Unmanned aerial vehicle cluster autonomous crossing method and system

Abstract

The invention relates to an unmanned aerial vehicle cluster autonomous crossing method and system. The method comprises the following steps: expanding channel capacity, carrying out time delay analysis, and introducing a non-orthogonal multiple access technology to carry out interference management and resource allocation so as to complete network deployment; establishing a linear state space model of a multi-unmanned aerial vehicle system for the unmanned aerial vehicle cluster, setting a cooperative control strategy and an obstacle avoidance strategy, and performing model performance analysis; acquiring environment information through each sensor arranged on each unmanned aerial vehicle in the unmanned aerial vehicle cluster to obtain an environment state; an artificial potential field method is adopted to complete a dynamic obstacle avoidance algorithm according to environment state design, and the path of the unmanned aerial vehicle is adjusted in real time according to the dynamic obstacle avoidance algorithm; establishing an unmanned aerial vehicle kinematic model, designing an optimization algorithm of path planning based on the unmanned aerial vehicle kinematic model, and performing real-time dynamic path adjustment according to the optimization algorithm; setting a security and fault-tolerant mechanism; and autonomous crossing of the unmanned aerial vehicle cluster is realized. And the high-efficiency collaboration and stability of the unmanned aerial vehicle cluster are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of UAV swarm control, and particularly to a method and system for autonomous crossing of UAV swarms. Background Art

[0002] Existing UAV swarm technologies are widely used in fields such as military, logistics, and monitoring. However, with the increasing application requirements, the expansion of swarm scale, and the complexity of the operating environment, traditional communication and control technologies face great challenges. Especially when performing efficient and reliable autonomous crossing of UAV formations in complex environments, existing communication technologies are difficult to meet the collaborative requirements of large-scale swarms due to problems such as bandwidth limitations, high latency, and unstable connections. Currently, some UAV swarms use 4G / 5G networks for communication. Although the data transmission rate has been improved to a certain extent, in the face of large-scale UAV swarms, hundreds or even thousands of UAV data communications need to be processed simultaneously, and the existing bandwidth capacity is difficult to meet. The transmission requirements of a large amount of data such as high-resolution sensor data, real-time video streams, and control instructions exceed the bearing capacity of the existing network, resulting in communication bottlenecks and affecting the collaborative efficiency of the swarm. Moreover, the autonomous crossing of UAV swarms has extremely high requirements for real-time performance, and communication latency will directly affect the collaborative reaction speed of UAVs. The latency of existing communication networks is usually in the millisecond level or even higher, which cannot meet the real-time obstacle avoidance and path planning requirements of UAVs in high-speed movement and complex environments, and may lead to collisions and mission failures. Furthermore, UAV swarms are vulnerable to security threats such as network attacks and signal interference, and existing technologies lack effective protection measures. In addition, for the failure of a single UAV, there is a lack of a fast detection and fault tolerance mechanism, which may lead to the collaborative failure of the entire swarm.

[0003] Therefore, when traditional UAV swarms are working, the collaboration and stability of UAV swarms are poor. Summary of the Invention

[0004] Based on this, in order to solve the above technical problems, a method and system for autonomous crossing of UAV swarms are provided, which can improve the efficient collaboration and stability of UAV swarms.

[0005] A method for autonomous crossing of UAV swarms, the method includes:

[0006] Obtain the network capacity requirements corresponding to the UAV swarm, and expand the channel capacity based on the network capacity requirements; perform latency analysis on each UAV in the UAV swarm, and introduce non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements of the UAV swarm to complete network deployment;

[0007] Based on the network deployment, a linear state - space model of the multi - UAV system is established for the UAV cluster. Under the linear state - space model, a cooperative control strategy, an obstacle - avoidance strategy are set and the model performance is analyzed;

[0008] Under the linear state - space model, environmental information is collected by each sensor set on each UAV in the UAV cluster to obtain the environmental state; the artificial potential field method is used to design a dynamic obstacle - avoidance algorithm according to the environmental state, and the UAV path is adjusted in real time according to the dynamic obstacle - avoidance algorithm;

[0009] The autonomous crossing task of the UAV cluster is determined. Based on the autonomous crossing task, a kinematic model of the UAV is established according to the adjusted UAV path. An optimization algorithm for path planning is designed based on the kinematic model of the UAV, and the path is adjusted dynamically in real time according to the optimization algorithm;

[0010] A safety and fault - tolerance mechanism is set; if any UAV in the UAV cluster has a fault, the control law of the UAV cluster is adjusted based on the safety and fault - tolerance mechanism, and the non - fault UAVs are controlled to operate through the adjusted control law of the UAV cluster to achieve the autonomous crossing of the UAV cluster.

[0011] In one embodiment, the network capacity requirement corresponding to the UAV cluster is obtained, and the channel capacity is expanded based on the network capacity requirement; the delay analysis of each UAV in the UAV cluster is performed, and the non - orthogonal multiple access technology is introduced according to the connection requirement of the UAV cluster for interference management and resource allocation to complete the network deployment, including:

[0012] The network capacity requirement corresponding to the UAV cluster is obtained, and according to the network capacity requirement, a bandwidth allocation strategy is calculated based on the channel capacity, and the channel capacity is expanded according to the bandwidth allocation strategy;

[0013] The end - to - end delay, signal propagation delay, signal processing delay, transmission delay, and queuing delay of each UAV in the UAV cluster are determined, and the delay analysis is performed;

[0014] The connection requirement of the UAV cluster is determined, and the non - orthogonal multiple access technology is introduced according to the connection requirement for interference management and resource allocation to complete the network deployment.

[0015] In one embodiment, a linear state - space model of the multi - UAV system is established for the UAV cluster. Under the linear state - space model, a cooperative control strategy, an obstacle - avoidance strategy are set and the model performance is analyzed, including:

[0016] The dynamic models of each UAV in the UAV cluster are determined, and the dynamic models are linearized to obtain a linear state - space model of the multi - UAV system;

[0017] Each of the drones obtains the status information of itself and its neighboring drones and conducts information interaction;

[0018] Based on the linear state space model and the status information of each drone, state estimation is performed on each drone;

[0019] Set cooperative control strategies and obstacle avoidance strategies according to the status information of each drone;

[0020] Based on the cooperative control strategies and obstacle avoidance strategies, perform performance analysis on the linear state space model of the multi-drone system to obtain the analysis results.

[0021] In one embodiment, environmental information is collected by each sensor set on each drone in the drone cluster to obtain the environmental status; the artificial potential field method is used to design a dynamic obstacle avoidance algorithm according to the environmental status, and the drone path is adjusted in real time according to the dynamic obstacle avoidance algorithm, including:

[0022] Collect environmental information by each sensor set on each drone in the drone cluster, perform data fusion on each piece of environmental information using a filter, and estimate the environmental status based on the fused data;

[0023] Use the artificial potential field method to determine the gravitational potential field and repulsive potential field according to the environmental status, perform gradient calculation, and design a dynamic obstacle avoidance algorithm;

[0024] Based on the dynamic obstacle avoidance algorithm and the environmental status, adjust the drone path in real time.

[0025] In one embodiment, use the artificial potential field method to design a dynamic obstacle avoidance algorithm according to the environmental status, including:

[0026] Determine dynamic obstacles based on the environmental status, obtain the obstacle data corresponding to the dynamic obstacles, and predict the obstacle movement data according to the obstacle data;

[0027] Set collision avoidance constraints according to the obstacle movement data;

[0028] Use the artificial potential field method to design a dynamic obstacle avoidance algorithm according to the environmental status and collision avoidance constraints.

