Unmanned aerial vehicle cluster collaborative path planning method based on improved grey wolf algorithm

Abstract

The invention discloses an unmanned aerial vehicle cluster collaborative path planning method based on an improved grey wolf algorithm. The method comprises the following steps: 1, constructing an environment model of a task scene of an unmanned aerial vehicle cluster and a multi-vehicle cooperative constraint condition, and initializing unmanned aerial vehicle motion parameters; 2, generating a high-quality initial feasible solution by adopting a greedy random initialization strategy, and reducing invalid search through priori knowledge; 3, performing global optimization in combination with dynamic weight adjustment and a multi-target fitness function, and balancing multi-target conflicts; outputting an optimized path; 4, repairing the path through local path re-planning or speed adjustment, and if repairing fails, feeding back conflicting individuals to the step 3 for re-optimization to form a closed-loop cooperation mechanism; and 5, generating a 3D path planning graph and a fitness curve through a drawing function, and verifying path security and algorithm convergence. According to the invention, the global search capability is enhanced, and efficient collaborative path planning of the unmanned aerial vehicle cluster in a complex environment is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous navigation and path planning of unmanned aerial vehicles (UAVs), and in particular to a collaborative path planning method for a UAV cluster based on an improved grey wolf algorithm. Background Art

[0002] In daily application scenarios, facing complex three-dimensional environments (such as buildings, trees, mountains and other obstacles), efficient and safe path planning remains a challenging problem.

[0003] Traditional path planning algorithms, such as A*, genetic algorithms, and particle swarm optimization (PSO), perform well for low-dimensional problems. However, they struggle to generate feasible paths in real time in complex, dynamic, and high-dimensional threat environments. They also suffer from deficiencies such as insufficient global optimization capabilities, inability to effectively avoid multiple types of threats, and a tendency to get stuck in local optima. They also lack multi-UAV collision avoidance and time synchronization mechanisms, leading to a high risk of mission failure. A paper (DOI: 10.16002 / j.cnki.10090312.2023.03.017) proposed a method based on an improved particle swarm optimization algorithm for UAV path planning. This method improves performance by setting random inertia weights and adding a perturbation particle update mechanism. Despite this, the algorithm still suffers from slow convergence and a tendency to get stuck in local optima when dealing with high-dimensional problems. Furthermore, it is not optimized for multi-UAV collaborative tasks, highlighting the limitations of traditional algorithms in complex environments. While the traditional gray wolf optimization algorithm possesses global search capabilities, its standard version converges slowly for multi-constrained, high-dimensional problems and is not optimized for UAV collaborative tasks. Summary of the Invention

[0004] In order to overcome the above technical problems, the purpose of the present invention is to provide a UAV cluster collaborative path planning method based on the improved gray wolf algorithm, enhance the global search capability through the simulated annealing mechanism and adaptive mutation strategy, and realize efficient collaborative path planning of UAV clusters in complex environments through dynamic weighted update and adaptive constraint processing mechanism.

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

[0006] A collaborative path planning method for UAV swarm based on an improved grey wolf algorithm includes the following steps:

[0007] Step 1: Environmental modeling and initialization: constructing the UAV swarm mission scenario and multi-machine collaboration constraints, initializing the UAV motion parameters (starting point, target point), and providing global environmental data and constraint rule libraries for subsequent steps;

[0008] Step 2: Generate the initial wolf pack positions. Based on the environmental model and constraints in step 1, a greedy random initialization strategy is used to generate a high-quality initial feasible solution. Prior knowledge is used to reduce invalid searches and provide a convergence starting point for the iterative optimization in step 3.

[0009] Step 3: Improve the Grey Wolf Algorithm for iterative optimization. Starting from the initial feasible solution in Step 2, the algorithm combines dynamic weight adjustment with the multi-objective fitness function for global optimization to balance multi-objective conflicts. During the iteration, the multi-machine collaborative constraints in Step 1 are called in real time to adjust the search direction and output the optimized path. The collaborative constraint processing in Step 4 is also triggered to correct the conflicting paths.

[0010] Step 4: Collaborative constraint processing and path repair. For the optimized path output from step 3, conflict detection is performed in combination with the collaborative constraints from step 1. The path is repaired through local path replanning or speed adjustment. If the repair fails, the conflicting individual is fed back to step 3 for re-optimization, forming a closed-loop collaborative mechanism.

