Method for collaborative control of unmanned aerial vehicle and ground vehicle under privacy protection and on-demand communication

Abstract

The application belongs to the technical field of unmanned aerial vehicle cluster control, and specifically discloses a privacy protection and on-demand communication unmanned aerial vehicle and ground vehicle cooperative control method. The method designs a privacy protection and on-demand communication unmanned aerial vehicle and ground vehicle cooperative controller, first designs a distributed dynamic protocol based on an unmanned aerial vehicle and ground vehicle cooperative system model; then designs a privacy protection mechanism, adds random noise in the data transmission process; establishes an on-demand communication mechanism, when the local event error is greater than a preset threshold, the trigger time is updated and the reference state information is sent, otherwise, it remains silent; finally, based on the received adjacent intelligent body information and the local state, the reference state signal is updated and the distributed cooperative control law is designed. The application effectively guarantees the stable operation and cooperative performance of the unmanned aerial vehicle and ground vehicle cooperative system while reducing the communication frequency and bandwidth occupation.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) swarm control technology, specifically relating to a method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication. Background Technology

[0002] With the development of intelligent unmanned system technology, air-ground collaborative systems composed of drones and ground vehicles have shown broad application prospects in areas such as target tracking, disaster relief, environmental monitoring, collaborative inspection, and complex scenario operations due to their complementary advantages of global perception in the air and near-field execution on the ground. To achieve mission collaboration, drones and ground vehicles typically need to continuously exchange location information, status information, mission information, or control information via wireless networks.

[0003] Existing air-ground collaborative systems typically employ periodic sampling and fixed-frequency communication. Each platform continuously broadcasts its own state or neighbor information at preset intervals to achieve task collaboration. To address the privacy protection requirements in distributed control, various methods have been proposed, including homomorphic encryption and state decomposition. Homomorphic encryption uses encryption technology to process sensitive information, enabling the system to perform calculations directly on encrypted data without decryption; however, this method incurs significant communication and computational burdens, making it unsuitable for scenarios with limited computing and network resources. State decomposition works by abstracting the nodes of an agent into two or more virtual child nodes: one child node remains private and does not participate in inter-node communication, while the remaining child nodes actively interact with neighboring nodes; however, this method is only effective in specific scenarios with simple system architectures, and its effectiveness is limited in other application scenarios. As the system scales up, this method becomes unsuitable.

[0004] Patent Document 1 discloses a data privacy protection method for resource-constrained UAV swarm networks. This method utilizes homomorphic encryption to address the difficulty in effectively balancing resource efficiency, privacy security, and accurate convergence in UAV swarm network data privacy protection methods, effectively protecting the privacy of the initial position of the UAVs. However, this method requires significant computational resources, making it difficult to use resource-constrained physical devices. Patent Document 2 discloses an event-driven privacy protection method for collaborative formation of UAV swarm systems. This method introduces a control term based on an event-triggered strategy on the basis of traditional privacy-preserving collaborative algorithms and uses a class of finite-time masking functions to protect real communication data, achieving collaborative formation of UAV swarm systems. However, the periodic continuous communication method causes significant bandwidth consumption and energy overhead, and high-frequency information interaction increases the number of observation samples for attackers, increasing the possibility of them statistically inferring and reconstructing the sensitive state of the system.

[0005] Patent Document 1: Chinese Invention Patent Application, Publication No.: CN120640275A, Publication Date: 2025.09.12.

[0006] Patent document 2 Chinese invention patent application, publication number: CN119937596A, publication date: 2025.05.06. Summary of the Invention

