Unmanned aerial vehicle swarm routing system fusing blockchain macro task and micro decision instruction

Abstract

This invention discloses a drone swarm navigation system that integrates macro-level tasks and micro-level decision-making instructions from blockchain. Under the premise that blockchain supports decentralized task distribution, it enables the distribution of macro-level task instructions from blockchain nodes to drones, forming micro-level decision-making instructions for drone swarm navigation. It includes: a task software platform containing blockchain nodes for issuing macro-level task instructions to each drone in the swarm, and an observation information calculation module for calculating observation information for each drone; and a drone software system for generating and executing micro-level decision-making instructions based on macro-level instruction information and drone status information. Operators can issue macro-level navigation instructions to each drone in the swarm via the blockchain. The drone software system receives the observation information integrating macro-level instructions and drone swarm status through the task software platform, then uses a reinforcement learning model to generate and execute all the micro-level decision-making instructions needed to complete the macro-level instructions, ultimately completing the navigation task.

Description

Technical Field

[0001] This invention relates to a drone swarm navigation system that integrates macro-level tasks and micro-level decision-making instructions from blockchain, belonging to the field of computer software technology. Background Technology

[0002] In recent years, extensive research has been conducted both domestically and internationally on the autonomous collaborative task execution problem of UAV swarms. However, this technology is still in the algorithm research and technical verification stage, and there is still a considerable distance to go before practical application. Faced with the pathfinding problem of a UAV swarm, the traditional method involves using a central rule system to collect the status information of each UAV in real time, then calculating the next action command for each UAV based on its rules, and finally sending each action command to each UAV in the swarm for execution. This method makes the central rule system a computational bottleneck, and the quality of its rules directly affects the execution effect of the pathfinding task, often resulting in a high probability of task failure. Summary of the Invention

[0003] Purpose of the invention: To address the problems and shortcomings of existing technologies, this invention provides a system suitable for drone swarms to perform pathfinding tasks. Specifically, it is a drone swarm pathfinding system that integrates macro-level tasks and micro-level decision-making instructions from blockchain. The decision-making authority is delegated to individual drones. The upper layer only needs to transmit macro-level task instructions, while the lower-level drones generate micro-level decision-making instructions in real time based on status information and macro-level task instructions through a trained intelligent model to complete the pathfinding task.

[0004] This invention effectively ensures the security, efficiency, and intelligence of UAV pathfinding missions. Distributing macro-level task instructions via blockchain leverages the distributed, traceable, and tamper-proof characteristics of blockchain to guarantee security during the distribution process. Breaking down macro-level task instructions into micro-level decision instructions generates a large amount of data; off-chain transmission of this data improves system efficiency and ensures the real-time performance of subsequent reinforcement learning pathfinding model decisions. After macro-level task instructions are issued from above, the micro-level decision instructions to be executed on the UAVs are generated by the reinforcement learning pathfinding model, which helps reduce the probability of damage to individual UAVs during mission execution and increases the overall probability of successful execution of macro-level task instructions.

[0005] Technical solution: A drone swarm navigation system integrating macro-level tasks and micro-level decision-making instructions from blockchain, comprising: The physical hardware platform includes a physical drone and an external positioning sensor system mounted on it.

[0006] The task software platform includes a blockchain node for issuing and distributing macro-level task instructions to each drone in the swarm, and an observation information Obs calculation module for comprehensively analyzing the drone swarm's State information and calculating the unique observation information Obs for each drone.

[0007] The UAV software system includes a message interaction module, mainly used for communicating with the mission software platform to send state information (State) and acquire observation information (Obs); a policy module, mainly used to encapsulate the trained reinforcement learning pathfinding model and output micro-decision instructions (Cmd) based on the observation information (Obs); a UAV module, mainly used to encapsulate the mobility capabilities of different types of UAVs, provide specific implementations for the micro-decision instructions (Cmd) according to its programming interface, and read the UAV's state information (State) in real time; and a state processing module, used to determine the UAV's motion state based on the observation information (Obs) to decide whether to output a landing command.

[0008] A mission software platform and all the UAV software systems that communicate with it form a mission cluster. Multiple mission clusters in the entire UAV swarm issue macro commands through a blockchain composed of blockchain nodes in the mission software platform.