[0029] In one embodiment, establish a drone kinematic model according to the adjusted drone path, including:

[0030] Determine obstacle information according to the adjusted drone path, and calculate the obstacle avoidance penalty function according to the obstacle information;

[0031] Based on the obstacle avoidance penalty function, define the cost function of each unmanned aerial vehicle (UAV) in the UAV cluster and establish constraints;

[0032] According to the obstacle avoidance penalty function and constraints, and in combination with the flight parameters of each UAV in the UAV cluster, establish a UAV kinematic model.

[0033] In one embodiment, design an optimization algorithm for path planning based on the UAV kinematic model, and perform real-time dynamic path adjustment according to the optimization algorithm, including:

[0034] Based on the UAV kinematic model, use distributed model predictive control to optimize the flight paths of each UAV, use a consensus mechanism to perform distributed optimization on the coordination between each UAV, and use an adaptive rolling horizon control method to adjust the path in real time dynamically.

[0035] In one embodiment, the method further includes:

[0036] Design a reward function for the UAV cluster;

[0037] Update the reward strategy of each UAV in the UAV cluster based on the reward function.

[0038] In one embodiment, if any UAV in the UAV cluster has a fault, then based on the safety and fault tolerance mechanism, adjust the UAV cluster control law, and control the non-fault UAVs to operate through the adjusted UAV cluster control law, including:

[0039] Based on the safety and fault tolerance mechanism, introduce a fault detection filter in the UAV cluster and use the fault detection filter to detect the fault UAV;

[0040] Use a fault estimator to perform error estimation on the fault UAV, adjust the UAV cluster control law according to the estimated error, and perform fault compensation using the adjusted UAV cluster control law.

[0041] A UAV cluster autonomous crossing system, the system includes:

[0042] A network deployment module, configured to obtain the network capacity requirements corresponding to the UAV cluster, expand the channel capacity based on the network capacity requirements; perform delay analysis on each UAV in the UAV cluster, and introduce non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements of the UAV cluster to complete network deployment;

[0043] The model establishment module is used to establish a linear state - space model of the multi - UAV system for the UAV cluster based on the network deployment. Under the linear state - space model, a cooperative control strategy, an obstacle - avoidance strategy are set and the model performance is analyzed;

[0044] The path adjustment module is used to, under the linear state - space model, collect environmental information through various sensors set on each UAV in the UAV cluster to obtain the environmental state; design and complete a dynamic obstacle - avoidance algorithm according to the environmental state by using the artificial potential field method, and adjust the UAV path in real time according to the dynamic obstacle - avoidance algorithm;

[0045] The optimization module is used to determine the autonomous crossing task of the UAV cluster, establish a UAV kinematic model based on the adjusted UAV path according to the autonomous crossing task, design an optimization algorithm for path planning based on the UAV kinematic model, and perform real - time dynamic path adjustment according to the optimization algorithm;

[0046] The safety and fault - tolerance module is used to set up a safety and fault - tolerance mechanism; if any UAV in the UAV cluster has a fault, then based on the safety and fault - tolerance mechanism, adjust the UAV cluster control law, and control the non - faulty UAVs to operate through the adjusted UAV cluster control law to achieve the autonomous crossing of the UAV cluster.

[0047] The above - mentioned UAV cluster autonomous crossing method and system, by expanding the channel capacity, conducting delay analysis, interference management and resource allocation, complete the network deployment, and can realize real - time data transmission and information sharing between UAVs and with the ground control center; by constructing a linear state - space model and setting a cooperative control strategy and an obstacle - avoidance strategy, it allows UAVs to realize autonomous path planning and obstacle - avoidance operations on the basis of obtaining their own and neighboring UAV state information, ensuring the high - efficiency cooperation and stability of the UAV cluster; using various sensors to collect environmental information, each UAV can sense the surrounding obstacles and environmental changes in real time, dynamically adjust the flight path, and avoid collisions and dangers; because a dynamic obstacle - avoidance algorithm is designed, a UAV kinematic model is established based on the UAV path according to the autonomous crossing task and the path is optimized, ensuring that in a complex environment, the UAV cluster can quickly adjust the flight formation according to the actual situation and achieve the optimal execution of the task; by setting up a safety and fault - tolerance mechanism, it can ensure the normal operation of the remaining cluster and the continuation of the task when individual UAVs fail or communication is interrupted; improving the high - efficiency cooperation and stability of the UAV cluster. Brief Description of the Drawings

[0048] Figure 1 It is an application environment diagram of the UAV cluster autonomous crossing method in an embodiment;

[0049] Figure 2Schematic diagram of the process of the method for autonomous crossing of an unmanned aerial vehicle (UAV) cluster in an embodiment;

[0050] Figure 3 Block diagram of the structure of the UAV cluster autonomous crossing system in an embodiment;

[0051] Figure 4 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0052] In order to make the objectives, technical solutions, and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to 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.

[0053] The method for autonomous crossing of a UAV cluster provided by the embodiments of the present application can be applied to an application environment as Figure 1 shown. As Figure 1 shown, the application environment includes a computer device 110. The computer device 110 can obtain the network capacity requirements corresponding to the UAV cluster, expand the channel capacity based on the network capacity requirements; perform delay analysis on each UAV in the UAV cluster, and introduce non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements of the UAV cluster to complete network deployment; the computer device 110 can establish a linear state space model of a multi-UAV system for the UAV cluster based on the network deployment, and under the linear state space model, set cooperative control strategies, obstacle avoidance strategies, and perform model performance analysis; the computer device 110 can, under the linear state space model, collect environmental information through various sensors set on each UAV in the UAV cluster to obtain the environmental state; design and complete a dynamic obstacle avoidance algorithm according to the environmental state using the artificial potential field method, and adjust the UAV path in real time according to the dynamic obstacle avoidance algorithm; the computer device 110 can determine the autonomous crossing task of the UAV cluster, establish a UAV kinematic model based on the adjusted UAV path according to the autonomous crossing task, design an optimization algorithm for path planning based on the UAV kinematic model, and perform real-time dynamic path adjustment according to the optimization algorithm; the computer device 110 can set up safety and fault tolerance mechanisms; if any UAV in the UAV cluster has a fault, the control law of the UAV cluster is adjusted based on the safety and fault tolerance mechanisms, and the non-fault UAVs are controlled to operate through the adjusted control law of the UAV cluster to achieve autonomous crossing of the UAV cluster. Among them, the computer device 110 can be but is not limited to various personal computers, laptop computers, smart phones, robots, unmanned aerial vehicles, tablet computers, and other devices.

[0054] In one embodiment, as Figure 2 shown, a method for autonomous crossing of a UAV cluster is provided, including the following steps:

[0055] Step 202, obtain the network capacity requirements corresponding to the UAV cluster, and expand the channel capacity based on the network capacity requirements; perform latency analysis on each UAV in the UAV cluster, and introduce non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements of the UAV cluster to complete network deployment.

[0056] To achieve the autonomous crossing of the UAV cluster, it is first necessary to deploy the communication network in a cluster. Specifically, in this embodiment, deploying the communication network in a cluster requires implementing network capacity and bandwidth allocation, ultra-low latency communication and latency analysis, and large-scale connection and interference management. Among them, when the computer device deploys the communication network in a cluster, it can utilize the ultra-low latency, ultra-high bandwidth, and large connection characteristics of 6G technology to enhance the real-time perception and cooperation ability of the UAV cluster, so as to achieve efficient formation crossing in complex environments.

