Unmanned aerial vehicle cluster cloud side-end collaborative navigation control decision-making method

Abstract

The invention discloses an unmanned aerial vehicle cluster cloud side-end collaborative navigation control decision-making method, and belongs to the technical field of positioning navigation and control. The method comprises the following steps: an unmanned aerial vehicle cluster performs a task at a low altitude, and a ground operator performs real-time monitoring through a ground station cloud background; detecting the communication rate between the ground station cloud background and the unmanned aerial vehicle cluster; establishing a multi-target particle swarm optimization algorithm navigation decision evaluation function and constraint conditions to obtain a navigation decision solution of the unmanned aerial vehicle cluster; the unmanned aerial vehicle cluster flies according to the navigation decision solution; the whole-network communication rate between the unmanned aerial vehicles is monitored in real time in the unmanned aerial vehicle cluster; the unmanned aerial vehicle cluster carries out inter-vehicle communication interaction and carries out navigation flight according to a task allocation result; and the unmanned aerial vehicle individuals in the cluster carry out navigation decision solution calculation and autonomous navigation flight. The method can be effectively applied to unmanned aerial vehicle cluster cooperative task execution scenes with complex physical and electromagnetic environments such as urban space in the future, and the problem of accurate, reliable and robust navigation control capability of an unmanned aerial vehicle platform is solved.

Description

Technical Field

[0001] The present invention belongs to the field of positioning, navigation and control technology, and in particular refers to a method for unmanned aerial vehicle cluster cloud-edge collaborative navigation control decision-making. Background Art

[0002] When drone swarms perform tasks such as logistics transportation, line inspection, and regional mapping in urban environments, they often face situations such as spatial occlusion and electromagnetic signal interference that lead to limited communications, which puts forward new requirements for drone swarm navigation and decision-making technology.

[0003] Traditional UAV navigation decision-making technologies mainly include three methods: background manual remote control, preset route autonomous flight, and sensor-based autonomous navigation control. Among them, manual remote control is the safest, but it is difficult to meet the large-capacity, fast and real-time requirements of cluster tasks, and this method will not be available under the condition of missing control links due to limited communication; the preset route autonomous flight method can be used in known static environments, but it will be difficult to achieve adaptive navigation decisions in unknown or dynamic environments; sensor-based autonomous navigation control can achieve autonomous navigation of individual UAVs, but there is currently a lack of cluster collaborative navigation decision-making methods, and the credibility of navigation decisions is not high. Summary of the invention

[0004] In view of this, the present invention provides a drone cluster cloud-edge-end collaborative navigation control decision method, which solves the control problem of drone clusters in complex communication scenarios and realizes the collaborative navigation decision-making capabilities of individual drones, clusters and background. The present invention can be applied to multi-drone mission scenarios such as regional inspections and logistics transportation to realize the collaborative navigation control capabilities of drone clusters.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for unmanned aerial vehicle cluster cloud-edge collaborative navigation control decision-making includes the following steps:

[0007] Step 1: The drone swarm performs its mission at low altitude, and the ground operator monitors the status of the drone swarm itself and its environment in real time through the ground station cloud background;

[0008] Step 2: Detect the communication rate between the ground station cloud background and the drone cluster. If the average communication rate between any drone in the cluster and the ground station cloud background is greater than the minimum situation real-time transmission requirement, execute step 3, otherwise go to step 5;

[0009] Step 3: The cloud backend of the ground station analyzes the situation information transmitted back by the drone cluster in real time, thereby establishing a navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm. The situation information includes the distribution of environmental obstacles, the drone's own motion state, and the positions of other drones in the cluster. The navigation decision solution is searched based on the multi-objective particle swarm optimization algorithm to obtain the navigation decision solution of the drone cluster.

[0010] Step 4: The ground station cloud backend sends the navigation decision solution to the drone cluster, which receives it and flies according to the navigation decision solution.

[0011] Step 5: The UAV cluster monitors the communication rate of the entire network between machines in real time. If the average communication rate of the entire network is greater than the minimum real-time interactive transmission requirement, execute step 6, otherwise go to step 7;

[0012] Step 6: The drone cluster communicates with each other and performs inter-cluster task allocation based on the distributed contract network algorithm. The drone cluster performs navigation flight according to the task allocation results.