[0007] The purpose of this invention is to propose a collaborative control method for UAVs and ground vehicles under privacy protection and on-demand communication, so as to effectively suppress the leakage of sensitive states while reducing the communication burden and ensuring the operational performance of the UAV and ground vehicle collaborative system.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: A method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication includes the following steps: Step 1. Establish a multi-agent collaborative system model consisting of UAVs and ground vehicles, and construct a communication topology diagram; Step 2. Design a collaborative controller for UAVs and ground vehicles based on privacy protection and on-demand communication. The specific process is as follows: First, based on the multi-agent cooperative system model established in step 1, a distributed dynamic protocol is designed. Then, a privacy protection mechanism is designed to add random noise during data transmission; at the same time, an on-demand communication mechanism is established, which updates the trigger time and sends a reference status signal when the error of a local event exceeds a preset threshold, otherwise it remains silent. Finally, based on the received neighbor agent information and the local state update reference state signal, a distributed cooperative control law is designed. Step 3. Based on the privacy-preserving and on-demand communication-based UAV-ground vehicle collaborative controller designed in Step 2, realize collaborative control of UAV and ground vehicle.

[0009] Furthermore, based on the aforementioned method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication, this invention also proposes a corresponding collaborative control system for UAVs and ground vehicles under privacy protection and on-demand communication, which adopts the following technical solution: A collaborative control system for drones and ground vehicles, featuring privacy protection and on-demand communication, includes the following modules: The model building module is used to build a multi-agent collaborative system model consisting of UAVs and ground vehicles, and to construct a communication topology diagram; The controller design module is used to design a collaborative controller for UAVs and ground vehicles based on privacy protection and on-demand communication. The specific process is as follows: First, based on a multi-agent cooperative system model, a distributed dynamic protocol is designed; Then, a privacy protection mechanism is designed to add random noise during data transmission; at the same time, an on-demand communication mechanism is established, which updates the trigger time and sends a reference status signal when the error of a local event exceeds a preset threshold, otherwise it remains silent. Finally, based on the received neighbor agent information and the local state update reference state signal, a distributed cooperative control law is designed. The UAV and ground vehicle collaborative control module is used to realize collaborative control between UAVs and ground vehicles based on a privacy-preserving and on-demand communication-based UAV and ground vehicle collaborative controller.

[0010] The present invention has the following advantages: As described above, this invention proposes a collaborative control method for UAVs and ground vehicles under privacy protection and on-demand communication. This method constructs an on-demand communication mechanism centered on event triggering, ensuring that data transmission only occurs when the state error exceeds a preset threshold, significantly reducing the communication frequency and network resource consumption of the air-to-ground collaborative system. Furthermore, this invention designs a privacy protection mechanism by introducing random noise into the transmitted data to mask the true information, effectively preventing eavesdroppers from obtaining the real information while ensuring consensus performance. This method requires no additional overhead and is relatively simple to compute. Simultaneously, this invention introduces a time-varying step size, ensuring that the system can still achieve mean-square consistency under noise interference. This invention is applicable to various air-to-ground collaborative scenarios, such as UAV-to-ground vehicle collaboration, robot formation, and intelligent transportation, and has strong versatility, scalability, and engineering application value. Attached Figure Description

[0011] Figure 1 This is a flowchart of the collaborative control method between UAVs and ground vehicles under privacy protection and on-demand communication in an embodiment of the present invention; Figure 2 This is a communication topology diagram in an embodiment of the present invention; Figure 3 The output state response curves of different intelligent agents in the embodiments of the present invention are shown. Figure 4 This is a diagram showing the disturbed interaction data of different intelligent agents in an embodiment of the present invention; Figure 5 This is a diagram showing the triggering times of different intelligent agents in embodiments of the present invention. Detailed Implementation

[0012] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments: Example 1 To address the problems of high communication resource consumption, easy eavesdropping of sensitive status information, and difficulty in balancing privacy protection and collaborative control performance in UAV-ground vehicle collaborative systems operating in open communication environments, this invention proposes a collaborative control method for UAVs and ground vehicles under privacy protection and on-demand communication. This method reduces communication frequency while effectively protecting the sensitive status of the system and ensuring the collaborative control performance and convergence accuracy of the system.

[0013] like Figure 1 As shown in Embodiment 1, a method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication is described, including the following steps: Step 1. Establish a multi-agent collaborative system model consisting of UAVs and ground vehicles, and construct a communication topology diagram.

