Target control method, device and equipment for cooperative task and medium
By constructing a target state matrix and dynamically adjusting strategies, and combining machine learning and heterogeneous communication, the problems of unreasonable motion planning and unstable communication in traditional target control methods under complex environments are solved, and efficient, flexible and stable collaborative task execution is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-08
- Publication Date
- 2026-03-31
AI Technical Summary
Traditional target control methods suffer from problems such as unreasonable motion planning, unstable communication, lack of flexibility and fault tolerance in complex environments and dynamic task scenarios, resulting in low task execution efficiency.
By constructing a target state matrix, dynamically adjusting motion parameters and cooperative strategies, optimizing the cooperative strategy library using machine learning algorithms, assigning roles by combining an improved Hungarian algorithm and a distributed computing framework, and setting up sliding mode fault-tolerant control laws and heterogeneous communication networks, real-time monitoring and rapid response can be achieved.
It improves the flexibility and efficiency of task execution, enhances the stability and fault tolerance of the system, and ensures rapid recovery of normal operation in abnormal situations.
Smart Images
Figure CN121764136A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of automation control technology, and in particular to a target control method, apparatus, device, and medium for collaborative tasks. Background Technology
[0002] With the rapid development of automation and artificial intelligence, the demand for multi-target systems to perform complex collaborative tasks is exploding. In numerous fields, such as intelligent transportation, industrial automation, and training applications, the efficiency and stability of multi-target system collaborative operations are crucial. However, traditional target control methods have revealed many insurmountable shortcomings when dealing with complex environments and dynamic task scenarios.
[0003] In multi-target collaborative tasks, traditional methods are prone to irrational motion planning. They often rely on static environmental information and pre-defined rules, failing to perceive and adapt to dynamic environmental changes in real time. This leads to target movement path conflicts, resource waste, and other problems, severely impacting task efficiency. Regarding communication, traditional methods suffer from poor stability in complex environments, making them highly susceptible to communication interruptions. Particularly in multi-target collaborative operations, the conflict between the demand for large amounts of data transmission and limited communication bandwidth becomes acute, hindering the timely and accurate transmission of critical information and significantly impeding the smooth operation of collaborative tasks.
[0004] Furthermore, traditional methods rely excessively on predefined rules, lacking the flexibility to handle unforeseen circumstances. When faced with sudden threats, such as unexpected enemy interference during operations or mission changes, traditional methods struggle to quickly adjust strategies, severely impacting the efficiency and effectiveness of collaborative operations. In heterogeneous network environments, traditional methods suffer from high data transmission latency, hindering timely synchronization of critical status updates. This leads to deviations in target coordination, making information distortion and misunderstandings during mission execution more likely, further reducing the reliability of collaborative operations. When actuators malfunction, existing systems lack rapid and effective compensation mechanisms, failing to quickly restore targets to normal operating status, which can easily trigger mission interruptions and failures, resulting in significant losses. Summary of the Invention
[0005] Purpose of the invention: The purpose of this invention is to provide a target control method, apparatus, device, and medium for collaborative tasks.
[0006] Technical solution: The target control method for collaborative tasks according to the present invention is characterized by comprising the following steps:
[0007] Obtain collaborative task instructions and target status information;
[0008] The collaborative task instructions are parsed and target information is extracted. The target information is then converted into a target vector, and a target state matrix is constructed based on the target vector.
[0009] The target's state information is integrated into a target state matrix according to priority for comprehensive analysis to determine whether the target state is performing the task according to the predetermined motion plan. If the target state deviates from the predetermined motion plan or the task scenario changes, the target's motion parameters and collaborative strategies are dynamically adjusted according to a preset collaborative control strategy library. The collaborative strategies include obstacle avoidance strategies, speed adjustment strategies, and target tracking strategies, which are continuously optimized and updated through machine learning algorithms to adapt to different task scenarios and emergencies. Otherwise, operation continues.
[0010] Furthermore, the target's state information is integrated into a target state matrix according to priority for comprehensive analysis, including:
[0011] The targets are initially assigned roles based on the improved Hungarian algorithm;
[0012] A distributed computing framework is used to perform an initial evaluation of the target's current state information.
[0013] Based on the results of the initial evaluation, a reinforcement learning model is used to perform a secondary evaluation on the matching degree of the initial role assignment. When the change in environmental threat exceeds a set threshold, the role switching is automatically triggered to adjust the target's role to adapt to new task requirements or environmental changes.