[0013] In step 7, each drone in the cluster performs navigation decision solution calculations and performs autonomous navigation flight without relying on communication with other drones and the ground station cloud background.

[0014] Furthermore, the specific method of step 3 is:

[0015] Step 301, based on the situation information transmitted back by the drone cluster in real time, a navigation decision evaluation function and constraint conditions of a multi-objective particle swarm optimization algorithm are established, as shown in the following formula:

[0016] { y = minutes [ L P , R P , i P ] H p ≥ nH minutes n ( n − 1 ) l minutes ≤ l p ≤ n ( n − 1 ) l max

[0017] In the formula, is the UAV range constraint, is the turning radius constraint of the UAV, is the climbing angle constraint of the UAV cluster, is the UAV cluster flight altitude constraint, The distance constraint term between drone clusters together constitutes the navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm; is the number of swarm drones, The minimum flight altitude of the drone. is the minimum collision avoidance distance between aircraft, is the maximum communication distance between machines;

[0018] Step 302, initializing particle population parameters and external file parameters, wherein the particle population parameters include the population size , initial particle position and velocity, maximum number of iterations , convergence function value , external archive parameters include the maximum size of external archives , weighted dominance coefficient ;

[0019] Step 303, determining the Pareto dominance relationship between particles, and adding particles in a Pareto dominance position to an external archive;

[0020] Step 304, determine the weighted dominance relationship between the remaining particles according to the following formula, and add the particles in the weighted dominance position to the external file:

[0021] { ∀ i , f i ( x 1 ) − f i ( x 2 ) + r i ∑ i ≠ j 1 ... k [ f j ( x 1 ) − f j ( x 2 ) ] ≤ 0 ∃ i , f i ( x 1 ) − f i ( x 2 ) + r i ∑ i ≠ j 1 ... k [ f j ( x 1 ) − f j ( x 2 ) ] < 0

[0022] In the formula, is the evaluation function of the multi-objective particle swarm optimization algorithm, which is , and , is a feasible solution to the problem. If the above formula is satisfied, then Weighted Domination ;

[0023] Step 305: If the number of particles in the external file reaches the maximum size , then randomly remove particles that are in weighted dominance but not in Pareto dominance until the number of particles in the external archive is less than the maximum size;

[0024] Step 306, select the global optimal solution from the external archive, that is, the particle position that has the most weighted dominance over other particles, and replace and update the global extreme value in the position and velocity iteration formula of the standard particle swarm algorithm. The velocity position iteration formula of a single particle is as follows:

[0025]

[0026]

[0027] In the formula, and For the The particle speed and position after iterations, is the inertia weight, is the learning factor; For the The individual extreme value of a single particle after iterations, that is, the historical optimal position of a single particle; For the The global extreme value after iterations, which is replaced and updated by the global optimal solution selected from the external archive after each iteration;

[0028] Step 307, repeat steps 303 to 306 until the navigation decision evaluation function value reaches the convergence function value , or the iteration reaches the maximum number of iterations, and the optimized UAV cluster navigation decision solution is obtained.

[0029] Furthermore, the specific method of step 6 is:

[0030] Step 601, numbered The UAV senses the situation change during the mission and, as the bidder, transmits the bidding document containing the task bidding information and the price value according to the information transmission order with the neighboring drone when the step time arrives;

[0031] Step 602: The drone that receives the bid document conducts a situation assessment on the task and determines whether to participate in the bidding for the task. If it decides to participate in the bidding for the task, it calculates its own cost value based on the comprehensive performance evaluation index and transmits it to its neighboring drones when the next long time arrives. Otherwise, it only transmits the bid document when the next long time arrives. The calculation formula of the comprehensive performance evaluation index cost value is as follows:

[0032]

[0033] In the formula, They are environmental constraints, self-constraints, and task constraints. is the weight coefficient;

[0034] Step 603: After multiple information exchanges, each drone has learned the cost of the drones participating in the bidding, and the single or multiple drones with the lowest global cost will perform the task.

[0035] Furthermore, the specific method of step 7 is:

[0036] Step 701, the individual drone obtains its own status and observation information of the environment through the onboard sensor;

[0037] Step 703, according to the current observation information, the current linear velocity and angular velocity control amount are calculated through the proximal optimization strategy network, and sent to the drone onboard controller;

[0038] Step 702: All drones perform autonomous navigation control to achieve individual navigation decision-making effects in the current state.