[0057] In one embodiment, a method for autonomous crossing of a UAV cluster may further include the process of network deployment. The specific process includes: obtaining the network capacity requirements corresponding to the UAV cluster, and based on the network capacity requirements, calculating the bandwidth allocation strategy based on the channel capacity, and expanding the channel capacity according to the bandwidth allocation strategy; determining the end-to-end latency, signal propagation latency, signal processing latency, transmission latency, and queuing latency of each UAV in the UAV cluster, and performing latency analysis; determining the connection requirements of the UAV cluster, and introducing non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements to complete network deployment.

[0058] The computer device can obtain the network capacity requirements corresponding to the UAV cluster. Among them, the communication requirements in the UAV cluster include control signals, cooperation information, sensing data, etc. In this embodiment, let the number of UAVs in the cluster be N, then the total network capacity requirement C total can be expressed as: C total = N × R uav ; where R uav represents the average data rate requirement of a single UAV, with the unit of bps. According to Shannon's theorem, the capacity C of a single channel is: C = B × log2(1 + γ); where B represents the channel bandwidth, with the unit of Hz; γ represents the signal-to-noise ratio. In a 6G network, to meet the high data rate requirements, it is necessary to make full use of the large bandwidth characteristics of the terahertz (THz) band. Then, to meet the total capacity requirements, it is necessary to reasonably allocate the bandwidth. The computer device can calculate the bandwidth allocation strategy based on the channel capacity according to the network capacity requirements. In this embodiment, let the total bandwidth provided by the 6G network be B total , then it is necessary to satisfy: C total ≤ B total × log2(1 + γ avg ); where γ avgDenote the average signal-to-noise ratio. Through transformation, the required average signal-to-noise ratio can be obtained as follows:

[0059] In this embodiment, to improve the signal-to-noise ratio, key 6G technologies such as massive MIMO technology and adaptive beamforming can be adopted. Specifically, in a 6G network, introducing massive MIMO (Massive Multiple Input Multiple Output) technology can significantly improve the spectral efficiency. The channel capacity is extended as: C = M × B × log2(1 + γ); by increasing the number of antennas M, without increasing the bandwidth B, the channel capacity C can be improved to meet the high data rate requirements of the UAV swarm.

[0060] Since the real-time cooperative control of the UAV swarm has strict requirements on communication latency, the end-to-end latency T total consists of: T total = T prop + T proc + T trans + T queue ; where T prop represents the signal propagation latency; T proc represents the signal processing latency; T trans represents the transmission latency; T queue represents the queuing latency. Among them, the signal propagation latency is determined by the physical distance: where d represents the distance between the UAV and the base station or other UAVs, in meters; c represents the speed of light, approximately 3×10 8 m / s; in air communication, T prop is relatively small and can be approximately considered a constant.

[0061] The signal processing latency T proc is related to the hardware performance and algorithm complexity and can be reduced by optimizing the algorithm and upgrading the hardware. Among them, the transmission latency T trans is related to the packet length L and the transmission rate R: where L represents the packet length, in bits; R represents the transmission rate, in bps; by increasing the transmission rate R, T trans can be effectively reduced.

[0062] The queuing latency T queue can be analyzed using the M / M / 1 queuing model, and the average queuing latency is: where λ represents the average arrival rate, in packets per second; μ represents the average service rate, in packets per second; represents the system utilization; to reduce T queue , it is necessary to ensure that ρ is as low as possible, that is, the service rate μ is greater than the arrival rate λ.

[0063] In this embodiment, ultra-low latency communication can be achieved in the following ways: High bandwidth and high rate: Utilize the large bandwidth characteristics of 6G to increase the transmission rate R and reduce T trans Optimize network architecture: Use edge computing and fog computing to delegate some computing tasks to drones or edge nodes to reduce T proc ; Efficient queuing mechanism: introduce priority scheduling and resource reservation to reduce T queue .

[0064] The computer device can determine the connection requirements of the drone cluster. In this embodiment, the 6G network needs to support the access of a large number of devices. Assume that the drone density is δ, the unit is frame / square kilometer, and the coverage area is S, then the total number of connections is: N = δ × S. In order to support large-scale connections, 6G introduces non-orthogonal multiple access (NOMA) technology. NOMA allows multiple users to share the same spectrum resources, separated by power domain or code domain. In power domain NOMA, the received signal is: Where y represents the received signal; P i represents the transmission power of the i-th user; h i represents the channel gain; x i represents the transmitted signal of the i-th user; n represents the noise. Through the successive interference cancellation (SIC) technology, the receiving end can decode the user signals one by one.

[0065] Next, the computer device can perform interference management and resource allocation. First, the computer device can calculate the signal to interference noise ratio SINR. For the i-th drone, its SINR is: Among them, |h i | 2 represents the square of the channel gain; N0 represents the noise power spectral density; P j represents the transmission power of the jth user; h j Represents the channel gain. When optimizing resource allocation, the goal is to maximize the total system capacity. The optimization problem can be expressed as: Constraints include: Transmit power limit: 0≤P i ≤P max ; Interference temperature limit: The total interference shall not exceed a certain threshold. Computer equipment can use optimization algorithms to optimize resource allocation. Specifically, computer equipment can use Lagrange multiplier method or convex optimization technology to solve optimization problems. Among them, the Lagrange function is: Where λ represents the Lagrange multiplier; P total represents the total power limit; by solving and Get the optimal power allocation solution.

[0066] Computer devices can use interference cancellation techniques to reduce interference. Computer devices can utilize beamforming techniques to concentrate signal energy in the target direction and reduce interference in other directions. In a multi-user environment, computer devices can adjust the signal transmission method to make interference superimpose in a certain subspace at the receiving end, thereby reducing the impact on the effective signal.

[0067] In this embodiment, a communication network deployment plan for the autonomous crossing method of large-scale UAV clusters is set up. In-depth research is carried out in aspects such as network capacity and bandwidth allocation, ultra-low latency communication and delay analysis, large-scale connection and interference management, etc. For the first time, the key technologies of 6G are systematically applied to the deployment of UAV cluster communication networks; network capacity calculation, delay analysis and interference management methods for UAV clusters are proposed, combined with advanced optimization algorithms to achieve efficient utilization of resources; the provided theoretical models and algorithms provide a feasible technical solution for the actual deployment of 6G communication network UAV clusters, and have important engineering application value.

[0068] Step 204, based on the network deployment, establish a linear state space model of the multi-UAV system for the UAV cluster. Under the linear state space model, set the cooperative control strategy, obstacle avoidance strategy and conduct model performance analysis.

[0069] After the computer device performs network deployment, it can design and implement a distributed cooperative control strategy based on the network deployment. First, the computer device can perform dynamic modeling and state estimation of the multi-UAV system, design a distributed cooperative control algorithm, and analyze stability and convergence, so as to design a distributed cooperative control strategy.

[0070] In one embodiment, a method for autonomous crossing of a UAV cluster may further include the process of designing a distributed cooperative control strategy. The specific process includes: determining the dynamic models of each UAV in the UAV cluster, linearizing the dynamic models to obtain a linear state space model of the multi-UAV system; each UAV obtains its own and the state information of neighboring UAVs and conducts information interaction; according to the linear state space model and the state information of each UAV, perform state estimation on each UAV; set the cooperative control strategy and obstacle avoidance strategy according to the state information of each UAV; based on the cooperative control strategy and obstacle avoidance strategy, perform performance analysis on the linear state space model of the multi-UAV system to obtain the analysis result.