[0009] Furthermore, the UAV software system runs on the onboard computer of the physical UAV, interacting in real time with the mission software platform running on the ground station for state information (State) and observation information (Obs). Based on the macro-level mission instructions issued by the blockchain nodes, it continuously generates and executes micro-level decision instructions (Cmd) through the onboard reinforcement learning pathfinding model, ultimately completing the pathfinding task. The task steps are as follows: Step 1) Each drone in the drone swarm connects to the task software platform of its respective task cluster and obtains its own macro-level task instructions from the blockchain node. The instructions contain the flight target points for each drone in the cluster for this mission. The swarm also initializes the observation information.

[0010] Step 2) Each drone in the drone swarm obtains its own drone module's State information and sends it to the task software platform.

[0011] Step 3) The task software platform integrates the state information (State) of each UAV in the UAV swarm, calculates the observation information (Obs) for each UAV, and sends it to the message interaction module of each UAV.

[0012] Step 4) Each drone in the drone swarm sends its observation information (Obs) to the status processing module to determine its current flight status, including whether the macro-task has been completed and whether there is an obstacle collision. If any of these states are present, proceed to step 7.

[0013] Step 5) Each drone in the drone swarm sends the observation information Obs into the strategy module to calculate the micro-decision instruction Cmd that needs to be executed this time.

[0014] Step 6) Each drone in the drone swarm sends the micro-decision command Cmd into the drone module for execution, and the drone begins a new movement. After 1 second of unit command execution time, the process jumps back to step 2).

[0015] Step 7) After the drone completes the macro-level mission instructions or collides with an obstacle, the drone executes the landing instruction LandCmd in the micro-level decision instructions Cmd.

[0016] Furthermore, the physical drone is equipped with an external positioning sensor system, using UWB as the horizontal positioning sensor and LiDAR as the vertical positioning sensor.

[0017] Furthermore, blockchain nodes are used to issue macro-level mission instructions to each drone in the fleet. The content of each drone's macro-level mission instruction is the three-dimensional coordinates of the target point that the drone needs to reach for this mission.

[0018] Furthermore, the Observation Information Obs calculation module integrates the UAV swarm state information (State) in real time. The UAV swarm state information (State) includes the UAV's three-dimensional coordinates, three-dimensional velocity, three-dimensional acceleration, and attitude angle. It calculates the specific Observation Information (Obs) for each UAV, including: relative position to the target, UAV yaw angle direction, distance information from obstacles, and UAV flight status information.

[0019] Furthermore, the message interaction module is mainly used to define the format of the interface for the UAV and the mission software platform to send State information and receive Obs observation information, as well as to support the specific implementation of various communication protocols, including HTTP protocol, RPC protocol and message queue.

[0020] Furthermore, the strategy module encapsulates the trained reinforcement learning pathfinding model, whose input and output are both one-dimensional vectors. The strategy module sets up an abstract class `StrategyAgent` to represent the abstract pathfinding model. The abstract class `StrategyAgent` contains: an abstract method `transformObs` to process the observation information `Obs` input of the pathfinding model, transforming the observation information `Obs` into a one-dimensional vector input; an abstract method `chooseAction` encapsulating the trained pathfinding model, used to output its computational results; an abstract method `transformCmds` to process the one-dimensional vector output of the pathfinding model, transforming the one-dimensional vector output into a micro-decision instruction `Cmd` according to the reinforcement learning model's definition of the output result; and a concrete method `getCmds` to sequentially call the above three methods, outputting the micro-decision instruction `Cmd` that should be executed based on the current UAV observation information `Obs` after the pathfinding model's judgment. Each specific reinforcement learning pathfinding model inherits from the abstract pathfinding model class `StrategyAgent` and implements the corresponding abstract methods according to its own model processing requirements.

[0021] Furthermore, the micro-decision command Cmd represents the basic mobility capabilities of the UAV, and in this UAV swarm pathfinding system, it includes the takeoff command TakeoffCmd, the landing command LandCmd, the moveByPositionCmd command to fly to the target position, and the moveByVelocity command to fly at the target speed.