[0039] The beneficial effects of the present invention compared with the prior art are:

[0040] (1) The present invention is aimed at the application of adaptive navigation decision-making and control in the actual flight of UAVs in complex physical and electromagnetic environments. It combines individual UAV nodes, UAV cluster systems and background cloud systems, and designs a UAV cluster cloud-edge collaborative navigation control decision-making architecture and method. It can provide UAVs with accurate, reliable and robust navigation decision-making capabilities according to different communication conditions during the execution of missions.

[0041] (2) The present invention adopts a multi-objective particle swarm optimization navigation decision algorithm based on an improved external archive update strategy, which can improve the accuracy of navigation decision solutions while improving search efficiency, meeting the cloud background's demand for high-performance navigation decision assistance for drone clusters.

[0042] (3) The present invention adopts a UAV cluster task allocation algorithm based on an improved distributed contract network, which can realize cluster task allocation capabilities without central nodes and background control, and meet the distributed collaborative navigation decision-making needs of UAV clusters. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the principle of the present invention.

[0044] Figure 2 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0046] A UAV cluster cloud-edge collaborative navigation control decision method, such as Figure 1 and 2 As shown, the following steps are included:

[0047] Step 1: The drone swarm performs its mission at low altitude, and the ground operator monitors the status of the drone swarm itself and its environment in real time through the ground station cloud background;

[0048] Step 2: Detect the communication rate between the ground station cloud background and the drone cluster. If the average communication rate between any drone in the cluster and the ground station cloud background is greater than the minimum situation real-time transmission requirement, execute step 3, otherwise go to step 5;

[0049] Step 3: The cloud backend of the ground station analyzes the situation information (including the distribution of environmental obstacles, the motion status of the drone itself, the positions of other drones in the cluster, etc.) transmitted back in real time by the drone cluster, so as to establish the navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm, search for the navigation decision solution based on the multi-objective particle swarm optimization algorithm, and obtain the navigation decision solution of the drone cluster; the specific method is:

[0050] Step 301, based on the situation information transmitted back by the drone cluster in real time, a navigation decision evaluation function and constraint conditions of a multi-objective particle swarm optimization algorithm are established, as shown in the following formula:

[0051] { y = minutes [ L P , R P , i P ] H p ≥ nH minutes n ( n − 1 ) l minutes ≤ l p ≤ n ( n − 1 ) l max

[0052] In the formula, is the UAV range constraint, is the turning radius constraint of the UAV, is the climbing angle constraint of the UAV cluster, is the UAV cluster flight altitude constraint, It is the distance constraint item between UAV clusters, and together they constitute the navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm. is the number of swarm drones, The minimum flight altitude of the drone. is the minimum collision avoidance distance between aircraft, is the maximum communication distance between machines, and its value is determined by the specific decision-making problem of UAV cluster navigation.

[0053] Step 302: Initialize particle population parameters (including population size , initial particle position and velocity, maximum number of iterations , convergence function value ) and external file parameters (including the maximum size of external files , weighted dominance coefficient );

[0054] Step 303, determining the Pareto dominance relationship between particles, and adding particles in a Pareto dominance position to an external archive;

[0055] Step 304, determine the weighted dominance relationship between the remaining particles according to the following formula, and add the particles in the weighted dominance position to the external file:

[0056] { ∀ i , f i ( x 1 ) − f i ( x 2 ) + r i ∑ i ≠ j 1 ... k [ f j ( x 1 ) − f j ( x 2 ) ] ≤ 0 ∃ i , f i ( x 1 ) − f i ( x 2 ) + r i ∑ i ≠ j 1 ... k [ f j ( x 1 ) − f j ( x 2 ) ] < 0

[0057] In the formula, is the evaluation function of the multi-objective particle swarm optimization algorithm, which is , and , is a feasible solution to the problem. If the above formula is satisfied, then Weighted Domination ;

[0058] Step 305: If the number of particles in the external file reaches the maximum size , then randomly remove particles that are in weighted dominance but not in Pareto dominance until the number of particles in the external archive is less than the maximum size;