[0014] Furthermore, after continuous optimization and updates through machine learning algorithms, including:
[0015] Establish a heterogeneous communication network, and determine the applicable scenarios and switching mechanisms for different communication technologies based on the heterogeneous communication network;
[0016] Under the conditions of the applicable scenarios and switching mechanisms, the link quality is judged, an adaptive relay selection mechanism is established, a target relay target is dynamically selected, and a mesh communication topology is constructed based on the target relay target.
[0017] In the mesh communication topology, a TDMA / CSMA hybrid protocol is set up, and the communication cycle in the TDMA / CSMA hybrid protocol is divided into critical state broadcast time slots and non-critical data compression time slots to meet the communication requirements of the target collaborative control task.
[0018] Furthermore, the critical status broadcast time slots mainly transmit location data and threat data, wherein the location data is the spatial location of each target, and the threat data includes environmental threat information; the non-critical data compression time slots mainly transmit the target's device status information, auxiliary log data, etc., and perform double compression on the non-critical data.
[0019] Furthermore, based on a pre-defined collaborative control strategy library, the target's motion parameters and collaborative strategies are dynamically adjusted, including:
[0020] Construct a kinematic model, wherein the kinematic model is:
[0021]
[0022] in, Let be the rate of change of the position of target i. Let i be the velocity of target i. The direction of movement is the target i, and j represents the adjacent target. β is the synergistic gain coefficient, and β is the threat avoidance coefficient. The position of target i. Let j be the position of the target. Let i be the set of neighboring targets. The degree of environmental threat faced by target i This is a reference threshold.
[0023] Furthermore, based on a pre-defined collaborative control strategy library, dynamically adjusting the target's motion parameters and collaborative strategies also includes:
[0024] A sliding mode fault-tolerant control law is set up, and a fault diagnosis sensor is used to detect actuator deviation. When a deviation is detected in the actuator, a compensation term is automatically activated to maintain continued operation.
[0025] Furthermore, based on a pre-defined collaborative control strategy library, dynamically adjusting the target's motion parameters and collaborative strategies also includes:
[0026] By introducing transfer learning technology, optimization strategies from similar historical task scenarios can be quickly applied to the current task, shortening training time. Furthermore, adversarial training mechanisms are used to generate adversarial examples, enabling machine learning models to continuously learn and optimize in an adversarial environment.
[0027] The target control device for collaborative tasks according to the present invention includes:
[0028] The acquisition module is used to acquire collaborative task instructions and target status information;
[0029] The construction module is used to parse the collaborative task instructions and extract target information, convert the target information into a target vector, and construct a target state matrix based on the target vector;
[0030] The processing module is used to integrate the target's state information into a target state matrix according to priority for comprehensive analysis, and to determine whether the target state is performing the task according to the predetermined motion plan. If the target state deviates from the predetermined motion plan or the task scenario changes, the module dynamically adjusts the target's motion parameters and collaborative strategies according to a preset collaborative control strategy library. The collaborative strategies include obstacle avoidance strategies, speed adjustment strategies, and target tracking strategies, and are continuously optimized and updated through machine learning algorithms to adapt to different task scenarios and emergencies. Otherwise, the module continues to operate.
[0031] Beneficial Effects: Compared with existing technologies, this invention has the following significant advantages: By accurately parsing collaborative task instructions and constructing a target state matrix, this invention achieves dynamic task planning, avoiding path conflicts and resource waste caused by static planning. Simultaneously, by integrating target state information for real-time monitoring, it dynamically adjusts according to the collaborative control strategy library as soon as deviations or changes in the task scenario are detected, enhancing flexibility and adaptability. Utilizing machine learning algorithms to continuously optimize the collaborative strategy library and intelligently select the optimal strategy improves collaborative operation efficiency. Furthermore, through dynamic adjustment and intelligent optimization, it can quickly restore normal target operation in abnormal situations, significantly improving system stability and fault tolerance. Attached Figure Description
[0032] Figure 1 A flowchart of a target control method for collaborative tasks;
[0033] Figure 2 This is a schematic diagram of the drone's operation process;
[0034] Figure 3 This is a schematic diagram illustrating changes in location information.
[0035] Figure 4 This is a schematic diagram illustrating the initial stage of dynamic changes in the coordinates of a UAV.