[0022] Furthermore, the UAV module encapsulates the ability of different types of UAVs to execute micro-decision commands (Cmd). The UAV module sets up an abstract class `Uxv`, which contains abstract methods `takeoff`, `land`, `moveByPosition`, and `moveByVelocity` to correspond to the execution methods of each micro-decision command (Cmd). These execution methods include `TakeoffCmd` (takeoff from the spot), `LandCmd` (landing from the spot), `MoveByPositionCmd` (flying towards the target location), and `MoveByVelocity` (flying at the target speed). It also includes concrete methods `executeCmds`, which take the micro-decision command (Cmd) as a method parameter and are used to select the corresponding micro-decision command (Cmd) to execute when called externally. Specific types of UAVs inherit from the abstract class `Uxv` and implement the specific micro-decision command (Cmd) execution methods according to their programming interfaces. The UAV module also needs to provide a method `getState` to return the UAV's state information (`State`).

[0023] Furthermore, the state processing module is used to determine the UAV's motion state based on the observation information Obs, including the determination that the macro-task has been completed and the determination of obstacle collision. If the above states occur, the Land command is output to control the UAV to land.

[0024] This invention offers the following advantages: The UAV swarm pathfinding system integrating macro-level tasks and micro-level decision-making instructions provided by this invention transmits macro-level task instructions within the UAV swarm via blockchain. The UAV software system employs a reinforcement learning pathfinding model to generate and execute micro-level decision-making instructions, thus completing the pathfinding task. This system endows UAV pathfinding tasks with security, high efficiency, and intelligence, and is compatible with various reinforcement learning algorithms with different implementations and specific UAVs with different control methods. It helps solve the problem of splitting and executing macro-level instructions in UAV swarm pathfinding. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the structure of the UAV swarm pathfinding system that integrates macro-level tasks and micro-level decision-making instructions from blockchain in this implementation example. Figure 2 This is a schematic diagram of the software platform and UAV software system modules for this implementation example. Figure 3 This is a sequence diagram showing the interaction between various modules in the UAV software system in this implementation example; Figure 4 This is a diagram showing the access to the external sensors used by the physical drone in this example. Detailed Implementation

[0026] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading the present invention, any modifications of the present invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0027] This implementation example uses the scenario of a drone swarm crossing a battlefield air defense zone as the basic scenario. When a drone reaches the target location specified by the macro command, it is considered that the mission has been successfully completed. Alternatively, if a drone collides with a battlefield obstacle, i.e., another drone or the air defense zone, it is considered that the mission has failed and a landing command (Land) will be triggered.

[0028] like Figure 1 As shown, this example demonstrates a drone swarm navigation system that integrates macro-level blockchain tasks and micro-level decision-making instructions, including: The physical hardware platform includes a physical drone and an external positioning sensor system mounted on it.

[0029] The mission software platform includes a blockchain node for issuing macro-level mission instructions to each drone in the swarm, and an observation information Obs calculation module for calculating the unique observation information Obs for each drone by integrating the drone swarm's state information (State).

[0030] The UAV software system includes a message interaction module, mainly used for communicating with the mission software platform to send state information (State) and acquire observation information (Obs); a strategy module, mainly used to encapsulate the pathfinding model trained through reinforcement learning and output micro-decision instructions (Cmd) based on the observation information (Obs); a UAV module, mainly used to encapsulate the mobility capabilities of different types of UAVs, provide specific implementations for the micro-decision instructions (Cmd) according to its programming interface, and read the UAV's state information (State) in real time; and a state processing module, used to determine the UAV's motion state based on the observation information (Obs) to decide whether to output a landing command.

[0031] A mission software platform and all the UAV software systems that communicate with it form a mission cluster. Multiple mission clusters in the entire UAV swarm issue macro commands through a blockchain composed of blockchain nodes in the mission software platform.

[0032] Furthermore, such as Figure 2 As shown, the UAV software system runs on the onboard computer of the physical UAV, interacting in real time with the mission software platform running on the ground station for state information (State) and observation information (Obs). Based on the macro-level mission instructions issued by the blockchain nodes, it continuously generates micro-level decision instructions (Cmd) through the onboard reinforcement learning pathfinding model and executes them, ultimately completing the pathfinding task. The task steps are as follows: Step 1) Each drone in the drone swarm connects to the mission software platform of its respective mission cluster and obtains its own macro-level mission instructions from the blockchain node. These instructions specify the flight target points for each drone within the cluster for this mission. Initialization operations are then performed, including setting initial target points, obstacle detection settings, and drone initial state settings. In this embodiment, the target points are the battlefield objective points that each drone needs to reach, and the obstacles are the battlefield air defense zones. The drone swarm commander assigns a target point to each drone in the mission cluster via the blockchain node.