[0059] Step 306, select the global optimal solution (i.e., the particle position that has the most weighted dominance over other particles) from the external archive, and replace and update the global extreme value in the position and velocity iteration formula of the standard particle swarm algorithm. The velocity position iteration formula of a single particle is as follows:

[0060]

[0061]

[0062] In the formula, and For the Particle velocity and position after iterations. is the inertia weight, is the learning factor, and its value is determined by the specific navigation decision problem; For the The individual extreme value of a single particle after iterations, that is, the historical optimal position of a single particle; For the The global extreme value after iterations, which is the global optimal solution selected from the external archive after each iteration.

[0063] Step 307, repeat steps 303 to 306 until the convergence condition is met (i.e., the navigation decision evaluation function value reaches the convergence function value ) or reaches the maximum number of iterations, and obtains the optimized UAV swarm navigation decision solution.

[0064] Step 4: The ground station cloud backend sends the navigation decision solution to the drone cluster, which receives it and flies according to the navigation decision solution.

[0065] Step 5: The UAV cluster monitors the communication rate of the entire network between machines in real time. If the average communication rate of the entire network is greater than the minimum real-time interactive transmission requirement, execute step 6, otherwise go to step 7;

[0066] Step 6: The drone cluster communicates with each other and performs inter-cluster task allocation based on the distributed contract network algorithm. The drone cluster performs navigation flight according to the task allocation results. The specific method is as follows:

[0067] Step 601, numbered The UAV senses the situation change during the mission and, as the bidder, transmits the bidding document containing the task bidding information and the price value according to the information transmission order with the neighboring drone when the step time arrives;

[0068] Step 602: The drone that receives the bid document conducts a situation assessment on the task and determines whether to participate in the bidding for the task based on the environment and its own status; if it decides to participate in the bidding for the task, it calculates its own cost value based on the comprehensive performance evaluation index and transmits it to its neighboring drones when the next long time arrives; otherwise, it only transmits the task information (i.e. the bid document) when the next long time arrives. The calculation formula of the comprehensive performance evaluation index cost value is as follows:

[0069]

[0070] In the formula, They are environmental constraints, self-constraints, and task constraints. is the weight coefficient, and its value is determined by the specific navigation decision problem.

[0071] Step 603: After multiple information exchanges, each drone has learned the cost of the drones participating in the bidding, and the single or multiple drones with the lowest global cost will perform the task.

[0072] Step 7: The individual drones in the cluster perform navigation decision solution calculations without relying on communication with other drones and the ground station cloud background, and perform autonomous navigation flight; the specific method is:

[0073] Step 701, the individual drone obtains its own status and observation information of the environment through the onboard sensor;

[0074] Step 702, according to the current observation information, the current linear velocity and angular velocity control amount are calculated through the proximal optimization strategy network, and sent to the drone onboard controller;

[0075] Step 703: All drones perform autonomous navigation control to achieve individual navigation decision-making effects in the current state.

[0076] The cloud-edge collaborative navigation decision system adopted in this method includes individual UAV autonomous navigation decision control nodes, unmanned cluster edge navigation decision network, and background cloud navigation planning and decision system; the individual UAV autonomous navigation decision control node uses UAV onboard visual navigation, satellite navigation, inertial navigation and other positioning means and airborne navigation decision algorithms to realize the navigation control capability of individual nodes of UAV cluster under non-communication conditions; the unmanned cluster edge navigation decision network uses the communication equipment between UAVs to exchange information between each node, and realizes the collaborative navigation decision capability of UAV cluster under inter-machine communication based on the airborne collaborative navigation decision algorithm; the background cloud navigation planning and decision system uses the remote communication between the front-end and back-end of the UAV, and realizes the intelligent navigation decision assistance of the background cloud to the UAV cluster under remote communication conditions based on the cloud-back-end intelligent navigation decision algorithm library.

[0077] During the collaborative execution of tasks by drone swarms, based on the cloud-edge collaborative navigation decision-making architecture, in the face of communication restricted environments caused by obstruction or electromagnetic interference and without interaction between groups, basic navigation decision-making and control capabilities such as obstacle avoidance flight can be achieved by relying on the individual nodes' own perception and navigation processing; after some communication capabilities between clusters are restored, drones interact with each other through edge network information, rely on group collaboration to achieve intelligent navigation decisions and correct individual autonomous navigation decision status; when the cluster has a good communication link with the background, the situation is transmitted back to the rear and fused, relying on the background cloud-based navigation decision control system to achieve intelligent assisted navigation decision control of the drone swarm.