[0014] Consider a combination of drones and ground vehicles The collaborative system model of multiple agents, treating each agent as a node, has the following dynamic equation: ; in , For intelligent agents The state signal, the initial state is denoted as ; For intelligent agents Control input; For intelligent agents Measurement output; , , These represent the dimensions of the system state, input, and output, respectively.

[0015] matrix , , The dimensions are compatible, and the values ​​differ for different agents, thus reflecting the heterogeneity of the system.

[0016] Construct a communication topology graph This is used to describe the information interaction relationship between drones and ground vehicles; in express A collection of intelligent agents ; Denotes the set of edges. , Represents intelligent agents Received from intelligent agent Information; Represents the adjacency matrix. , for The element; when hour, ,otherwise .

[0017] This embodiment considers a collaborative system between a drone and a ground vehicle, comprising 10 intelligent agents. ,like Figure 2 As shown, agents 1-5 are ground vehicles, and agents 6-10 are drones. The output responses of different agents (drones and ground vehicles) are as follows: Figure 3 As shown, the triggering time diagrams for different agents are as follows: Figure 5 As shown.

[0018] In this embodiment, potential attackers are assumed to be eavesdroppers. Eavesdroppers are external eavesdroppers familiar with the network topology and capable of monitoring the communication channels of agents. It is assumed that the eavesdropper can continuously monitor the external observable state set of the agents. Compared to other attack forms discussed in multi-agent system security research (such as honest but curious nodes, malicious information injectors, etc.), external eavesdroppers pose a potentially greater threat because they can infer the true state trajectory of agents based on observed communication information between agents, especially when sensitive information is involved.

[0019] Step 2. Design a collaborative controller for UAVs and ground vehicles based on privacy protection and on-demand communication. The specific process is as follows: First, based on the multi-agent cooperative system model established in step 1, a distributed dynamic protocol is designed. Then, a privacy protection mechanism is designed by adding random noise during data transmission. At the same time, an on-demand communication mechanism is established, whereby the trigger time is updated and a reference state signal is sent when the error of a local event exceeds a preset threshold; otherwise, the system remains silent. Finally, the reference state signal is updated based on the received information from neighboring agents and the local state, and a distributed cooperative control law is designed.

[0020] The specific process of designing the distributed dynamic protocol in this embodiment is as follows: To achieve average output consistency in the collaborative system between UAVs and ground vehicles, a hierarchical control structure is adopted, with each intelligent agent... The upper layer generates an internal reference state, and the lower-layer physical system tracks this reference state in real time. The distributed dynamic protocol is designed as follows: ; in and respectively intelligent agents and intelligent agents The reference state signal, whose initial conditions are initialized by the local output, i.e. , express initial value, .

[0021] For time-varying step size, ; Represents intelligent agents The set of neighbors (neighboring agents), .

[0022] and It is the gain matrix, and ; , This represents the unknown matrix to be solved.

[0023] matrix , The solution is obtained from formula (1).

[0024] (1) in for 3D identity matrix.

[0025] The privacy protection mechanism in this embodiment is as follows: To protect the privacy of drone-ground vehicle collaborative systems, noise is introduced when exchanging information with neighboring intelligent agents; that is, noise is added to the intelligent agents. When communicating with neighboring agents, the true reference state signal is not transmitted. Instead, it transmits signals with random noise added. Its expression is as follows: ; in Represents Laplace noise, which follows a Laplace distribution. .

[0026] For time-varying decay noise scale parameters, , , Given parameters, Indicates the time step. This represents a p-dimensional identity matrix.

[0027] Used to adjust noise intensity; when When the value is close to 1, the noise disturbance is strong, and the privacy protection effect is good; when When the value is close to 0, the transmitted signal is closer to the true reference state signal.

[0028] The design reflects a balance between privacy protection and collaborative control performance. Noise variables corresponding to different agents, different times, and different vector components are defined to be independent to avoid exploitation of statistical correlations between noises, thereby further improving the effectiveness of privacy protection and the feasibility of theoretical analysis.