[0033] Step 2) Each drone in the drone swarm obtains its own drone module's State information and sends it to the task software platform.

[0034] Step 3) The task software platform integrates the state information (State) of each UAV in the UAV swarm, calculates the observation information (Obs) for each UAV, and sends it to the message interaction module of each UAV.

[0035] Step 4) Each drone in the drone swarm sends its observation information (Obs) to the status processing module to determine its current flight status, including whether the macro-task has been completed and whether there is an obstacle collision. If any of these states are present, proceed to step 7.

[0036] Step 5) Each drone in the drone swarm sends the observation information Obs into the strategy module to calculate the micro-decision instruction Cmd that needs to be executed this time.

[0037] Step 6) Each drone in the drone swarm sends the micro-decision command Cmd into the drone module for execution, and the drone begins a new movement. After 1 second of unit command execution time, the process jumps back to step 2).

[0038] Step 7) After the drone completes the macro-level mission instructions or collides with an obstacle, the drone executes the landing instruction LandCmd in the micro-level decision instructions Cmd.

[0039] The execution status of each step within the drone software system is as follows: Figure 3 As shown.

[0040] Furthermore, such as Figure 4 As shown, the physical drone is equipped with an external positioning sensor system, using UWB as the horizontal positioning sensor and LiDAR as the vertical positioning sensor. In this example, four UWB base stations are arranged in a square shape for the UWB horizontal positioning sensor, and a UWB tag is attached to the drone as a positioning tag. The drone's horizontal coordinates are obtained by analyzing the distance between the positioning tag and the four base stations. The UWB tag is placed at the nose of the drone and connected via a UART serial port. The relative distance between the drone and the ground is calculated using the readings of the LiDAR vertical positioning sensor and the attitude angle readings of the drone's sensors, and this distance is used as the vertical coordinate. The LiDAR is placed on the belly of the drone, perpendicular to the ground, and connected via a UART serial port.

[0041] Furthermore, the blockchain nodes included in the mission software platform are used to issue macro-mission instructions to each drone in the fleet. The content of each drone's macro-mission instruction is the three-dimensional coordinates of the target point it needs to reach in this mission.

[0042] Furthermore, the observation information Obs calculation module integrates the UAV swarm state information (State) in real time, including the UAV's three-dimensional coordinates, three-dimensional velocity, three-dimensional acceleration, and attitude angle. It calculates the specific observation information (Obs) for each UAV, including its relative position to the target, yaw angle direction, distance information from obstacles, and flight status information.

[0043] Furthermore, the message interaction module is mainly used to define the format of the interface for the UAV and the mission software platform to send State information and receive Obs observation information, as well as to support the specific implementation of various communication protocols, including HTTP protocol, RPC protocol and message queue.

[0044] Furthermore, the strategy module encapsulates the trained reinforcement learning pathfinding model. The pathfinding model's input and output are both one-dimensional vectors. In this example, the input length is 14 bits, with 2 bits for the relative position to the target, 1 bit for the direction angle, and 11 bits for the distance information to obstacles in the eleven directions within a 180-degree arc. The output length is 1 bit, indicating the angular velocity at which the drone needs to change its current direction. The strategy module sets an abstract class `StrategyAgent` to represent the abstract pathfinding model. It includes the `transformObs` abstract method to process the observation information (Obs) input of the pathfinding model, converting it into a one-dimensional vector input; the `chooseAction` abstract method encapsulates the trained pathfinding model and outputs its calculation results; and the `transformCmds` abstract method processes the one-dimensional vector output of the pathfinding model, converting the one-dimensional vector into a vector. The output is transformed into a micro-decision instruction Cmd based on the reinforcement learning model's definition of the output result. In this example, the reinforcement learning model outputs the angular velocity of the next moment. Multiplying the angular velocity by the unit time and adding the current velocity azimuth angle yields the azimuth angle of the next moment. The UAV flies at a fixed speed of one meter per second. Multiplying the velocity value by the sine and cosine of the azimuth angle respectively yields the velocity components on each coordinate axis. Finally, the micro-decision instruction for flying at the target speed is obtained. It includes a specific method getCmds, which is used to sequentially call the above three methods. Based on the current observation information of the UAV, the Obs output is judged by the pathfinding model and outputs the micro-decision instruction Cmd that should be executed. Each specific reinforcement learning pathfinding model inherits the abstract pathfinding model class StrategyAgent and implements the corresponding abstract methods according to its own model processing requirements.