[0078] The working principle of this method is as follows: during the collaborative execution of tasks by drone clusters, based on the architecture of cloud-edge collaborative navigation decision-making, in the face of communication restricted environments caused by occlusion or electromagnetic interference and without interaction between groups, basic navigation decision-making and control capabilities such as obstacle avoidance flight can be achieved by relying on the perception and navigation processing of individual nodes themselves; after the partial communication capabilities between clusters are restored, drones interact with each other through edge network information, rely on group collaboration to achieve intelligent navigation decisions and correct individual autonomous navigation decision states; under the condition that the cluster has a good communication link with the background, the situation is transmitted back to the rear and fused, relying on the background cloud-based navigation decision control system to realize intelligent auxiliary navigation decision control of drone clusters.

[0079] The following is a detailed description of the UAV cluster navigation control decision under the area search task as an example:

[0080] A method for unmanned aerial vehicle cluster cloud-edge collaborative navigation control decision-making includes the following steps:

[0081] (1) A drone swarm consisting of n drones performs regional search missions at low altitudes in the city. Ground operators monitor the status of the drone swarm itself and its environment in real time through the drone backend ground station.

[0082] (2) The operator detects the communication rate between the ground station cloud backend and the drone cluster. If the average communication rate between any drone in the cluster and the ground station cloud backend is Both are greater than the minimum situation real-time transmission requirements , then go to step (3), otherwise go to step (5);

[0083] (3) The cloud backend of the ground station analyzes the situation information (including the distribution of environmental obstacles, the motion status of the drone itself, the position of other drones in the cluster, etc.) transmitted back in real time by the drone cluster, so as to establish the navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm, and search for the navigation decision solution based on the multi-objective particle swarm optimization algorithm to obtain the navigation decision solution of the drone cluster. The specific method is as follows:

[0084] (301) Based on the real-time situation information transmitted by the UAV cluster, the navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm are established, as shown in the following formula:

[0085] { y = minutes [ L P , R P , i P ] H p ≥ nH minutes n ( n − 1 ) l minutes ≤ l p ≤ n ( n − 1 ) l max

[0086] In the formula, is the UAV range constraint, is the turning radius constraint of the UAV, is the climbing angle constraint of the UAV cluster, is the UAV cluster flight altitude constraint, The distance constraint term between drone clusters together constitutes the navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm; is the number of swarm drones, The minimum flight altitude of the drone. is the minimum collision avoidance distance between aircraft, is the maximum communication distance between machines;

[0087] (302) Initialize particle population parameters and external archive parameters, where particle population parameters include population size , initial particle position and velocity, maximum number of iterations , convergence function value , external archive parameters include the maximum size of external archives , weighted dominance coefficient ;

[0088] (303) Determine the Pareto dominance relationship between particles and add the particles in the Pareto dominance position to the external archive;

[0089] (304) The weighted dominance relationship between the remaining particles is determined according to the following formula, and the particles in the weighted dominance position are added to the external archive:

[0090] { ∀ i , f i ( x 1 ) − f i ( x 2 ) + r i ∑ i ≠ j 1 ... k [ f j ( x 1 ) − f j ( x 2 ) ] ≤ 0 ∃ i , f i ( x 1 ) − f i ( x 2 ) + r i ∑ i ≠ j 1 ... k [ f j ( x 1 ) − f j ( x 2 ) ] < 0

[0091] In the formula, is the evaluation function of the multi-objective particle swarm optimization algorithm, which is , and , is a feasible solution to the problem. If the above formula is satisfied, then Weighted Domination ;

[0092] (305) If the number of particles in the external archive reaches the maximum size , then randomly remove particles that are in weighted dominance but not in Pareto dominance until the number of particles in the external archive is less than the maximum size;

[0093] (306) The global optimal solution is selected from the external archive, that is, the position of the particle that has the most weighted dominance over other particles, and the global extreme value in the position and velocity iteration formula of the standard particle swarm algorithm is replaced and updated. The velocity position iteration formula of a single particle is as follows:

[0094]

[0095]

[0096] In the formula, and For the The particle speed and position after iterations, is the inertia weight, is the learning factor; For the The individual extreme value of a single particle after iterations, that is, the historical optimal position of a single particle; For the The global extreme value after iterations, which is replaced and updated by the global optimal solution selected from the external archive after each iteration;

[0097] (307) Repeat steps (303) to (306) until the navigation decision evaluation function value reaches the convergence function value , or the iteration reaches the maximum number of iterations, and the optimized UAV cluster navigation decision solution is obtained.

[0098] (4) The ground station cloud backend sends the navigation decision solution to the UAV cluster, which receives it and flies according to the navigation decision solution;

[0099] (5) The UAV cluster monitors the communication rate of the entire network between the UAVs in real time. If the average communication rate of the entire network is greater than the minimum real-time interactive transmission requirement, execute step (6); otherwise, go to step (7);

[0100] (6) The drone cluster communicates with each other and performs inter-cluster task allocation based on the distributed contract network algorithm. The drone cluster performs navigation and flight according to the task allocation results. The specific method is as follows:

[0101] (601) No. The UAV senses the situation change during the mission and, as the bidder, transmits the bidding document containing the task bidding information and the price value according to the information transmission order with the neighboring drone when the step time arrives;

[0102] (602) The drone that receives the bid document conducts a situation assessment on the task and determines whether to participate in the bidding for the task based on the environment and its own status; if it decides to participate in the bidding for the task, it calculates its own cost value based on the comprehensive performance evaluation index and transmits it to its neighboring drones when the next long-term arrival occurs; otherwise, it only transmits the task information (i.e., the bid document) when the next long-term arrival occurs. The calculation formula of the comprehensive performance evaluation index cost value is as follows:

[0103]

[0104] In the formula, They are environmental constraints, self-constraints, and task constraints. is the weight coefficient, and its value is determined by the specific navigation decision problem.

[0105] (603) After multiple information exchanges, each drone has learned the cost of the drones participating in the bidding, and the single or multiple drones with the lowest global cost will perform the task.

[0106] (7) Individual drones in the cluster perform navigation decision-making calculations without relying on communication with other drones or the ground station cloud backend, and perform autonomous navigation flight; the specific method is:

[0107] (701) Individual drones obtain their own status (own position) through onboard sensors and speed ) and observation information of the environment (relative distance to other drones or obstacles );

[0108] (702) According to the current observation information, the current linear velocity is obtained by calculating through the proximal optimization strategy network. and angular velocity control quantity and sends it to the drone onboard controller;

[0109] (703) All UAVs perform autonomous navigation control to achieve individual navigation decision-making effects under the current state.

[0110] In summary, the present invention realizes a navigation decision control method for UAV cluster applications, which has the advantages of strong environmental adaptability and excellent navigation decision performance.

[0111] The present invention can be applied to scenarios where drone clusters collaborate to perform tasks in complex physical and electromagnetic environments such as future urban spaces. Based on the cloud-edge collaborative navigation decision-making system, it can solve the problem of accurate, reliable and robust navigation control capabilities of drone platforms.

Claims

1. A UAV cluster cloud-edge collaborative navigation control decision method, characterized in that: The following steps are involved: Step 1: The drone swarm performs its mission at low altitude, and the ground operator monitors the status of the drone swarm itself and its environment in real time through the ground station cloud background; Step 2: Detect the communication rate between the ground station cloud background and the drone cluster. If the average communication rate between any drone in the cluster and the ground station cloud background is greater than the minimum situation real-time transmission requirement, execute step 3, otherwise go to step 5; Step 3: The cloud backend of the ground station analyzes the situation information transmitted back by the drone cluster in real time, thereby establishing a navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm. The situation information includes the distribution of environmental obstacles, the drone's own motion state, and the positions of other drones in the cluster. The navigation decision solution is searched based on the multi-objective particle swarm optimization algorithm to obtain the navigation decision solution of the drone cluster. Step 4: The ground station cloud backend sends the navigation decision solution to the drone cluster, which receives it and flies according to the navigation decision solution. Step 5: The UAV cluster monitors the communication rate of the entire network between machines in real time. If the average communication rate of the entire network is greater than the minimum real-time interactive transmission requirement, execute step 6, otherwise go to step 7; Step 6: The drone cluster communicates with each other and performs inter-cluster task allocation based on the distributed contract network algorithm. The drone cluster performs navigation flight according to the task allocation results. In step 7, each drone in the cluster performs navigation decision solution calculations and performs autonomous navigation flight without relying on communication with other drones and the ground station cloud background.