[0029] This invention introduces random perturbations into the transmitted information, making it impossible for external receivers or potential eavesdroppers to directly recover the agent's true reference state from the transmitted data, thereby achieving the effect of hiding sensitive information and protecting privacy.

[0030] In this embodiment, the on-demand communication mechanism is specifically as follows: Define discrete sampling time ; For the global sampling period, , , Represents natural numbers.

[0031] For intelligent agents Construct its local trigger time sequence ; in Represents intelligent agents The triggering time, , This represents the discrete sampling time.

[0032] At discrete sampling time intelligent agent Local event error for: ; in for Reference state signal at time, For intelligent agents exist Reference state signal at any given time.

[0033] Set the trigger condition as follows: ; in Represents the L2 norm; The threshold is determined by a time-varying function that is adaptively adjusted based on the consistency error of the reference state signal. ; in Indicates an adjustable parameter. This indicates the signal after random noise has been added. exist The value at any given moment.

[0034] When the above triggering conditions are met, the trigger time is updated and a reference status signal is sent; otherwise, silence is maintained. During the triggering communication process, the sent reference status information undergoes privacy protection processing.

[0035] The distributed cooperative control law is designed as follows: ; The reference state signal of the virtual layer is updated as follows: ; in Represents intelligent agents exist Reference state signal at time, Indicates time-varying step size exist The value at any given moment.

[0036] After completing the design of the UAV-ground vehicle cooperative controller based on privacy protection and on-demand communication, convergence and privacy analysis are performed on the UAV-ground vehicle cooperative system controlled by the UAV-ground vehicle cooperative controller based on privacy protection and on-demand communication.

[0037] The specific process of convergence analysis is as follows: If for any initial conditions, for each agent The following relationships hold true for all of them: The system is said to have achieved average output consistency; where... For intelligent agents The initial measurement output.

[0038] To achieve average output consistency, the system model and communication topology must satisfy the following conditions: I. For each intelligent agent , It is controllable, and It is detectable; among which Indicates the first The control pair consists of the system matrix and the input matrix of each agent. Indicates the first The detection pair is formed by the output matrix of each agent and the system matrix.

[0039] II. For each agent There exists a matrix and This satisfies the following discrete-time adjustment equation: .

[0040] III. Time-varying step size The following random approximation conditions must be met: ; ; ; in Represents the Laplace matrix, ; for The largest non-zero eigenvalue; for The largest eigenvalue; For degree matrix, , It is the sum of the weights of all edges in the connected component. Represents a diagonal matrix; Indicates time-varying step size The upper boundary.

[0041] when or At that time, noise and They are independent of each other.

[0042] IV. Communication Topology Diagram They are connected.

[0043] and If the Schur matrix is ​​used, then under the distributed cooperative control law, the UAV-ground vehicle cooperative system can achieve mean-square output consistency, i.e.: ; in This represents the vector formed by the measurement outputs of all agents. A vector representing the reference state signals of all agents. , col() represents a column vector. Represents an N-dimensional column vector of all ones. Indicates the Kronecker product. It indicates a desire for the expected value.

[0044] Choosing Lyapunov functions for: ; in , express Reference state signal at any given time; , This represents an N-dimensional identity matrix.

[0045] The Lyapunov function at the next moment for: .

[0046] Taking a one-step difference of the Lyapunov function along the system trajectory and obtaining its expected value, and considering that the noise satisfies the zero-mean property, the cross term can be eliminated after obtaining the expected value, resulting in: ; in for The mathematical expectation, for The mathematical expectation, Indicates the system at the 1st The convergence rate parameter at time t is: This indicates that the system state error has a decreasing trend under the mean. This represents the additional perturbation term caused by noise, discretization error, and triggering error; and , .

[0047] Based on the random approximation condition, we obtain ;Right now .

[0048] Because the reference state signal is implemented using piecewise constants, that is, for any ,have .

[0049] therefore ;in .

[0050] Therefore, we get: .

[0051] therefore: .

[0052] Based on the above convergence conditions, the UAV-ground vehicle collaborative system can achieve average output consistency in the mean square sense. Figure 3 It can be seen that all nodes have achieved average output consistency.