[0071] Among them, when the computer device performs dynamic modeling and state estimation of the multi-UAV system, it can first establish a UAV dynamic model. Specifically, there are N UAVs in the UAV cluster, and the dynamic model of each UAV can be described as: Where: Denote the state vector of the \(i\)-th UAV, such as position, velocity, attitude, etc.; Denote the control input of the \(i\)-th UAV; \(f\) i (·) represents the non-linear dynamics function of the \(i\)-th UAV. For simplicity of analysis, the dynamics model can be linearized to obtain a linear state space model: Where: Denote the system matrix; Denote the control matrix.

[0072] In distributed control, the UAVs need to obtain the state information of themselves and their neighbors. Based on the high bandwidth and low latency characteristics of the 6G communication network, efficient information interaction can be achieved. Set the communication topology as a directed graph Where: Is the set of nodes, representing the UAV numbers; Is the set of edges, representing the communication links. Define the adjacency matrix \(A = [a\) ij , if \((j, i)\in\varepsilon\), then \(a\) ij > 0; otherwise \(a\) ij = 0, and the state estimation equation is: Where, Denote the estimation of the \(i\)-th UAV's own state; Denote the neighbor set of the \(i\)-th UAV.

[0073] When the computer device designs the distributed cooperative control algorithm, the UAV cluster needs to achieve the following control objectives: formation maintenance: the UAVs maintain the preset relative positions and attitudes; path tracking: the entire formation moves along the predetermined trajectory; obstacle avoidance ability: avoid obstacles and other UAVs in the dynamic environment. The computer device can design a cooperative control law, adopt a consensus control strategy, and the control input is designed as: Where, Denote the reference state of the \(i\)-th UAV, that is, the expected formation shape; Denote the state feedback gain matrix; \(b\) ij > 0 denotes the cooperative control weight; Is used to make the UAV track the reference state; Achieve the state consensus between UAVs.

[0074] Next, the computer device can introduce the artificial potential field method to achieve the obstacle avoidance function. The total control input is adjusted to: Where, Denote the input calculated by the cooperative control law; Denote the obstacle avoidance control input. The obstacle avoidance control input adopts the negative gradient of the potential field function: The obstacle avoidance potential field function is defined as: ||x i -o k ||<d safe ; where M represents the number of obstacles; o k represents the position of the k-th obstacle; η k represents the repulsive force coefficient, controlling the obstacle avoidance force; d safe represents the safety distance.

[0075] When performing stability and convergence analysis, the computer device can construct a Lyapunov function. Specifically, in this embodiment, to analyze the stability of the system, a Lyapunov function is constructed: where x i (t) represents the state vector of the i-th UAV; represents the reference state of the i-th UAV; P i is a symmetric positive definite matrix. Then, the computer device can calculate the time derivative of the Lyapunov function: Substituting the dynamic equation and control law of the UAV, we get: Under appropriate conditions (such as when P i , K i is properly selected), it can be proved that thus the system is stable.

[0076] In this embodiment, the computer device can use graph theory and matrix analysis methods to study the convergence of the system. Specifically, define the Laplacian matrix L = [l ij , where: Define the state error vector where, The error dynamics of the system can be expressed as: where A represents a block diagonal matrix composed of each A i ; B represents a block diagonal matrix composed of each B i ; K represents a block diagonal matrix composed of each K i ; represents the Kronecker product. By analyzing the eigenvalues and using the second Lyapunov method, it can be proved that the error e(t) converges to zero, that is, the system achieves consistency and tracking performance.

[0077] In this embodiment, a distributed cooperative control strategy for large-scale UAV swarm formation autonomous crossing is elaborated in detail from three aspects: dynamic modeling and state estimation of the multi-UAV system, design of distributed cooperative control algorithms, and analysis of stability and convergence. Combining the advantages of 6G networks, a distributed control method suitable for large-scale UAV swarms is proposed, introducing a consensus algorithm and an artificial potential field obstacle avoidance strategy to integrate the functions of cooperative control and obstacle avoidance; specific control laws and stability proofs are provided, providing a theoretical basis for the design and implementation of actual systems.

[0078] Step 206, in the linear state space model, collect environmental information through various sensors set on each UAV in the UAV swarm to obtain the environmental state; use the artificial potential field method to design a dynamic obstacle avoidance algorithm according to the environmental state, and adjust the UAV path in real time according to the dynamic obstacle avoidance algorithm.

[0079] In the large-scale UAV swarm autonomous crossing task, intelligent perception and dynamic obstacle avoidance are key technologies to ensure the safe and efficient operation of UAVs. Based on the high bandwidth and low latency characteristics of 6G communication technology, combined with advanced sensors and artificial intelligence algorithms, real-time perception of complex environments and dynamic obstacle avoidance of UAVs can be achieved.

[0080] Computer equipment can achieve intelligent perception and dynamic obstacle avoidance through an environmental perception model, multi-sensor data fusion, dynamic obstacle avoidance algorithm design, and path planning and real-time adjustment mechanisms.

[0081] In one embodiment, a method for autonomous crossing of a UAV swarm can also include a process of realizing intelligent perception and dynamic obstacle avoidance. The specific process includes: collecting environmental information through various sensors set on each UAV in the UAV swarm, and performing data fusion on each environmental information using a filter, and estimating the environmental state based on the fused data; using the artificial potential field method to determine the gravitational potential field and repulsive potential field according to the environmental state, and performing gradient calculation to design a dynamic obstacle avoidance algorithm; adjusting the UAV path in real time based on the dynamic obstacle avoidance algorithm and the environmental state.

[0082] UAVs need to obtain accurate information about the surrounding environment to achieve dynamic obstacle avoidance. Commonly used sensors include lidar, millimeter-wave radar, vision cameras, and ultrasonic sensors, and each sensor has different measurement characteristics and error models. In this embodiment, let the measurement value of the k-th sensor of the i-th UAV be z ik , and its true environmental state be s i , then the measurement model can be expressed as: z ik = h ik (s i ) + v ik ; where h ik (·) represents the measurement function of the k-th sensor; vik denotes the measurement noise, assumed to follow a zero-mean Gaussian distribution with covariance R ik . To fuse multi-sensor data, in this embodiment, an Extended Kalman Filter (EKF) can be adopted, and the state prediction equation can be expressed as: Measurement update equation: z i (t) = H i s i (t) + v i (t); where f(·) represents the state transition function; w i (t - 1) represents the process noise with covariance Q i ; H i represents the measurement matrix; v i (t) represents the measurement noise with covariance R i . The filtering process includes two steps: prediction and update. Among them, the prediction step is: where P i (t) represents the state covariance matrix; represents the Jacobian matrix of the state transition matrix. The update step can include: where K i (t) represents the Kalman gain matrix; I represents the identity matrix.

[0083] In this embodiment, leveraging the high bandwidth and low latency characteristics of the 6G network, the drones can share sensing data to form a distributed sensing network. Each drone sends its own sensing result s i (t) to its neighbors The fused state estimate is: where w ij represents the weight coefficient, satisfying represents the fused environmental state estimate.

[0084] In one embodiment, a method for autonomous flight of a drone swarm can further include the process of designing a dynamic obstacle avoidance algorithm. The specific process includes: determining dynamic obstacles based on the environmental state, obtaining obstacle data corresponding to the dynamic obstacles, and predicting the obstacle motion data according to the obstacle data; setting collision avoidance constraints according to the obstacle motion data; and designing and completing the dynamic obstacle avoidance algorithm using the artificial potential field method based on the environmental state and collision avoidance constraints.