[0036] Figure 5 This is a schematic diagram of the intermediate stage of the dynamic change of UAV coordinates;
[0037] Figure 6 This is a schematic diagram of the final stage of the dynamic change of UAV coordinates.
[0038] Figure 7 A flowchart of a target control method for collaborative tasks;
[0039] Figure 8Example block diagram of a target control device for collaborative tasks;
[0040] Figure 9 This is a schematic diagram of the electronic device. Detailed Implementation
[0041] The technical solution of the present invention will be further described below with reference to the accompanying drawings.
[0042] like Figure 1 As shown, the target control method for collaborative tasks according to the present invention includes the following steps:
[0043] In step S101, the collaborative task instructions and target status information are obtained.
[0044] Among them, collaborative task instructions can be used to guide multiple targets to collaboratively complete a specific task. They typically include information such as the task's objectives, requirements, and time limits. The target's status information can refer to the target's real-time status data during the task execution process.
[0045] It is understood that the embodiments of this application provide a data foundation for subsequent dynamic adjustment and optimization by obtaining collaborative task instructions and target status information.
[0046] In step S102, the collaborative task instructions are parsed and target information is extracted. The target information is then converted into a target vector, and a target state matrix is constructed based on the target vector.
[0047] Among them, target information can be used to guide the movement and behavior of the target, the target vector can include multiple dimensions to represent different target parameters, and the target state matrix can be used to represent the current state of the target and its relationship with the target.
[0048] It is understood that this application embodiment, by parsing collaborative task instructions and extracting target information, further quantifies it into target vectors and constructs a target state matrix accordingly. This ensures an accurate understanding of the task objectives and requirements, providing clear guidance for subsequent operations. The quantification of target vectors enables more efficient data processing and analysis, while the target state matrix comprehensively reflects the real-time state of the target, not only improving task execution efficiency but also enhancing flexibility and adaptability, enabling it to cope with complex and ever-changing task environments. Simultaneously, by monitoring the target state in real time, resource allocation can be optimized, reducing energy consumption and costs.
[0049] like Figure 2As shown, in a forest fire rescue mission, multiple drones participated in the collaborative operation. Suppose the received collaborative mission instruction is: "Three drones depart from the temporary command base at 3:00 AM to conduct fire reconnaissance on the fire area located at coordinates (30, 50). After discovering the fire source, they guide the fire fighting forces to accurately extinguish it. During the mission, they need to avoid areas with dense smoke and high-voltage lines. After completing the mission, they return to the temporary command base."
[0050] Collaborative task instruction parsing and target information extraction:
[0051] Task initiated at 3:00 AM.
[0052] Departure point: Temporary command base.
[0053] Target location: The fire zone at coordinates (30, 50).
[0054] Mission objectives: Fire reconnaissance, guiding fire suppression, avoiding dangerous areas (areas with dense smoke and high-voltage lines), and returning to base.
[0055] Target information is transformed into a target vector:
[0056] Time information: Convert 3 AM to the number of seconds remaining until midnight of the current day, assuming it is 10800 seconds, represented by the vector
[10800] .
[0057] Location information: Assuming the coordinates of the temporary command base are (0, 0) and the coordinates of the target location are (30, 50), then the location vector is [(0, 0), (30, 50)].
[0058] Task content: Fire reconnaissance, fire suppression guidance, avoiding dangerous areas, and returning to base are coded separately. Fire reconnaissance is 001, fire suppression guidance is 002, avoiding dangerous areas is 003, and returning to base is 004. Therefore, the task content vector is [001, 002, 003, 004].
[0059] Therefore, the target vector is [10800,(0,0),(30,50),[001,002,003,004]].
[0060] As the task progresses, the status information in the matrix will be updated in real time. For example... Figure 4 , Figure 5 and Figure 6The diagram illustrates the positional changes of the drones relative to the temporary command base and target location at different times. When drone #1 detects a fire, the "Fire source detected?" message changes to "Yes," and the drone transmits the fire's location information to the command center, guiding firefighting forces to the affected area. If drone #2 encounters a dense smoke area, the "Danger area encountered?" message changes to "Yes," triggering "Avoidance mode." The drone automatically adjusts its flight path, and its current position coordinates in the matrix change in real time. Furthermore, as flight time increases, the remaining battery power gradually decreases. When the battery level drops to a certain point, the drone prioritizes returning to base. At this time, the task execution priority and related statuses in the matrix are adjusted accordingly to ensure successful mission completion.