[0053] This invention employs a time-varying step size and combines it with stochastic approximation theory to achieve consistent convergence of the mean square output of the algorithm for heterogeneous cooperative systems composed of UAVs and ground vehicles.

[0054] The specific process of privacy analysis is as follows: Define real numbers If there exists a positive integer , so that: ; This is called the initial state set. and It is a set of adjacent datasets, that is and The difference only in the initial state of an agent does not exceed .

[0055] in and Representing adjacent initial state sets respectively and The Middle The initial measurement output of each agent.

[0056] definition If for any two adjacent initial state sets , and observation set They all Then it is called a mechanism. satisfy - Differential privacy.

[0057] in This represents a random mechanism acting on the initial set of states; and These respectively represent when the input dataset is and Time mechanism The random output.

[0058] and These respectively represent the input as and Time mechanism The random output falls into the observation set The probability of.

[0059] Describing the natural constant e The power is used to characterize the maximum deviation between the output distributions of adjacent datasets; If is a constant, The smaller the value, the less sensitive the mechanism output is to changes in a single initial state, and thus the stronger its privacy protection capabilities.

[0060] Define dataset and Observation set and noise sequences Given an initial dataset Under the condition of noise sample paths, the system state trajectory is denoted as ,mechanism In time Sensitivity to time for: ; in To represent any time range, supremum represents the supremum. and They represent in Given an initial dataset at any given time and The system state trajectory under the following conditions; Describing the L1 norm, express The signal after random noise has been added at any time. express Random noise at any given moment.

[0061] Under the aforementioned method for collaborative control of UAVs and ground vehicles with privacy protection and on-demand communication, the mechanism... The sensitivity requirement is met: ; in Indicating sensitivity The upper realm, Represents a product of products. This represents the smallest eigenvalue of the degree matrix.

[0062] Sensitivity represents neighboring datasets and In the same observation set The upper bound of the trajectory deviation generated reflects the degree to which the output responds to changes in the input, and is a key quantity in the design of privacy mechanisms. In privacy protection mechanisms, the goal is to prevent attackers from effectively recovering sensitive information from the input data through the output. Higher sensitivity means that even small changes in the input can easily cause significant changes in the output trajectory, thus requiring stronger privacy protection measures.

[0063] Privacy noise is defined as independent Laplace noise, whose components are independent and identically distributed, satisfying the following conditions: ,mechanism Over any time range - Differential privacy, and its corresponding privacy budget for: .

[0064] Since the noise injected by each agent at each time step in step 2 is independent and follows a Laplace distribution, therefore, in a given dataset... At that time, the mechanism outputs a sequence The joint probability density function is for: ; in Represents intelligent agents Given an initial dataset The system state trajectory below, Represents the set of all agent-time index pairs. , This represents the Tth sampling time.

[0065] For any adjacent dataset ~ Using the triangle inequality, we get: ; in, and These represent the initial datasets, respectively. The system state trajectory and joint probability density function under the given conditions. Describing the natural constant e Power of 1.

[0066] Therefore, mechanism In the time interval The above satisfies - Differential privacy. Different agents output trajectories such as... Figure 4 As shown, the privacy protection mechanism in step 2 effectively masks the real information by introducing random perturbations during the original data processing, making it difficult for external observers to accurately deduce the original sensitive data from the published results.

[0067] This invention achieves a high level of privacy protection. Privacy protection; among which Indicators of privacy The upper boundary.

[0068] This invention achieves this by rationally selecting the time-varying step size. and noise parameters In this form, a predefined level of differential privacy is achieved.

[0069] Step 3. Based on the privacy-preserving and on-demand communication-based UAV-ground vehicle collaborative controller designed in Step 2, realize collaborative control of UAV and ground vehicle.