[0085] Among them, the artificial potential field method can regard the targets and obstacles in the environment as a gravitational field and a repulsive field, and the total potential field function U(x) is: U(x) = U att (x) + Urep (x); where U att (x) represents the gravitational potential field that attracts the drone towards the target; U rep (x) represents the repulsive potential field that pushes the drone away from obstacles. The gravitational potential field function is expressed as: The repulsive potential field function is expressed as: where k att represents the gravitational coefficient; x represents the current position of the drone; x goal represents the target position; k rep represents the repulsive coefficient; o k represents the position of the k-th obstacle; d0 represents the repulsive influence range.

[0086] In this embodiment, the obstacle avoidance control input is the negative gradient of the potential field function, expressed as: The gradient of the gravitational potential field is: The gradient of the repulsive potential field is: where represents the gradient of the potential field function with respect to position, indicating the direction and magnitude of the force; u obs represents the obstacle avoidance control input.

[0087] When the computer device fuses obstacle avoidance and motion control, the total control input is: u = u ctrl + u obs ; where u ctrl represents the input from the cooperative control strategy, such as distributed cooperative control; u obs represents the obstacle avoidance control input. To prevent the control input from being too large, u can be normalized: where u max is the maximum allowable control input amplitude.

[0088] When the computer device implements the path planning and real-time adjustment mechanism, it first needs to use the A* algorithm for initial path planning. Specifically, before the task starts, the drone needs to plan the initial path. The A* algorithm is a commonly used heuristic search algorithm, and its cost function is: f(n) = g(n) + h(n); where f(n) represents the total cost of node n; g(n) represents the actual cost from the starting point to node n; h(n) represents the heuristic estimated cost from node n to the target. The heuristic function h(n) usually uses the Euclidean distance: h(n) = ||x n - x goal ||.

[0089] Next, the computer device can perform real-time adjustment based on the dynamic window method. In a dynamic environment, the path needs to be adjusted in real time. The Dynamic Window Approach (DWA) takes into account the kinematics and dynamic constraints of the UAV and searches for the optimal control input in the velocity space (v, ω). The evaluation function is expressed as: G = α·heading(v, ω) + β·velocity(v) + γ·obstacle(v, ω); where, heading(v, ω) represents the heading score, and the closer it is to the target direction, the higher the score; velocity(v) represents the velocity score, and the faster the velocity, the higher the score; obstacle(v, ω) represents the obstacle score, and the farther away from the obstacle, the higher the score; α, β, γ represent the weight coefficients.

[0090] Among them, the process of the real-time path adjustment algorithm may include: Perception update: Obtain the current environmental information through multi-sensor fusion; Dynamic window calculation: Calculate the feasible velocity space according to the current velocity v of the UAV cur and the acceleration limit: V = {(v, ω) | v min ≤ v ≤ v max , ω min ≤ ω ≤ ω max}; Evaluation function calculation: Calculate the evaluation function G for each velocity pair (v, ω) in V; Select the optimal control input: Select the (v * , ω * ) that maximizes G; Update the motion state: Apply (v * , ω * ) to update the velocity and position of the UAV, where, v, ω represent the linear velocity and angular velocity; v min , v max represent the velocity limits; ω min , ω max represent the angular velocity limits.

[0091] In this embodiment, from three aspects of the environmental perception model and multi-sensor data fusion, dynamic obstacle avoidance algorithm design, and path planning and real-time adjustment mechanism, the intelligent perception and dynamic obstacle avoidance method for large-scale UAV clusters based on 6G communication technology is elaborated; by utilizing the high performance of the 6G network, distributed multi-sensor data fusion and sharing are realized, and the accuracy of environmental perception is improved; combined with the extended Kalman filter, multi-sensor fusion, artificial potential field method, and dynamic window method, a complete set of intelligent perception and obstacle avoidance solutions is proposed, providing an operable guidance for the implementation of the actual system, being able to adapt to the dynamic and complex environment, and improving the autonomy and safety of the UAV cluster.

[0092] Step 208: Determine the autonomous crossing mission of the UAV swarm. Based on the autonomous crossing mission, establish a UAV kinematic model according to the adjusted UAV path. Design an optimization algorithm for path planning based on the UAV kinematic model, and perform real-time dynamic path adjustment according to the optimization algorithm.

[0093] In the autonomous crossing mission of a large-scale UAV swarm based on 6G communication technology, autonomous path planning is the key to ensuring the efficient and safe execution of the mission. Among them, the goal of autonomous path planning is to determine an optimal path for each UAV in the UAV swarm, so that while meeting the mission requirements, it can avoid collisions and obstacles and maintain the formation. Set the working space as a three-dimensional bounded space which contains static and dynamic obstacles; the position of the i-th UAV at time t is represented as q i (t); the path planning problem can be formulated as an optimization problem: The constraints can include: Kinematic constraints: Collision avoidance constraints: Obstacle avoidance constraints: where N represents the number of UAVs; J i represents the cost function of the i-th UAV; u i (t) represents the control input of the i-th UAV; d min represents the minimum safety distance between UAVs; represents the obstacle space.

[0094] Among them, the cost function J of each UAV i contains multiple components: where represents the target position of the i-th UAV; w1, w2, w3 represent weight coefficients used to balance different optimization goals; φ i (t) represents the obstacle avoidance penalty function. The obstacle avoidance penalty function can be expressed as: where M represents the number of obstacles; o k represents the position of the k-th obstacle; η k represents the penalty coefficient related to obstacle k.

[0095] In one embodiment, a method for autonomous crossing of a drone swarm may further include the process of establishing a drone dynamics model. The specific process includes: determining obstacle information based on the adjusted drone path, and calculating an obstacle avoidance penalty function based on the obstacle information; defining a cost function for each drone in the drone swarm based on the obstacle avoidance penalty function, and establishing constraint conditions; establishing a drone kinematic model based on the obstacle avoidance penalty function, the constraint conditions, and the flight parameters of each drone in the drone swarm.

[0096] Among them, considering the kinematic and dynamic constraints of the drone, for path planning, the simplified kinematic model is: The constraint conditions are: ||v i (t)||≤v max ; ||a i (t)||≤a max ; where v i (t) represents the speed of the i-th drone; a i (t) represents the acceleration of the i-th drone; v max represents the maximum allowable speed; a max represents the maximum allowable acceleration.

[0097] In one embodiment, a method for autonomous crossing of a drone swarm may further include the process of designing a path planning optimization algorithm. The specific process includes: optimizing the flight paths of each drone using distributed model predictive control based on the drone kinematic model, and performing distributed optimization on the coordination between each drone using a consensus mechanism, and using an adaptive rolling horizon control method to dynamically adjust the path in real time.

[0098] In one embodiment, the computer device may also design a reward function for the drone swarm; update the reward strategy for each drone in the drone swarm based on the reward function.

[0099] In the centralized method, the computer device needs to calculate the optimal paths of all drones simultaneously to solve the global optimization problem, which has a too high computational complexity in a large-scale swarm. To solve the scalability problem of a large-scale swarm, a distributed optimization method is adopted. Each drone solves a local optimization problem based on its own goals and the information obtained from its neighbors. In this embodiment, distributed model predictive control allows each drone to optimize its path within a finite prediction horizon H while coordinating with its neighbors. The local optimization problem of the i-th drone is: The constraint conditions may include: local dynamic constraints: q i (k + 1) = q i (k) + Δt·v i (k); v i (k + 1) = vi (k) + Δt·a i (k); Control input constraint: ||v i (k)|| ≤ v max ; ||a i (k)|| ≤ a max ; Collision avoidance with neighbors: where U i = {u i (0), …, u i (H - 1)} represents the control input sequence; Δt represents the time step; represents the neighbor set of the i-th UAV.