[0061] In step S103, the target's state information is integrated into the target state matrix according to priority for comprehensive analysis to determine whether the target state is performing the task according to the predetermined motion plan. If the target state deviates from the predetermined motion plan or the task scenario changes, the target's motion parameters and collaborative strategies are dynamically adjusted according to the preset collaborative control strategy library, and continuously optimized and updated through machine learning algorithms to adapt to different task scenarios and emergencies; otherwise, operation continues.
[0062] The collaborative strategy can include obstacle avoidance strategy, speed adjustment strategy, and target tracking strategy. The obstacle avoidance strategy can be used to guide the target on how to avoid obstacles when it encounters them. The speed adjustment strategy can be used to adjust the target's movement speed according to task requirements and actual conditions. The target tracking strategy can be used to guide the target on how to accurately track and reach the designated target position.
[0063] It is understood that the embodiments of this application integrate target status information into a target status matrix and perform comprehensive analysis based on priority, thereby achieving comprehensive monitoring and accurate judgment of target status. Simultaneously, by dynamically adjusting and optimizing the collaborative control strategy, and by introducing machine learning algorithms for continuous optimization and updating, it can better adapt to different task scenarios and unforeseen circumstances, improving task execution efficiency, flexibility, and resource allocation efficiency, and enhancing decision-making quality.
[0064] In this embodiment, the target's state information is integrated into the target state matrix according to priority for comprehensive analysis, including: initial role assignment of the target based on the improved Hungarian algorithm; initial evaluation of the target's current state information using a distributed computing framework; and secondary evaluation of the matching degree of the initial role assignment based on the results of the initial evaluation using a reinforcement learning model. When the change in environmental threat exceeds a set threshold, role switching is automatically triggered to adjust the target's role to adapt to new task requirements or environmental changes.
[0065] Among them, the improved Hungarian algorithm can be an optimization based on the original Hungarian algorithm, which is particularly suitable for the initial role allocation problem of UAV targets to ensure optimal resource allocation; the distributed computing framework can be a software tool and library for implementing distributed computing, which improves the speed and efficiency of the initial evaluation of target state information by processing multiple small problems in parallel; and the reinforcement learning model uses the strategies learned by the agent in interacting with the environment to perform secondary evaluation and adjustment of target role allocation to improve matching degree and task execution efficiency.
[0066] It is understood that the embodiments of this application utilize an improved Hungarian algorithm to achieve efficient and accurate matching of UAV targets and task roles. A distributed computing framework processes target status information in real time, providing data support for subsequent dynamic role allocation based on a reinforcement learning model. When environmental changes exceed a set threshold, a role switching mechanism is automatically triggered, ensuring that the target can quickly adapt to new tasks or environmental changes. This not only improves the overall efficiency of task execution, enhances flexibility and the rationality of resource allocation, but also improves the accuracy of decision-making and the robustness of the system, enabling it to better cope with complex and ever-changing real-world environments and unexpected situations.
[0067] In this embodiment, the motion parameters and collaborative strategies of the target are dynamically adjusted according to a preset collaborative control strategy library, including: constructing a kinematic model;
[0068] The operational model is as follows:
[0069]
[0070] in, Let be the rate of change of the position of target i. Let i be the velocity of target i. The direction of movement is the target i, and j represents the adjacent target. β is the synergistic gain coefficient, and β is the threat avoidance coefficient. The position of target i. Let j be the position of the target. Let i be the set of neighboring targets. The degree of environmental threat faced by target i This is a reference threshold.
[0071] In this embodiment, the motion parameters and collaborative strategies of the target are dynamically adjusted according to a preset collaborative control strategy library. The method also includes setting a sliding mode fault-tolerant control law, using a fault diagnosis sensor to detect actuator deviation, and automatically activating a compensation term when an actuator deviation is detected to maintain continued operation.
[0072] Understandably, this application achieves real-time monitoring and fault response of the actuator status by integrating fault diagnosis sensors, sliding mode fault-tolerant control laws, and automatic compensation mechanisms. The fault diagnosis sensors can quickly detect actuator deviations and issue alarms, while the sliding mode fault-tolerant control laws ensure stable operation through strategy adjustments during faults, demonstrating strong robustness and rapid response capabilities. Simultaneously, the automatic compensation mechanism is automatically activated upon detecting deviations, compensating for the impact of faults by adjusting control parameters or actuator outputs, ensuring continuous normal operation, improving stability and reliability, reducing maintenance costs, and effectively enhancing safety and task execution capabilities.