[0070] While existing on-demand communication methods reduce communication frequency to some extent, they primarily focus on communication efficiency and control performance, lacking systematic mitigation of the risk of data privacy leaks. Furthermore, the collaborative system comprised of UAVs and ground vehicles is characterized by heterogeneity, dynamic coupling, and network openness, making existing methods difficult to directly apply to the privacy protection requirements of on-demand communication scenarios. Addressing the high communication resource consumption issues in existing air-to-ground collaborative systems during continuous communication, this invention proposes a collaborative control method for UAVs and ground vehicles under on-demand communication with privacy protection. It designs an on-demand communication mechanism, transmitting information only when certain conditions are met. Simultaneously, noise is introduced into the communication data to interfere with sensitive information in the system, and a distributed collaborative control law is designed to achieve task collaboration between UAVs and ground vehicles. This method includes: establishing a collaborative system model of UAVs and ground vehicles; designing event triggering conditions to achieve on-demand communication; constructing a privacy protection mechanism and a noise design method; updating the local state based on received information; and achieving collaborative system operation while satisfying privacy constraints. Compared with existing technologies, this invention effectively ensures stable system operation and collaborative performance while reducing communication frequency and bandwidth consumption, making it suitable for scenarios such as air-to-ground collaborative sensing, collaborative inspection, target tracking, and emergency search and rescue.

[0071] Example 2 This embodiment 2 describes a collaborative control system for UAVs and ground vehicles under privacy protection and on-demand communication. This system is based on the same inventive concept as the collaborative control method for UAVs and ground vehicles under privacy protection and on-demand communication in embodiment 1 above.

[0072] A collaborative control system for drones and ground vehicles, featuring privacy protection and on-demand communication, includes the following modules: The model building module is used to build a multi-agent collaborative system model consisting of UAVs and ground vehicles, and to construct a communication topology diagram; The controller design module is used to design a collaborative controller for UAVs and ground vehicles based on privacy protection and on-demand communication. The specific process is as follows: First, based on a multi-agent cooperative system model, a distributed dynamic protocol is designed; Then, a privacy protection mechanism is designed to add random noise during data transmission; at the same time, an on-demand communication mechanism is established, which updates the trigger time and sends a reference status signal when the error of a local event exceeds a preset threshold, otherwise it remains silent. Finally, based on the received neighbor agent information and the local state update reference state signal, a distributed cooperative control law is designed. The UAV and ground vehicle collaborative control module is used to realize collaborative control between UAVs and ground vehicles based on a privacy-preserving and on-demand communication-based UAV and ground vehicle collaborative controller.

[0073] It should be noted that any content not mentioned in the above-described functional modules of the system described in Embodiment 2 can be referred to the step description of the corresponding method in Embodiment 1 above, and will not be repeated in detail here.

[0074] Of course, the above description is only a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. It should be noted that any equivalent substitutions or obvious modifications made by those skilled in the art under the guidance of this specification fall within the scope of this specification and should be protected by the present invention.