[0011] Step 5: Output and verification. Verify the algorithm convergence based on the multi-objective fitness function data from step 3. Combine the repaired path from step 4 with the mission scenario and multi-machine collaboration constraints from step 1. Generate a 3D path planning diagram and fitness curve through the drawing function to verify the path safety and algorithm convergence.

[0012] The step 1 is specifically as follows:

[0013] Step 1.1: Initialize the UAV cluster parameters, set the starting point (UAV.S), end point (UAV.G), number N, communication radius R, and maximum speed V_max of the UAV;

[0014] Step 1.2: Build a 3D threat field model, define obstacles and radar threat areas (UAV, Menace) and constraints (speed, no-fly altitude, range, etc.).

[0015] The mission scenario is three-dimensional terrain, obstacles, and threat area distribution;

[0016] The multi-machine collaboration constraints are speed limit and obstacle avoidance safety distance.

[0017] The step 2 is specifically as follows:

[0018] Step 2.1: Use the greedy random initialization strategy (PopsInit function) to generate the initial path based on historical task data;

[0019] Step 2.2: Insert random track points between the start and end points to ensure path continuity and obstacle avoidance feasibility;

[0020] Step 2.3: Assign an initial velocity to each UAV to satisfy the constraints, ensure the quality of the initial solution, and reduce invalid searches.

[0021] The step 3 is specifically as follows:

[0022] Step 3.1: Dynamic position update: Update the positions of α, β, and δ wolves according to the standard GWO rule. Introduce a linear attenuation coefficient a to balance exploration and development. The linear attenuation coefficient formula is:

[0023]

[0024] a: The dynamic attenuation coefficient of the Grey Wolf Algorithm, which is used to balance global exploration and local development capabilities and control the UAV path search range (global exploration → local optimization); iter: the current number of iterations; Max_iter: the maximum number of algorithm iterations;

[0025] Step 3.2: Simulated annealing mechanism, in each iteration, calculate the fitness of the new solution f new , if it is worse than the current solution f current , accept inferior solutions with probability P to avoid premature convergence. The probability P acceptance formula of the simulated annealing mechanism is

[0026]

[0027] f new : The fitness value of the newly generated solution, reflecting the comprehensive performance indicators of the path such as fuel consumption, threat cost, and number of collisions; the smaller the value, the better the path; f current : The fitness value of the current solution, that is, the performance index corresponding to the optimal path currently retained by the algorithm; T: Temperature parameter, which controls the probability of accepting an inferior solution and allows the drone path to temporarily degrade to explore a better solution. The initial value is set to T0 and decays with the number of iterations. The formula is:

[0028] T=T0·γ iter

[0029] Where: T0: initial temperature (100-1000), which needs to be adjusted according to the complexity of the task; γ: cooling coefficient (0<γ<1), which controls the temperature drop rate (usually 0.9-0.99); iter: current iteration number;

[0030] Step 3.3: Adaptive mutation strategy. If the optimal solution does not improve after K consecutive iterations, apply Gaussian perturbations to the wolf positions α, β, and δ. The Gaussian mutation formula is:

[0031]

[0032] X: The current position vector of the gray wolf (i.e., path coordinates or speed parameters), the dimension is consistent with the drone state variable; The mean is 0 and the variance is σ 2 Gaussian distributed random number is used to perturb the position; σ: Perturbation intensity, which is dynamically adjusted with the number of iterations. The position of the track point is dynamically adjusted to avoid threats. The formula is:

[0033]

[0034] Where: σ0: initial perturbation intensity (0.1-1.0), controlling the initial amplitude of the mutation; Max_iter: maximum number of algorithm iterations (100-500); trigger condition: when the global optimal solution does not improve in K consecutive iterations, the mutation operation is triggered (K is usually set to 5-10).

[0035] The step 4 is specifically as follows:

[0036] Step 4.1: Call the BoundAdjust function to delete track points that violate the safety distance (UAV.ds) or altitude constraints, and use cubic spline interpolation to generate a smooth path;

[0037] Step 4.2: Integrate the collision penalty term f in the objective function (ObjFun.m) c and the time synchronization error term f m , to ensure cluster coordination, the coordination penalty term in the objective function

[0038] f c =p5·col_times,f m =p4·|t i -UAV.tc|

[0039] f c : Collision penalty item, reflecting the number of collisions between drones; p5: Collision penalty weight (1.0-2.0), which needs to be adjusted according to the mission safety requirements to ensure a safe interval between paths and avoid drone collisions; col_times: The total number of collisions, counted by the path detection module (TrackDetect); f m : time synchronization error term, reflecting the deviation between the actual flight time of the UAV and the coordinated time; p4: time synchronization weight (1.0-1.5), controlling the UAV to reach the target point according to the coordinated time; t i : The actual flight time of the i-th UAV; UAV.tc: The preset collaborative time threshold.