[0073] Specifically, suppose that during flight, a drone relies on servos on its wings to adjust its attitude and direction. The servos are the actuators, and the fault diagnosis sensors can be various sensors installed inside the servos or on the connecting circuitry, such as current sensors and angle sensors. The current sensor can monitor changes in current during servo operation, while the angle sensor can provide real-time feedback on the servo's rotation angle.
[0074] While UAV 1 was en route to the target location (30, 50) for reconnaissance, it suddenly encountered strong airflow interference. At this time, the fault diagnosis sensor detected a sudden increase in the servo current controlling the wing direction of UAV 1, and the actual rotation angle of the servo deviated from the preset angle. For example, the preset rotation was 10 degrees to maintain a stable flight attitude, but it only rotated 8 degrees in reality.
[0075] Once this actuator deviation is detected, the compensation term in the sliding mode fault-tolerant control law will be automatically activated, and the additional control quantity that needs to be applied will be quickly calculated. By increasing the control voltage on the servo, the servo will overcome strong airflow interference and get as close as possible to the preset rotation angle, so that the UAV 1 can continue to perform the mission according to the predetermined motion plan.
[0076] In this embodiment of the application, the motion parameters and collaborative strategies of the target are dynamically adjusted according to the preset collaborative control strategy library. It also includes: introducing transfer learning technology to quickly apply the optimization strategies in similar historical task scenarios to the current task, shortening the training time, and using adversarial training mechanism to generate adversarial samples, so that the machine learning model can continuously learn and optimize in the adversarial environment.
[0077] Understandably, the transfer learning approach in this application enables the model to quickly adapt to new tasks and significantly shortens training time by transferring knowledge, reducing data dependence, and improving learning efficiency. Simultaneously, it helps the model achieve better performance on the target task by leveraging useful features from the source task. The adversarial training mechanism, on the other hand, enhances the model's robustness and generalization ability, ensuring stable performance even when faced with input perturbations or malicious attacks, thereby strengthening the model's security. Furthermore, the combination of these two techniques reduces model maintenance costs and minimizes the need for manual intervention and model adjustments.
[0078] In this embodiment of the application, after continuous optimization and updating through machine learning algorithms, the process includes: establishing a heterogeneous communication network; determining the applicable scenarios and switching mechanisms for different communication technologies based on the heterogeneous communication network; judging link quality under the conditions of the applicable scenarios and switching mechanisms; establishing an adaptive relay selection mechanism; dynamically selecting target relay targets; and constructing a mesh communication topology based on the target relay targets. In the mesh communication topology, a TDMA / CSMA hybrid protocol is set, and the communication cycle in the TDMA / CSMA hybrid protocol is divided into critical state broadcast time slots and non-critical data compression time slots to meet the communication requirements of the target collaborative control task.
[0079] The critical status broadcast time slots mainly transmit location data and threat data. The location data includes the spatial location of each target, and the threat data includes environmental threat information. The non-critical data compression time slots mainly transmit the target's device status information, auxiliary log data, etc., and perform double compression on the non-critical data.
[0080] It is understood that the heterogeneous communication network in this application embodiment significantly improves the flexibility and adaptability of communication by integrating multiple communication technologies, enabling the flexible selection of the most suitable communication technology according to actual needs. Simultaneously, the introduction of adaptive relay selection mechanisms and mesh communication topologies enhances network reliability and stability, reduces the probability of data transmission interruptions, and ensures network connectivity and data integrity. Furthermore, the TDMA / CSMA hybrid protocol and dynamic resource allocation strategy optimize the utilization and allocation of communication resources, improve channel utilization and data transmission efficiency, reduce data transmission latency and jitter, enhance network response speed and stability, and significantly improve network throughput and overall performance, thereby better meeting the growing data transmission demands and diverse application scenarios.
[0081] In the exercise scenario, multiple drones collaborated on tasks, requiring the construction of a heterogeneous communication network to meet communication demands. The network encompasses three communication technologies: satellite, 4G / LTE, and Bluetooth, suitable for long-distance, high-volume data transmission under base station coverage, and short-range low-volume data transmission scenarios, respectively, and includes corresponding automatic switching mechanisms. Link quality is assessed by monitoring signal strength and bit error rate. When link quality is poor, drones can select a relay based on the location of surrounding drones and communication load. Once a relay is selected, a mesh communication topology is formed among the drones, enhancing communication reliability and flexibility. Furthermore, a TDMA / CSMA hybrid protocol is implemented, allocating critical status broadcast slots and non-critical data compression slots to transmit critical location and threat data versus non-critical equipment status data, thereby improving communication efficiency and meeting the communication requirements of collaborative control tasks.