Claims

1. A method for collaborative control of unmanned aerial vehicles and ground vehicles under privacy protection and on-demand communication, characterized in that, Includes the following steps: Step 1. Establish a multi-agent collaborative system model consisting of UAVs and ground vehicles, and construct a communication topology diagram; Consider a combination of drones and ground vehicles The collaborative system model of multiple agents, treating each agent as a node, has the following dynamic equation: ； in , For intelligent agents The state signal, whose initial state is denoted as . ; For intelligent agents Control input; For intelligent agents The measurement output; , , The dimensions of the system state, input, and output are represented respectively; matrices , , Dimensionality is compatible, and its value differs for different agents; Step 2. Design a collaborative controller for UAVs and ground vehicles based on privacy protection and on-demand communication. The specific process is as follows: First, based on the multi-agent cooperative system model established in step 1, a distributed dynamic protocol is designed. The designed distributed dynamic protocol is as follows: ； in and respectively intelligent agents and intelligent agents The reference state signal, , express initial value, ; For time-varying step size, Time-varying step size The following random approximation conditions must be met: ； ； ； in Represents the Laplace matrix, , It is an adjacency matrix; for The largest non-zero eigenvalue; for The largest eigenvalue; For degree matrix, , It is the sum of the weights of all edges in the connected component. Represents a diagonal matrix; Indicates time-varying step size The upper bound; Represents intelligent agents The neighborhood group, ; and It is the gain matrix, and ; , Represents the unknown matrix to be solved; matrix , The solution is obtained from formula (1); (1) in for 3D identity matrix; Then, a privacy protection mechanism is designed to add random noise during data transmission; at the same time, an on-demand communication mechanism is established, which updates the trigger time and sends a reference status signal when the error of a local event exceeds a preset threshold, otherwise it remains silent. The privacy protection mechanism specifically includes: Noise is added when exchanging information with neighboring agents, i.e., the agent... When communicating with neighboring agents, the true reference state signal is not transmitted. Instead, it transmits signals with random noise added. Its expression is as follows: ； in Represents Laplace noise, which follows a Laplace distribution. ; For time-varying decay noise scale parameters, , , Given parameters, Indicates the time step; , used to adjust noise intensity; The on-demand communication mechanism is as follows: Define discrete sampling time ; in For the global sampling period, , , Represent natural numbers; For intelligent agents Construct its local trigger time sequence ; in Represents intelligent agents The triggering time, ; Represents discrete sampling time; At discrete sampling time intelligent agent Local event error for: ； in For intelligent agents exist Reference state signal at time, For intelligent agents exist The reference state signal at any given time; the trigger condition is set as follows: ; in This indicates a search for the norm. The threshold is determined by a time-varying function; ； in Indicates an adjustable parameter; This indicates the signal after random noise has been added. exist The value at time, for Element; When the above triggering conditions are met, update the triggering time and send a reference status signal; otherwise, remain silent. Finally, based on the received neighbor agent information and the local state update reference state signal, a distributed cooperative control law is designed. Step 3. Based on the privacy-preserving and on-demand communication-based UAV-ground vehicle collaborative controller designed in Step 2, realize collaborative control of UAV and ground vehicle.

2. The method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication as described in claim 1, characterized in that, In step 1, Construct a communication topology graph This is used to describe the information interaction relationship between drones and ground vehicles; in express A collection of intelligent agents ; Denotes the set of edges. ; in Represents intelligent agents Received from intelligent agent Information; Represents the adjacency matrix; , for The element; when hour, ,otherwise .

3. The method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication as described in claim 2, characterized in that, In step 2, the designed distributed cooperative control law is as follows: ； The reference status signal is updated as follows: ； in Represents intelligent agents exist Reference state signal at time, express The time-varying step size at any given moment.

4. The method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication as described in claim 3, characterized in that, In step 2, after completing the design of the UAV and ground vehicle collaborative controller based on privacy protection and on-demand communication, convergence analysis and privacy analysis are performed on the UAV and ground vehicle collaborative system controlled by the controller.

5. The method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication as described in claim 4, characterized in that, In step 2, the specific process of convergence analysis is as follows: If for any initial conditions, for each agent The following relationships hold true for all of them: ； The system is said to have achieved average output consistency. in For intelligent agents Initial measurement output; If the system model and communication topology satisfy the following conditions: I. For each intelligent agent , It is controllable, and It is detectable; in Indicates the first The control pair consists of the system matrix and the input matrix of each agent. Indicates the first The detection pair consists of the output matrix of each agent and the system matrix; II. For each agent There exists a matrix and This satisfies the following discrete-time adjustment equation: ； III. When or At that time, noise and They are independent of each other; IV. Communication Topology Diagram It is strongly connected; and If the Schur matrix is ​​used, then under the distributed cooperative control law, the UAV-ground vehicle cooperative system can achieve mean-square output consistency, i.e.: ； in This represents the vector formed by the measurement outputs of all agents. A vector representing the reference state signals of all agents. Represents an N-dimensional column vector of all ones. Indicates the Kronecker product. This indicates a demand for expectation; , col() represents a column vector; Choosing Lyapunov functions for: ； in , express Reference state signal at any given time; ； in Represents an N-dimensional identity matrix; The Lyapunov function at the next moment for: ； Taking a one-step difference along the system trajectory of the above Lyapunov function and obtaining the expected value, we get: ； in for The mathematical expectation, for The mathematical expectation; Indicates the system at the 1st The convergence rate parameter at time t; This represents the additional perturbation term caused by noise, discretization error, and triggering error; and , ; Based on the random approximation condition, we obtain: ;Right now ; Because the reference state signal is implemented using piecewise constants, that is, for any ,have ; therefore ; in ; Therefore, we get: ； therefore: ; Based on the above convergence conditions, the UAV-ground vehicle collaborative system can achieve average output consistency in the mean square sense.