[0040] Beneficial effects of the present invention:

[0041] 1. Dynamic weighting mechanism: Optimize the gray wolf position update strategy through linear attenuation coefficient to improve convergence speed;

[0042] 2. Annealing-mutation hybrid mechanism: This mechanism uses simulated annealing to accept inferior solutions and Gaussian mutation perturbations, completely escaping local optima and enhancing global search capabilities.

[0043] 3. Dynamic parameter and constraint adaptive processing: Temperature T and disturbance intensity σ are automatically adjusted with iteration, reducing reliance on manual parameter adjustment; track points that do not meet constraints are repaired through interpolation to ensure path feasibility;

[0044] 4. Collaborative control mechanism: Integrates time synchronization and collision penalty to achieve efficient collaboration among multiple drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 Flowchart for improving the gray wolf optimization algorithm.

[0046] Figure 2 Block diagram of the UAV swarm path planning system.

[0047] Figure 3 This is the effect diagram of improving the 3D path planning of the Grey Wolf algorithm.

[0048] Figure 4 This is the effect diagram of the traditional grey wolf algorithm 3D path planning.

[0049] Figure 5 Schematic diagram of the fitness curve of the improved GWO algorithm.

[0050] Figure 6 Schematic diagram of the fitness curve of the traditional GWO algorithm. DETAILED DESCRIPTION

[0051] The present invention will be further described in detail below with reference to the accompanying drawings.

[0052] This paper proposes a collaborative path planning method for UAV clusters based on an improved gray wolf optimization algorithm. By integrating a simulated annealing mechanism with an adaptive mutation strategy, the global search capability and dynamic environment adaptability of the algorithm are significantly improved.

[0053] The following steps are included:

[0054] Step 1: Environmental modeling and initialization: constructing the UAV swarm mission scenario and multi-machine collaboration constraints, initializing the UAV motion parameters (starting point, target point), and providing global environmental data and constraint rule libraries for subsequent steps;

[0055] Step 2: Generate the initial wolf pack positions. Based on the environmental model and constraints in step 1, a greedy random initialization strategy is used to generate a high-quality initial feasible solution. Prior knowledge is used to reduce invalid searches and provide a convergence starting point for the iterative optimization in step 3.

[0056] Step 3: Improve the Grey Wolf Algorithm for iterative optimization. Starting from the initial feasible solution in Step 2, the algorithm combines dynamic weight adjustment with the multi-objective fitness function for global optimization to balance multi-objective conflicts. During the iteration, the multi-machine collaborative constraints in Step 1 are called in real time to adjust the search direction and output the optimized path. The collaborative constraint processing in Step 4 is also triggered to correct the conflicting paths.

[0057] Step 4: Collaborative constraint processing and path repair. For the optimized path output from step 3, conflict detection is performed in combination with the collaborative constraints from step 1. The path is repaired through local path replanning or speed adjustment. If the repair fails, the conflicting individual is fed back to step 3 for re-optimization, forming a closed-loop collaborative mechanism.

[0058] Step 5: Output and verification. Verify the algorithm convergence based on the multi-objective fitness function data from step 3. Combine the repaired path from step 4 with the mission scenario and multi-machine collaboration constraints from step 1. Generate a 3D path planning diagram and fitness curve through the drawing function to verify the path safety and algorithm convergence.

[0059] The step 1 is specifically as follows:

[0060] Step 1.1: Initialize the UAV cluster parameters, set the starting point (UAV.S), end point (UAV.G), number N, communication radius R, and maximum speed V_max of the UAV;

[0061] Step 1.2: Build a 3D threat field model, define obstacles and radar threat areas (UAV, Menace) and constraints (speed, no-fly altitude, range, etc.).

[0062] The mission scenario is three-dimensional terrain, obstacles, and threat area distribution;

[0063] The multi-machine collaboration constraints are speed limit and obstacle avoidance safety distance.