[0082] The target control method for collaborative tasks proposed in this application achieves dynamic task planning by accurately parsing collaborative task instructions and constructing a target state matrix, avoiding path conflicts and resource waste caused by static planning. Simultaneously, it integrates target state information for real-time monitoring; once deviations or changes in the task scenario are detected, it immediately adjusts dynamically according to the collaborative control strategy library, enhancing flexibility and adaptability. The collaborative strategy library is continuously optimized using machine learning algorithms, intelligently selecting the optimal strategy and improving collaborative operation efficiency. Furthermore, through dynamic adjustment and intelligent optimization, the system can quickly restore normal target operation in abnormal situations, significantly improving system stability and fault tolerance. Therefore, it solves the problems of existing technologies lacking flexibility, real-time performance, and stability in complex environments and dynamic tasks, leading to unreasonable motion planning, unstable communication, weak ability to handle emergencies, and poor fault tolerance.
[0083] The target control method for cooperative tasks will be illustrated below through a specific embodiment, such as... Figure 7 As shown, after a sudden earthquake disaster in a certain area, emergency rescue operations were quickly launched. In order to efficiently carry out the search and rescue mission, five drones took off quickly from the temporary command center located at coordinates (0,0). Their target was the severely affected area, the center of which was located at coordinates (100,150) with a radius of 5 kilometers.
[0084] These drones are equipped with advanced heat source signal identification systems. They dart through the air, searching in real time for potential survivors. Whenever a suspected heat source signal is detected, the drone immediately conducts a detailed scan to ensure that no chance of survival is missed.
[0085] Meanwhile, drones also bear the important responsibility of transmitting high-definition disaster images to ground rescue teams. They capture images of the disaster site from different angles and heights, transmitting real-time footage back to the command center to provide rescue teams with detailed disaster information and help them formulate more accurate rescue plans.
[0086] During the flight, the drones demonstrated exceptional obstacle avoidance capabilities. Faced with dynamic obstacles such as collapsed buildings during the earthquake, they were able to quickly adjust their flight paths to ensure safe flight. This not only guaranteed the drones' own safety but also provided strong support for the successful completion of the rescue mission. The five drones efficiently completed all their assigned tasks within two hours and then returned to the temporary command center along the predetermined route.
[0087] Step 1: Task Instruction Parsing and Target Vector Construction. The task instructions are precisely parsed, specifying a task duration of 2 hours (7200 seconds), a starting point of (0,0), a target area center of (100,150), and a radius of 5 kilometers. Simultaneously, the task is encoded as: searching for survivors (code 101), transmitting imagery (code 102), obstacle avoidance (code 103), and returning to base (code 104), constructing a clear and specific target vector.
[0088] Step 2: Construct a target state matrix to provide data support for subsequent task execution.
[0089] Step 3: Dynamic Role Assignment and Communication Optimization. An improved Hungarian algorithm was used for role assignment, designating UAV1 and UAV2 as the main reconnaissance units, equipped with thermal imagers for priority searching; UAV3 and UAV4 served as communication relays, equipped with satellite links for image transmission; and UAV5 provided dynamic support, ready to fill in at any time. Simultaneously, a heterogeneous communication network was configured, including satellite communication, 4G / LTE, and a self-organizing mesh network, as well as a TDMA / CSMA hybrid protocol, to ensure uninterrupted communication.
[0090] Step Four: Task Execution and Dynamic Adjustment. During execution, the UAV can make dynamic adjustments based on the actual situation. For example, when UAV1 detects a sudden obstacle such as a collapsed high-rise building ahead while flying at (40,60), it can quickly trigger an obstacle avoidance strategy and correct its detour path; when UAV3 experiences a 4G link interruption due to electromagnetic interference, it can adaptively select UAV4 as a new relay to build a mesh topology and switch to satellite communication.
[0091] Step 5: Application of sliding mode fault-tolerant control in actuator failure. When the UAV2 encounters strong crosswinds during a search, causing abnormal servo current, it can quickly diagnose the fault and activate the sliding mode fault-tolerant control law for compensation, ensuring stable flight of the UAV.