6. The method for collaborative control of UAVs and ground vehicles under privacy protection and on-demand communication as described in claim 4, characterized in that, In step 2, the specific process of privacy analysis is as follows: Define real numbers If there exists a positive integer , so that: ； This is called the initial state set. and It is a set of adjacent datasets, that is and The difference only in the initial state of an agent does not exceed ; in and Representing adjacent initial state sets respectively and The Middle Initial measurement output of each agent; definition If for any two adjacent initial state sets , and observation set They all Then it is called a mechanism. satisfy -Differential privacy; in This represents a random mechanism acting on the initial set of states; It is a constant; and These respectively represent when the input dataset is and Time mechanism Random output; and These respectively represent the input as and The random output of time mechanism M falls into the observation set The probability of; Define dataset and Observation set and noise sequences Given an initial dataset Under the conditions of noise sample paths, the system state trajectory is denoted as ,mechanism In time Sensitivity to time for: ； in Represents any time range, and supremum represents the supremum. and They represent in Given an initial dataset at any given time and The system state trajectory under the following conditions; Represents the L1 norm; express The signal after random noise has been added at any time. express Random noise at any given moment; Under the aforementioned method for collaborative control of UAVs and ground vehicles with privacy protection and on-demand communication, the mechanism The sensitivity requirement is met: ; in Indicating sensitivity The upper realm, Represents a product of products. The smallest eigenvalue of the degree matrix; Sensitivity represents neighboring datasets and In the same observation set The upper bound of the trajectory deviation generated reflects the degree of response of the output to changes in the input; Privacy noise is defined as independent Laplace noise, whose components are independent and identically distributed, satisfying the following conditions: ,mechanism Over any time range - Differential privacy, and its corresponding privacy budget for: ; Since the noise injected by each agent at each time step in step 2 is independent and follows a Laplace distribution, therefore, in a given dataset... At that time, the mechanism outputs a sequence joint probability density function for: ； in Represents intelligent agents Given an initial dataset The system state trajectory below, Represents the set of all agent-time index pairs. , This indicates the T-th sampling time. For any adjacent dataset ~ Using the triangle inequality, we get: ； in, and These represent the initial datasets, respectively. The system state trajectory and joint probability density function under the given conditions; Therefore, mechanism In the time interval The above satisfies - Differential privacy.

7. A collaborative control system for UAVs and ground vehicles under privacy protection and on-demand communication for implementing the collaborative control method for privacy protection and on-demand communication as described in claim 1, characterized in that, A collaborative control system for drones and ground vehicles, featuring privacy protection and on-demand communication, includes the following modules: The model building module is used to build a multi-agent collaborative system model consisting of UAVs and ground vehicles, and to construct a communication topology diagram; The controller design module is used to design a collaborative controller for UAVs and ground vehicles based on privacy protection and on-demand communication. The specific process is as follows: First, based on a multi-agent cooperative system model, a distributed dynamic protocol is designed; Then, a privacy protection mechanism is designed to add random noise during data transmission; at the same time, an on-demand communication mechanism is established, which updates the trigger time and sends a reference status signal when the error of a local event exceeds a preset threshold, otherwise it remains silent. Finally, based on the received neighbor agent information and the local state update reference state signal, a distributed cooperative control law is designed. The UAV and ground vehicle collaborative control module is used to realize collaborative control between UAVs and ground vehicles based on a privacy-preserving and on-demand communication-based UAV and ground vehicle collaborative controller.

Images