[0100] To ensure the coordination among UAVs, a consensus mechanism is adopted. The consensus update rule is: where α represents the consensus step coefficient; the update rule adjusts the paths of UAVs towards the average position of neighbors, which helps to maintain the formation shape.

[0101] Due to the existence of collision avoidance constraints, the optimization problem is usually non-convex. Therefore, sequential convex programming (SCP) is adopted. Sequential convex programming is an iterative method that approximates the non-convex problem as a series of convex optimization problems. In the l-th iteration, around the solution linearize the non-convex constraints: After linearization, the collision avoidance constraints become linear constraints, making the problem solvable by convex optimization methods.

[0102] In a dynamic environment, UAVs need to adjust their paths in real time. The receding horizon control (RHC) method solves the optimization problem at each time step, executes the first-step control input, and then re-optimizes at the next time step. Among them, the ultra-reliable low-latency communication (URLLC) provided by 6G communication technology is crucial for real-time coordination. Latency analysis: Let T comm be the communication latency, T comp be the computation time, and T total be the total time of each control cycle: T total = T comm + T comp + T exec ; To ensure real-time performance, it is necessary to satisfy T total ≤ Δt.

[0103] For dynamic obstacles, it is necessary to predict their future positions. Assume that the obstacle moves at a constant speed: Modified collision avoidance constraint: ||q i (k) - o k (k)|| ≥ dsafe To enhance the response ability of obstacle avoidance, the artificial potential field method can be integrated. The total potential field function: U(q i ) = U att (q i ) + U rep (q i ); Attractive potential field: Repulsive potential field: Control input based on gradient descent: where k att , k rep represent the coefficients of the attractive and repulsive potential fields; d safe represents the safety distance threshold.

[0104] In this embodiment, to further improve adaptability, multi-agent reinforcement learning (MARL) can be adopted. The reward r i (t) of the i-th unmanned aerial vehicle (UAV) at time t: Each UAV updates its policy π i to maximize the expected cumulative reward: where γ represents the discount factor, 0 < γ ≤ 1; T represents the time horizon.

[0105] In this embodiment, for the first time, 6G communication technology is combined with the autonomous path planning of large-scale UAV clusters. Utilizing its ultra-low latency and high reliability, real-time and efficient path planning is achieved; the integration of distributed optimization algorithms, artificial potential field methods, and multi-agent reinforcement learning is introduced, and an autonomous path planning method suitable for dynamic and complex environments is proposed; considering factors such as UAV dynamics constraints, communication latency, and obstacle avoidance in practice, it has high engineering feasibility and can guide actual UAV cluster tasks.

[0106] Step 210, set up a safety and fault tolerance mechanism; if any UAV in the UAV cluster has a fault, then based on the safety and fault tolerance mechanism, adjust the UAV cluster control law, and control the non-fault UAVs to operate through the adjusted UAV cluster control law to achieve autonomous crossing of the UAV cluster.

[0107] In the autonomous crossing task of large-scale UAV clusters, the design of the safety and fault tolerance mechanism is also crucial. Based on the high reliability and low latency characteristics of 6G communication technology, a highly robust UAV cluster system can be constructed.

[0108] In one embodiment, a method for autonomous crossing of an unmanned aerial vehicle (UAV) cluster may further include a process of setting up a safety and fault tolerance mechanism. The specific process includes: introducing a fault detection filter into the UAV cluster based on the safety and fault tolerance mechanism, and using the fault detection filter to detect faulty UAVs; using a fault estimator to estimate the errors of the faulty UAVs, and adjusting the UAV cluster control law according to the estimated errors, and using the adjusted UAV cluster control law for fault compensation.

[0109] In actual operation, UAVs may fail due to hardware failures, communication interruptions, or external interferences. To ensure the continuity of the overall function of the cluster, a fault tolerance control strategy needs to be designed so that the cluster can still operate normally even if some UAVs fail. Assume that the UAV cluster consists of N UAVs, and F of them may fail. To describe the fault tolerance control strategy, the following model is introduced:

[0110] UAV state equation: Where, represents the state vector of the i-th UAV; represents the control input; represents the fault vector, describing the possible faults of the UAV; A i , B i , D i represent the system matrices; D i f i (t) represents the impact of the fault on the system.

[0111] To detect faults in a timely manner, a fault detection filter is introduced. The residual generation is expressed as: r i (t) = H i x i (t) - y i (t); where, r i (t) represents the residual vector; H i represents the measurement matrix; y i (t) represents the actual measured output. Residual evaluation function: Where, S i is the residual covariance matrix. When J i (t) > γ i , it is considered that the i-th UAV has failed; where, γ i represents the fault detection threshold, which is set based on statistical characteristics.

[0112] After detecting a fault, the control law needs to be adjusted to compensate for the fault impact. The fault tolerance control law is expressed as: Where, represents the control input under normal conditions; K i represents the fault tolerance gain matrix; Represents the fault estimated value. The fault estimator is expressed as: where, L i represents the fault estimation gain matrix, which needs to be designed to minimize the estimation error.

[0113] In a UAV swarm, communication links may be subject to attacks such as eavesdropping, tampering, and denial of service. To ensure communication security, a security protocol needs to be designed. Utilizing the high bandwidth characteristics of 6G networks, advanced encryption algorithms are adopted for communication protection. The encryption at the sender is expressed as: The decryption at the receiver is expressed as: where, m i (t) represents the plaintext message; c i (t) represents the ciphertext message; represents the encryption and decryption functions; k i represents the key; the key k i can adopt a symmetric key or public key encryption mechanism.

[0114] To prevent fake nodes, identity authentication is required. Identity authentication based on the Elliptic Curve Digital Signature Algorithm (ECDSA): When generating a signature, the sender generates a signature σ i which is expressed as: When verifying the signature, the receiver verifies the signature and is expressed as: where, priv i , pub i represent the private key and public key of the i-th UAV. Utilizing the secure channel of 6G networks, Quantum Key Distribution (QKD) technology is adopted to achieve secure sharing of keys.

[0115] Considering the influence of the fault-tolerant control strategy on the system stability, Lyapunov stability analysis is adopted. Construct a Lyapunov function: where, represents the state error; P i is a symmetric positive definite matrix. The derivative of the Lyapunov function: To ensure it is necessary to design appropriate K i and P i .

[0116] The Markov chain model is adopted to describe the reliability of the system, and the state transition probability matrix is set. Among them, it is assumed that the system has two states: normal (N) and faulty (F). where, p NN represents the probability that the system remains normal when it is normal; p NF = 1 - p NN represents the probability that the system changes from normal to faulty; p FFIndicates the probability that the system remains in a faulty state after a fault; p FN = 1 - p FF Indicates the probability that the system recovers from a faulty state to a normal state. In the steady state, the probability that the system is in a normal state is π N : Among them, by adjusting the fault tolerance strategy, p is increased FN , improving the reliability of the system.

[0117] In this embodiment, the mean time between failures (MTBF) and the mean time to repair (MTTR) can be respectively expressed as: Among them, λ represents the failure rate; μ represents the repair rate; the system availability can be expressed as: By increasing the MTBF and reducing the MTTR, the system availability can be improved.

