obtain the execution control law; The coordinated control and execution of the drone cluster are achieved according to the consistency control law and the execution control law.

6. The method according to claim 5, characterized in that The method further comprises: Constructing a task cost function according to the position distance between the UAV and the task point and the task allocation matrix; the task allocation matrix represents the corresponding relationship between the UAV and the task point; Based on the task cost function, a task allocation optimization function is constructed with the goal of minimizing the total task cost of each drone in the cluster and the constraint that each task point corresponds to only one drone; and the task allocation matrix is ​​optimized according to the task allocation optimization function.

7. A drone cluster obstacle avoidance and reconstruction device, characterized in that: The device comprises: Environmental perception module, used for real-time monitoring of obstacles; Data sharing module, used for real-time monitoring of each drone in the cluster; The data processing module is used to construct an environmental space model based on the environmental perception module and the data sharing module; based on the environmental space model, a path planning objective function is constructed with the shortest obstacle avoidance path as the goal; a flight path is obtained according to the path planning objective function; when the environmental space model changes, the flight path is updated and reconstructed using the gradient descent method.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images