2. The method for cloud-edge-end collaborative navigation control decision-making of a drone cluster according to claim 1 is characterized in that: The specific method of step 3 is: Step 301, based on the situation information transmitted back by the drone cluster in real time, a navigation decision evaluation function and constraint conditions of a multi-objective particle swarm optimization algorithm are established, as shown in the following formula: In the formula, is the UAV range constraint, is the turning radius constraint of the UAV, is the climbing angle constraint of the UAV cluster, is the UAV cluster flight altitude constraint, The distance constraint term between drone clusters together constitutes the navigation decision evaluation function and constraint conditions of the multi-objective particle swarm optimization algorithm; is the number of swarm drones, The minimum flight altitude of the drone. is the minimum collision avoidance distance between aircraft, is the maximum communication distance between machines; Step 302, initializing particle population parameters and external file parameters, wherein the particle population parameters include the population size , initial particle position and velocity, maximum number of iterations , convergence function value , external archive parameters include the maximum size of external archives , weighted dominance coefficient ; Step 303, determining the Pareto dominance relationship between particles, and adding particles in a Pareto dominance position to an external archive; Step 304, determine the weighted dominance relationship between the remaining particles according to the following formula, and add the particles in the weighted dominance position to the external file: In the formula, is the evaluation function of the multi-objective particle swarm optimization algorithm, which is , and , is a feasible solution to the problem. If the above formula is satisfied, then Weighted Domination ; Step 305: If the number of particles in the external file reaches the maximum size , then randomly remove particles that are in weighted dominance but not in Pareto dominance until the number of particles in the external archive is less than the maximum size; Step 306, select the global optimal solution from the external archive, that is, the particle position that has the most weighted dominance over other particles, and replace and update the global extreme value in the position and velocity iteration formula of the standard particle swarm algorithm. The velocity position iteration formula of a single particle is as follows: In the formula, and For the The particle speed and position after iterations are is the inertia weight, is the learning factor; For the The individual extreme value of a single particle after iterations, that is, the historical optimal position of a single particle; For the The global extreme value after iterations, which is replaced and updated by the global optimal solution selected from the external archive after each iteration; Step 307, repeat steps 303 to 306 until the navigation decision evaluation function value reaches the convergence function value , or the iteration reaches the maximum number of iterations, and the optimized UAV cluster navigation decision solution is obtained.

3. The method for cloud-edge-end collaborative navigation control decision-making of a drone cluster according to claim 2 is characterized in that: The specific method of step 6 is: Step 601, numbered The UAV senses the situation change during the mission and, as the bidder, transmits the bidding document containing the task bidding information and the price value according to the information transmission order with the neighboring drone when the step time arrives; Step 602, the drone that receives the bid performs a situation assessment on the task and determines whether to participate in the bidding for the task; If it decides to bid for the task, it calculates its own cost based on the comprehensive performance evaluation index and transmits it to its neighboring drones when the next long time arrives; Otherwise, only the bid will be passed on when the next long time arrives; The calculation formula for the comprehensive performance evaluation index cost is as follows: In the formula, They are environmental constraints, self-constraints, and task constraints. is the weight coefficient; Step 603: After multiple information exchanges, each drone has learned the cost of the drones participating in the bidding, and the single or multiple drones with the lowest global cost will perform the task.

4. The method for cloud-edge-end collaborative navigation control decision-making of a drone cluster according to claim 3 is characterized in that: The specific method of step 7 is: Step 701, the individual drone obtains its own status and observation information of the environment through the onboard sensor; Step 703, according to the current observation information, the current linear velocity and angular velocity control amount are calculated through the proximal optimization strategy network, and sent to the drone onboard controller; Step 702: All drones perform autonomous navigation control to achieve individual navigation decision-making effects in the current state.

Images