[0045] Furthermore, the micro-decision command Cmd represents the basic mobility capabilities of the UAV, which in this system includes the takeoff command TakeoffCmd, the landing command LandCmd, the moveByPositionCmd command to fly to the target position, and the moveByVelocity command to fly at the target speed.

[0046] Furthermore, the UAV module encapsulates the ability of different types of UAVs to execute micro-decision commands (Cmd). It sets up an abstract class `Uxv` containing abstract methods `takeoff`, `land`, `moveByPosition`, and `moveByVelocity` to execute each micro-decision command (Cmd), and a concrete method `executeCmds` that takes the micro-decision command (Cmd) as a parameter. When called externally, `executeCmds` selects the appropriate micro-decision command (Cmd) to execute. Specific UAV types inherit from the abstract class `Uxv` and implement their specific micro-decision command (Cmd) execution methods according to their programming interfaces. The UAV module also needs to provide a method `getState` to return the UAV's state information (`State`). In this example, the UAV uses a quadcopter with a PX4 flight controller. The onboard computer communicates directly with the flight controller using the MAVLink protocol, and the UAV's movement is controlled using MAVROS interfaces. Additionally, this example encapsulates a flight log module to record the UAV's position and speed information every 0.1 seconds during flight for reconstructing the UAV's flight trajectory.

[0047] Furthermore, the state processing module is used to determine the UAV's motion state based on the observation information Obs, including the determination that the macro-task has been completed and the determination of obstacle collision. If the above states occur, the Land command is output to control the UAV to land.

[0048] This invention helps to solve the problem of splitting and executing macroscopic commands in the pathfinding problem of UAV swarms.

Claims

1. A drone swarm navigation system integrating macro-level tasks and micro-level decision-making instructions from blockchain, characterized in that, include: The physical hardware platform includes a physical drone and an external positioning sensor system mounted on it; The task software platform includes a blockchain node for issuing and distributing macro-level task instructions to each drone in the swarm, and an observation information Obs calculation module for integrating the drone swarm's State information to calculate the unique observation information Obs for each drone. The UAV software system includes a message interaction module for communicating with the mission software platform to send state information (State) and acquire observation information (Obs); a policy module for encapsulating a trained reinforcement learning pathfinding model and outputting micro-decision instructions (Cmd) based on the observation information (Obs); a UAV module for encapsulating the mobility capabilities of different types of UAVs, providing specific implementations for the micro-decision instructions (Cmd) according to its programming interface, and reading the UAV's state information (State) in real time; and a state processing module for judging the UAV's motion state based on the observation information (Obs) to determine whether to output a landing command. A mission software platform and all the UAV software systems that communicate with it form a mission cluster. Multiple mission clusters in the entire UAV swarm issue macro commands through a blockchain composed of blockchain nodes in the mission software platform.

2. The UAV swarm navigation system integrating macro-level blockchain tasks and micro-level decision-making instructions as described in claim 1, characterized in that, The UAV software system runs on the onboard computer of the physical UAV, interacting in real time with the mission software platform running on the ground station for state information (State) and observation information (Obs). Based on the macro-level mission instructions issued by the blockchain nodes, it continuously generates and executes micro-level decision instructions (Cmd) through the onboard reinforcement learning pathfinding model, ultimately completing the pathfinding task. The task steps are as follows: Step 1) Each drone in the drone swarm connects to the task software platform of its respective task cluster and obtains its own macro-task instructions from the blockchain node. The instructions contain the flight target points of each drone in the cluster for this mission. And perform the initialization operation of the observation information; Step 2) Each drone in the drone swarm obtains its own drone module's State information and sends it to the task software platform; Step 3) The task software platform integrates the state information (State) of each UAV in the UAV swarm, calculates the observation information (Obs) for each UAV, and sends it to the message interaction module of each UAV. Step 4) Each drone in the drone swarm sends its observation information (Obs) to the status processing module to determine its current flight status, including whether the macro-task has been completed and whether there is an obstacle collision. If any of these conditions are met, proceed to step 7. Step 5) Each drone in the drone swarm sends its observation information Obs into the strategy module to calculate the micro-decision instruction Cmd that needs to be executed this time; Step 6) Each drone in the drone swarm sends the micro-decision command Cmd into the drone module for execution, and the drone performs a new movement; after the unit command execution set time, it jumps back to step 2); Step 7) After the drone completes the macro-level mission instructions or collides with an obstacle, the drone executes the landing instruction LandCmd in the micro-level decision instructions Cmd.