[0092] Step Six: Optimization of Transfer Learning and Adversarial Training. By using historical earthquake rescue mission data as the source domain for transfer learning, the training time was greatly shortened; at the same time, adversarial examples simulating random building collapses were generated for adversarial training, which significantly enhanced the robustness of the model.
[0093] In summary, the five drones successfully completed the survivor search and image transmission mission within 1 hour and 50 minutes; they successfully avoided obstacles 23 times and switched communication links 5 times during the flight, with no mission interruption throughout the entire process; thanks to optimization measures such as dynamic role allocation and data compression, the overall energy consumption was reduced by 18%.
[0094] Next, referring to the accompanying drawings, a target control device for collaborative tasks is described according to an embodiment of this application.
[0095] Figure 8 This is a block diagram of a target control device for collaborative tasks according to an embodiment of this application.
[0096] like Figure 8 As shown, the target control device 10 for collaborative tasks includes: an acquisition module 100, a construction module 200, and a processing module 300.
[0097] The acquisition module 100 is used to acquire the collaborative task instructions and the target's status information; the construction module 200 is used to parse the collaborative task instructions and extract the target information, convert the target information into a target vector, and construct a target status matrix based on the target vector; the processing module 300 is used to integrate the target's status information into the target status matrix according to priority for comprehensive analysis, and determine whether the target status is executing the task according to the predetermined motion plan. If the target status deviates from the predetermined motion plan or the task scenario changes, the motion parameters and collaborative strategies of the target are dynamically adjusted according to the preset collaborative control strategy library. The collaborative strategies include obstacle avoidance strategy, speed adjustment strategy, and target tracking strategy, and are continuously optimized and updated through machine learning algorithms to adapt to different task scenarios and emergencies. Otherwise, the operation continues.
[0098] It should be noted that the foregoing explanation of the target control method embodiment for collaborative tasks also applies to the target control device for collaborative tasks in this embodiment, and will not be repeated here.
[0099] The target control device for collaborative tasks proposed in this application achieves dynamic task planning by accurately parsing collaborative task instructions and constructing a target state matrix, avoiding path conflicts and resource waste caused by static planning. Simultaneously, it integrates target state information for real-time monitoring; once deviations or changes in the task scenario are detected, it immediately adjusts dynamically according to the collaborative control strategy library, enhancing flexibility and adaptability. The collaborative strategy library is continuously optimized using machine learning algorithms, intelligently selecting the optimal strategy and improving collaborative operation efficiency. Furthermore, through dynamic adjustment and intelligent optimization, the system can quickly restore normal target operation in abnormal situations, significantly improving system stability and fault tolerance. Therefore, it solves the problems of existing technologies lacking flexibility, real-time performance, and stability in complex environments and dynamic tasks, leading to unreasonable motion planning, unstable communication, weak ability to handle emergencies, and poor fault tolerance.
[0100] like Figure 9 The diagram shown is a structural schematic of an electronic device provided in an embodiment of this application. The electronic device includes:
[0101] The memory 901, the processor 902, and the computer program stored on the memory 901 and capable of running on the processor 902.
[0102] When the processor 902 executes the program, it implements the target control method for collaborative tasks provided in the above embodiments.
[0103] Furthermore, electronic devices also include:
[0104] Communication interface 903 is used for communication between memory 901 and processor 902.
[0105] The memory 901 is used to store computer programs that can run on the processor 902.
[0106] The memory 901 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.
[0107] If the memory 901, processor 902, and communication interface 903 are implemented independently, then the communication interface 903, memory 901, and processor 902 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 9 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.
[0108] Optionally, in a specific implementation, if the memory 901, processor 902, and communication interface 903 are integrated on a single chip, then the memory 901, processor 902, and communication interface 903 can communicate with each other through an internal interface.
[0109] The processor 902 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.
[0110] This application also provides a computer-readable storage medium storing a computer program or instructions thereon, which, when executed by a processor, implements the target control method for cooperative tasks as described above.