[0118] In this embodiment, there are three aspects: the fault tolerance control strategy of the UAV swarm, the secure communication protocol based on 6G network, and the system stability and reliability analysis; the quantum key distribution and high-reliability low-latency communication in 6G communication technology are introduced to enhance the communication security and real-time performance of the UAV swarm; a fault tolerance control strategy based on a fault detection filter is proposed, the stability analysis is carried out by combining the Lyapunov method, and the Markov chain model is used to evaluate the system reliability; specific control laws, encrypted communication mechanisms and reliability models are provided, providing a theoretical basis and technical guidance for the design and implementation of the actual system.

[0119] A method for autonomous crossing of a UAV swarm provided by this application uses 6G communication technology to build a low-latency, high-bandwidth communication network for a large-scale UAV swarm, realizing real-time data transmission and information sharing among UAVs in the swarm and with the ground control center; based on the low-latency characteristics of the 6G network, a distributed cooperative control algorithm is proposed, allowing UAVs to realize autonomous path planning and obstacle avoidance operations on the basis of obtaining their own and neighboring UAV state information, ensuring the efficient cooperation and stability of the UAV swarm; combined with the high-precision perception and positioning capabilities of the 6G network, each UAV in the swarm can real-time sense the surrounding obstacles and environmental changes, dynamically adjust the flight path, and avoid collisions and dangers; through the real-time perception of the task scenario, the present invention proposes an adaptive task allocation and path optimization method based on intelligent algorithms, ensuring that in a complex environment, the UAV swarm can quickly adjust the flight formation according to the actual situation and achieve the optimal execution of tasks; in order to improve the reliability of the system, this application designs a multi-layer security mechanism and fault tolerance strategy based on 6G communication technology, which can ensure the normal operation of the remaining swarm and the continuation of tasks in case of individual UAV failures or communication interruptions.

[0120] It should be understood that although the steps in the above flowchart are shown in sequence according to the indication of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the above flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0121] In one embodiment, as Figure 3 shown, a drone swarm autonomous crossing system is provided, including: a network deployment module 310, a model establishment module 320, a path adjustment module 330, an optimization module 340, and a safety and fault tolerance module 350, where:

[0122] The network deployment module 310 is used to obtain the network capacity requirements corresponding to the drone swarm, expand the channel capacity based on the network capacity requirements; perform delay analysis on each drone in the drone swarm, and introduce non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements of the drone swarm to complete network deployment;

[0123] The model establishment module 320 is used to establish a linear state space model of the multi-drone system for the drone swarm based on network deployment. Under the linear state space model, set cooperative control strategies, obstacle avoidance strategies, and perform model performance analysis;

[0124] The path adjustment module 330 is used to collect environmental information through various sensors set on each drone in the drone swarm under the linear state space model to obtain the environmental state; design and complete a dynamic obstacle avoidance algorithm according to the environmental state using the artificial potential field method, and adjust the drone path in real time according to the dynamic obstacle avoidance algorithm;

[0125] The optimization module 340 is used to determine the autonomous crossing task of the drone swarm, establish a kinematic model of the drone based on the adjusted drone path according to the autonomous crossing task, design an optimization algorithm for path planning based on the kinematic model of the drone, and perform real-time dynamic path adjustment according to the optimization algorithm;

[0126] The safety and fault tolerance module 350 is used to set safety and fault tolerance mechanisms; if any drone in the drone swarm has a fault, then adjust the control law of the drone swarm based on the safety and fault tolerance mechanism, and control the non-fault drones to run through the adjusted control law of the drone swarm to achieve the autonomous crossing of the drone swarm.

[0127] In one embodiment, the network deployment module 310 is further configured to obtain the network capacity requirements corresponding to the UAV cluster, and based on the network capacity requirements, calculate a bandwidth allocation strategy based on the channel capacity, and expand the channel capacity according to the bandwidth allocation strategy; determine the end-to-end delay, signal propagation delay, signal processing delay, transmission delay, and queuing delay of each UAV in the UAV cluster, and perform delay analysis; determine the connection requirements of the UAV cluster, and introduce non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements to complete network deployment.

[0128] In one embodiment, the model establishment module 320 is further configured to determine the dynamic models of each UAV in the UAV cluster, perform linearization processing on the dynamic models to obtain a linear state space model of the multi-UAV system; each UAV obtains its own and neighbor UAVs' state information and performs information interaction; perform state estimation on each UAV according to the linear state space model and the state information of each UAV; set cooperative control strategies and obstacle avoidance strategies according to the state information of each UAV; based on the cooperative control strategies and obstacle avoidance strategies, perform performance analysis on the linear state space model of the multi-UAV system to obtain an analysis result.

[0129] In one embodiment, the path adjustment module 330 is further configured to collect environmental information through various sensors installed on each UAV in the UAV cluster, perform data fusion on each environmental information using a filter, and estimate the environmental state based on the fused data; use the artificial potential field method to determine the gravitational potential field and repulsive potential field according to the environmental state, and perform gradient calculation to design and complete a dynamic obstacle avoidance algorithm; based on the dynamic obstacle avoidance algorithm and the environmental state, adjust the UAV path in real time.

[0130] In one embodiment, the path adjustment module 330 is further configured to determine dynamic obstacles based on the environmental state, obtain the obstacle data corresponding to the dynamic obstacles, and predict the obstacle motion data according to the obstacle data; set collision avoidance constraints according to the obstacle motion data; use the artificial potential field method to design and complete a dynamic obstacle avoidance algorithm according to the environmental state and the collision avoidance constraints.

[0131] In one embodiment, the optimization module 340 is further configured to determine obstacle information according to the adjusted UAV path, and calculate an obstacle avoidance penalty function according to the obstacle information; based on the obstacle avoidance penalty function, define the cost function of each UAV in the UAV cluster, and establish constraint conditions; according to the obstacle avoidance penalty function and the constraint conditions, combine the flight parameters of each UAV in the UAV cluster to establish a UAV kinematic model.

[0132] In one embodiment, the optimization module 340 is further configured to optimize the flight paths of the respective drones based on the drone kinematic model using distributed model predictive control, and use a consensus mechanism to perform distributed optimization on the coordination between the respective drones, and use an adaptive rolling horizon control method to dynamically adjust the paths in real time.

[0133] In one embodiment, the optimization module 340 is further configured to design a reward function for the drone swarm; update the reward strategies of the respective drones in the drone swarm based on the reward function.

[0134] In one embodiment, the safety and fault tolerance module 350 is further configured to introduce a fault detection filter into the drone swarm based on a safety and fault tolerance mechanism, and use the fault detection filter to detect faulty drones; use a fault estimator to estimate the errors of the faulty drones, and adjust the drone swarm control law according to the estimated errors, and perform fault compensation using the adjusted drone swarm control law.

[0135] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structural diagram may be as Figure 4 shown. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a method for autonomous crossing of a drone swarm. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device may be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the outer shell of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0136] Those skilled in the art can understand that Figure 4 the structure shown in

[0137] is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0138] 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 of the method for autonomous crossing of a drone swarm are implemented.

[0139] Those of ordinary skill in the art can understand that all or part of the processes in the above-described embodiment methods can be completed by instructing relevant hardware through a computer program. 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 above-described method embodiments. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0140] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, 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, it should be considered to be within the scope described in this specification.