3. The UAV swarm navigation system integrating macro-level blockchain tasks and micro-level decision-making instructions as described in claim 1, characterized in that, The physical drone is equipped with an external positioning sensor system, using UWB as the horizontal positioning sensor and lidar as the vertical positioning sensor.

4. The UAV swarm navigation system integrating macro-level blockchain tasks and micro-level decision-making instructions as described in claim 1, characterized in that, Blockchain nodes are used to issue macro-level mission instructions to each drone in the fleet. The content of each drone's macro-level mission instruction is the three-dimensional coordinates of the target point that the drone needs to reach in this mission.

5. The UAV swarm navigation system integrating macro-level tasks and micro-level decision-making instructions based on blockchain as described in claim 1, characterized in that, The Observation Information Obs calculation module integrates the UAV swarm status information (State) in real time. The UAV swarm status information (State) includes the UAV's three-dimensional coordinates, three-dimensional velocity, three-dimensional acceleration, and attitude angle. It calculates the specific Observation Information Obs for each UAV, including: relative position to the target, UAV yaw angle direction, distance information from obstacles, and UAV flight status information.

6. The UAV swarm navigation system integrating macro-level tasks and micro-level decision-making instructions based on blockchain as described in claim 1, characterized in that, The message interaction module is used to define the format of the interface for UAVs and mission software platforms to send State information and receive observation information Obs, as well as the specific implementations of various communication protocols, including HTTP, RPC, and message queues.

7. The UAV swarm navigation system integrating macro-level tasks and micro-level decision-making instructions based on blockchain as described in claim 1, characterized in that, The strategy module encapsulates a pre-trained reinforcement learning pathfinding model, where the input and output are both one-dimensional vectors. The strategy module sets up an abstract class `StrategyAgent` to represent the abstract pathfinding model. The abstract class `StrategyAgent` contains: an abstract method `transformObs` to process the observation information (Obs) input of the pathfinding model, transforming it into a one-dimensional vector input; an abstract method `chooseAction` encapsulating the trained pathfinding model and outputting its computational results; an abstract method `transformCmds` to process the one-dimensional vector output of the pathfinding model, transforming it into a micro-decision instruction `Cmd` according to the reinforcement learning model's definition of the output result; and a concrete method `getCmds` to sequentially call the above three methods, outputting the micro-decision instruction `Cmd` that should be executed based on the current UAV observation information (Obs) after the pathfinding model's judgment. Each specific reinforcement learning pathfinding model inherits from the abstract pathfinding model class `StrategyAgent` and implements the corresponding abstract methods according to its own model processing requirements.

8. The UAV swarm navigation system integrating macro-level tasks and micro-level decision-making instructions based on blockchain as described in claim 1, characterized in that, The micro-decision command Cmd represents the basic mobility capabilities of the UAV. In this UAV swarm pathfinding system, it includes the takeoff command TakeoffCmd, the landing command LandCmd, the moveByPositionCmd command to fly to the target position, and the moveByVelocity command to fly at the target speed.

9. The UAV swarm navigation system integrating macro-level tasks and micro-level decision-making instructions based on blockchain as described in claim 1, characterized in that, The UAV module encapsulates the ability of different types of UAVs to execute micro-decision commands (Cmd). The UAV module defines an abstract class `Uxv`, which contains abstract methods `takeoff`, `land`, `moveByPosition`, and `moveByVelocity` to execute each micro-decision command (Cmd). It also includes a concrete method `executeCmds`, which takes the micro-decision command (Cmd) as a parameter and is used by external calls to select the appropriate micro-decision command (Cmd) to execute. Specific UAV types inherit from the abstract class `Uxv` and implement their own micro-decision command (Cmd) execution methods based on their programming interfaces. The UAV module also needs to provide a method `getState` to return the UAV's state information (State).

10. The UAV swarm navigation system integrating macro-level tasks and micro-level decision-making instructions based on blockchain as described in claim 1, characterized in that, The status processing module is used to determine the UAV's motion status based on the observation information Obs, including the determination of the completion of macro-tasks and obstacle collision. If the above statuses occur, a landing command (Land) is output to control the UAV to land.

Images