Claims
1. A target control method for cooperative tasks, characterized in that, Includes the following steps: Obtain collaborative task instructions and target status information; The collaborative task instructions are parsed and target information is extracted. The target information is then converted into a target vector, and a target state matrix is constructed based on the target vector. The target's state information is integrated into a target state matrix according to priority for comprehensive analysis to determine whether the target state is performing the task according to the predetermined motion plan. If the target state deviates from the predetermined motion plan or the task scenario changes, the target's motion parameters and collaborative strategies are dynamically adjusted according to a preset collaborative control strategy library. The collaborative strategies include obstacle avoidance strategies, speed adjustment strategies, and target tracking strategies, which are continuously optimized and updated through machine learning algorithms to adapt to different task scenarios and emergencies. Otherwise, operation continues.
2. The target control method for cooperative tasks according to claim 1, characterized in that, The target's state information is integrated into a target state matrix according to priority for comprehensive analysis, including: The targets are initially assigned roles based on the improved Hungarian algorithm; A distributed computing framework is used to perform an initial evaluation of the target's current state information. Based on the results of the initial evaluation, a reinforcement learning model is used to perform a secondary evaluation on the matching degree of the initial role assignment. When the change in environmental threat exceeds a set threshold, the role is automatically switched to adjust the target's role to adapt to new task requirements or environmental changes.
3. The target control method for cooperative tasks according to claim 1, characterized in that, After continuous optimization and updates through machine learning algorithms, including: Establish a heterogeneous communication network, and determine the applicable scenarios and switching mechanisms for different communication technologies based on the heterogeneous communication network; Under the conditions of the applicable scenarios and switching mechanisms, the link quality is judged, an adaptive relay selection mechanism is established, a target relay target is dynamically selected, and a mesh communication topology is constructed based on the target relay target. In the mesh communication topology, a TDMA / CSMA hybrid protocol is set up, and the communication cycle in the TDMA / CSMA hybrid protocol is divided into critical state broadcast time slots and non-critical data compression time slots to meet the communication requirements of the target collaborative control task.
4. The target control method for cooperative tasks according to claim 3, characterized in that, The critical status broadcast time slots mainly transmit location data and threat data, wherein the location data is the spatial location of each target, and the threat data includes environmental threat information; the non-critical data compression time slots mainly transmit the target's device status information, auxiliary log data, etc., and perform double compression on the non-critical data.
5. The target control method for cooperative tasks according to claim 1, characterized in that, Based on a pre-defined collaborative control strategy library, the target's motion parameters and collaborative strategies are dynamically adjusted, including: Construct a kinematic model, wherein the kinematic model is: in, Let be the rate of change of the position of target i. Let i be the velocity of target i. The direction of movement is the target i, and j represents the adjacent target. β is the synergistic gain coefficient, and β is the threat avoidance coefficient. The position of target i. Let j be the position of the target. Let i be the set of neighboring targets. The degree of environmental threat faced by target i This is a reference threshold.
6. The target control method for cooperative tasks according to claim 5, characterized in that, Based on a pre-defined collaborative control strategy library, the target's motion parameters and collaborative strategies are dynamically adjusted, including: A sliding mode fault-tolerant control law is set up, and a fault diagnosis sensor is used to detect actuator deviation. When a deviation is detected in the actuator, a compensation term is automatically activated to maintain continued operation.
7. The target control method for cooperative tasks according to claim 6, characterized in that, Based on a pre-defined collaborative control strategy library, the target's motion parameters and collaborative strategies are dynamically adjusted, including: By introducing transfer learning technology, optimization strategies from similar historical task scenarios can be quickly applied to the current task, shortening training time. Furthermore, adversarial training mechanisms are used to generate adversarial examples, enabling machine learning models to continuously learn and optimize in an adversarial environment.
8. A target control device for collaborative tasks, characterized in that, include: The acquisition module is used to acquire collaborative task instructions and target status information; The construction module is used to parse the collaborative task instructions and extract target information, convert the target information into a target vector, and construct a target state matrix based on the target vector; The processing module is used to integrate the target's state information into a target state matrix according to priority for comprehensive analysis, and to determine whether the target state is performing the task according to the predetermined motion plan. If the target state deviates from the predetermined motion plan or the task scenario changes, the module dynamically adjusts the target's motion parameters and collaborative strategies according to a preset collaborative control strategy library. The collaborative strategies include obstacle avoidance strategies, speed adjustment strategies, and target tracking strategies, and are continuously optimized and updated through machine learning algorithms to adapt to different task scenarios and emergencies. Otherwise, the module continues to operate.
9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the target control method for a cooperative task as described in any one of claims 1-7.
10. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by the processor, they are used to implement the target control method for cooperative tasks as described in any one of claims 1-7.