[0141] The above-described embodiments merely represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. An autonomous crossing method for a UAV swarm, characterized in that The method includes: Obtaining the network capacity requirement corresponding to the UAV cluster, and expanding the channel capacity based on the network capacity requirement; performing delay analysis on each UAV in the UAV cluster, and introducing non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements of the UAV cluster to complete network deployment; Based on the network deployment, establishing a linear state space model of the multi-UAV system for the UAV cluster, and setting cooperative control strategies, obstacle avoidance strategies and performing model performance analysis under the linear state space model; Under the linear state space model, collecting environmental information through each sensor set on each UAV in the UAV cluster to obtain the environmental state; using the artificial potential field method to design and complete a dynamic obstacle avoidance algorithm according to the environmental state, and adjusting the UAV path in real time according to the dynamic obstacle avoidance algorithm; Determining the autonomous crossing task of the UAV cluster, establishing a UAV kinematic model based on the adjusted UAV path according to the autonomous crossing task, designing an optimization algorithm for path planning based on the UAV kinematic model, and performing real-time dynamic path adjustment according to the optimization algorithm; Setting up a safety and fault tolerance mechanism; if any UAV in the UAV cluster has a fault, then adjusting the UAV cluster control law based on the safety and fault tolerance mechanism, and controlling the non-fault UAVs to operate through the adjusted UAV cluster control law to achieve autonomous crossing of the UAV cluster.

2. The method for autonomous crossing of a drone swarm according to claim 1, wherein The obtaining the network capacity requirement corresponding to the UAV cluster, expanding the channel capacity based on the network capacity requirement; performing delay analysis on each UAV in the UAV cluster, and introducing non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements of the UAV cluster to complete network deployment includes: Obtaining the network capacity requirement corresponding to the UAV cluster, and based on the network capacity requirement, calculating a bandwidth allocation strategy according to the channel capacity, and expanding the channel capacity according to the bandwidth allocation strategy; Determining the end-to-end delay, signal propagation delay, signal processing delay, transmission delay, and queuing delay of each UAV in the UAV cluster, and performing delay analysis; Determining the connection requirements of the UAV cluster, and introducing non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements to complete network deployment.

3. The method for autonomous crossing of an unmanned aerial vehicle cluster according to claim 1, wherein Establishing a linear state space model of the multi-UAV system for the UAV cluster, and setting cooperative control strategies, obstacle avoidance strategies and performing model performance analysis under the linear state space model includes: Determining the dynamic model of each UAV in the UAV cluster, and performing linearization processing on the dynamic model to obtain a linear state space model of the multi-UAV system; Each UAV obtains its own and the state information of neighboring UAVs, and performs information interaction; Performing state estimation on each UAV according to the linear state space model and the state information of each UAV; Setting cooperative control strategies and obstacle avoidance strategies according to the state information of each UAV; Based on the collaborative control strategy and the obstacle avoidance strategy, perform a performance analysis on the linear state space model of the multi-UAV system to obtain the analysis results.

4. The method for autonomous crossing of an unmanned aerial vehicle cluster according to claim 1, wherein Collect environmental information through the sensors set on each UAV in the UAV cluster to obtain the environmental state; Use the artificial potential field method to design and complete a dynamic obstacle avoidance algorithm according to the environmental state, and adjust the UAV path in real time according to the dynamic obstacle avoidance algorithm, including: Collect environmental information through the sensors set on each UAV in the UAV cluster, and perform data fusion on each piece of the environmental information using a filter, and estimate the environmental state based on the fused data; Use the artificial potential field method to determine the gravitational potential field and the repulsive potential field according to the environmental state, and perform gradient calculation to design and complete the dynamic obstacle avoidance algorithm; Based on the dynamic obstacle avoidance algorithm and the environmental state, adjust the UAV path in real time.

5. The method for autonomous crossing of a drone swarm according to claim 4, wherein Use the artificial potential field method to design and complete a dynamic obstacle avoidance algorithm according to the environmental state, including: Determine dynamic obstacles based on the environmental state, obtain the obstacle data corresponding to the dynamic obstacles, and predict the obstacle movement data according to the obstacle data; Set collision avoidance constraints according to the obstacle movement data; Use the artificial potential field method to design and complete a dynamic obstacle avoidance algorithm according to the environmental state and the collision avoidance constraints.

6. The method for autonomous crossing of an unmanned aerial vehicle cluster according to claim 1, wherein Establish a UAV kinematic model according to the adjusted UAV path, including: Determine obstacle information according to the adjusted UAV path, and calculate an obstacle avoidance penalty function according to the obstacle information; Based on the obstacle avoidance penalty function, define the cost function of each UAV in the UAV cluster, and establish constraint conditions; According to the obstacle avoidance penalty function and the constraint conditions, combine the flight parameters of each UAV in the UAV cluster to establish a UAV kinematic model.

7. The method for autonomous crossing of a drone swarm according to claim 1, wherein Design an optimization algorithm for path planning based on the UAV kinematic model, and perform real-time dynamic path adjustment according to the optimization algorithm, including: Based on the UAV kinematic model, use distributed model predictive control to optimize the flight paths of each UAV, use a consensus mechanism to perform distributed optimization on the coordination between each UAV, and use an adaptive rolling horizon control method to adjust the path in real time dynamically.

8. The method for autonomous crossing of a drone swarm according to claim 7, wherein The method further includes: Design a reward function for the UAV cluster; Update the reward strategy for each UAV in the UAV cluster based on the reward function.

9. The method for autonomous crossing of a drone swarm according to claim 1, characterized in that, If any UAV in the UAV cluster has a fault, then adjust the UAV cluster control law based on the safety and fault tolerance mechanism, and control the non-fault UAVs to operate through the adjusted UAV cluster control law, including: Based on the safety and fault tolerance mechanism, introduce a fault detection filter in the UAV cluster, and use the fault detection filter to detect the fault UAV; Use a fault estimator to perform error estimation on the fault UAV, adjust the UAV cluster control law according to the estimated error, and perform fault compensation using the adjusted UAV cluster control law.

10. An autonomous crossing system for an unmanned aerial vehicle cluster, characterized in that, The system includes: A network deployment module, which is used to obtain the network capacity requirements corresponding to the UAV cluster, expand the channel capacity based on the network capacity requirements; conduct latency analysis on each UAV in the UAV cluster, and introduce non-orthogonal multiple access technology for interference management and resource allocation according to the connection requirements of the UAV cluster to complete network deployment; A model establishment module, which is used to establish a linear state space model of the multi-UAV system for the UAV cluster based on the network deployment. Under the linear state space model, set cooperative control strategies, obstacle avoidance strategies and conduct model performance analysis; A path adjustment module, which is used to collect environmental information through various sensors set on each UAV in the UAV cluster under the linear state space model to obtain the environmental state; design and complete a dynamic obstacle avoidance algorithm according to the environmental state using the artificial potential field method, and adjust the UAV path in real time according to the dynamic obstacle avoidance algorithm; An optimization module, which is used to determine the autonomous crossing task of the UAV cluster, establish a kinematic model of the UAV based on the adjusted UAV path according to the autonomous crossing task, design an optimization algorithm for path planning based on the kinematic model of the UAV, and perform real-time dynamic path adjustment according to the optimization algorithm; A safety and fault tolerance module, which is used to set up safety and fault tolerance mechanisms; if any UAV in the UAV cluster has a fault, then adjust the UAV cluster control law based on the safety and fault tolerance mechanism, and control the non-fault UAVs to operate through the adjusted UAV cluster control law to achieve the autonomous crossing of the UAV cluster.

Images