[0064] The step 2 is specifically as follows:

[0065] Step 2.1: Use the greedy random initialization strategy (PopsInit function) to generate the initial path based on historical task data;

[0066] Step 2.2: Insert random track points between the start and end points to ensure path continuity and obstacle avoidance feasibility;

[0067] Step 2.3: Assign an initial velocity to each UAV to satisfy the constraints, ensure the quality of the initial solution, and reduce invalid searches.

[0068] The step 3 is specifically as follows:

[0069] Step 3.1: Dynamic position update: Update the positions of α, β, and δ wolves according to the standard GWO rule. Introduce the linear attenuation coefficient a to balance exploration and development. The linear attenuation coefficient formula is:

[0070]

[0071] a: The dynamic attenuation coefficient of the Grey Wolf Algorithm, which is used to balance global exploration and local development capabilities and control the UAV path search range (global exploration → local optimization); iter: the current number of iterations; Max_iter: the maximum number of algorithm iterations;

[0072] Step 3.2: Simulated annealing mechanism, in each iteration, calculate the fitness of the new solution f new , if it is worse than the current solution f current , accept inferior solutions with probability P to avoid premature convergence. The probability P acceptance formula of simulated annealing mechanism is

[0073]

[0074] f new : The fitness value of the newly generated solution, reflecting the comprehensive performance indicators of the path such as fuel consumption, threat cost, and number of collisions; the smaller the value, the better the path; f current : The fitness value of the current solution, that is, the performance index corresponding to the optimal path currently retained by the algorithm; T: Temperature parameter, which controls the probability of accepting an inferior solution and allows the drone path to temporarily degrade to explore a better solution. The initial value is set to T0 and decays with the number of iterations. The formula is:

[0075] T=T0·γ iter

[0076] Where: T0: initial temperature (100-1000), which needs to be adjusted according to the complexity of the task; γ: cooling coefficient (0<γ<1), which controls the temperature drop rate (usually 0.9-0.99); iter: current iteration number;

[0077] Step 3.3: Adaptive mutation strategy. If the optimal solution does not improve after K consecutive iterations, apply Gaussian perturbations to the wolf positions α, β, and δ. The Gaussian mutation formula is:

[0078]

[0079] X: The current position vector of the gray wolf (i.e., path coordinates or speed parameters), the dimension is consistent with the drone state variable; The mean is 0 and the variance is σ 2 Gaussian distributed random number is used to perturb the position; σ: Perturbation intensity, which is dynamically adjusted with the number of iterations. The position of the track point is dynamically adjusted to avoid threats. The formula is:

[0080]

[0081] Where: σ0: initial perturbation intensity (0.1-1.0), controlling the initial amplitude of the mutation; Max_iter: maximum number of algorithm iterations (100-500); trigger condition: when the global optimal solution does not improve in K consecutive iterations, the mutation operation is triggered (K is usually set to 5-10).

[0082] The step 4 is specifically as follows:

[0083] Step 4.1: Call the BoundAdjust function to delete track points that violate the safety distance (UAV.ds) or altitude constraints, and use cubic spline interpolation to generate a smooth path;

[0084] Step 4.2: Integrate the collision penalty term f in the objective function (ObjFun.m) c and the time synchronization error term f m , to ensure cluster coordination, the coordination penalty term in the objective function

[0085] f c =p5·col_times,f m =p4·|t i -UAV.tc|

[0086] f c : Collision penalty item, reflecting the number of collisions between drones; p5: Collision penalty weight (1.0-2.0), which needs to be adjusted according to the mission safety requirements to ensure a safe interval between paths and avoid drone collisions; col_times: The total number of collisions, counted by the path detection module (TrackDetect); f m : time synchronization error term, reflecting the deviation between the actual flight time of the UAV and the coordinated time; p4: time synchronization weight (1.0-1.5), controlling the UAV to reach the target point according to the coordinated time; t i : The actual flight time of the i-th UAV; UAV.tc: The preset collaborative time threshold.

[0087] like Figure 1 As shown in the figure, the improved algorithm balances exploration and development through a dynamic attenuation coefficient, combines the annealing mechanism to accept probabilistic inferior solutions to escape the local optimum, and dynamically adjusts the track points when the path stagnates through a Gaussian perturbation strategy to ensure path security in complex threat environments.

[0088] like Figure 2 As shown in the figure, the system architecture integrates task initialization, hybrid optimization engine and collaborative control module to detect collision risk and time synchronization error in real time, and generate an optimized path that meets the safety distance (UAV.ds) and collaborative time (UAV.tc) constraints.

[0089] contrast Figure 3 and Figure 4 As can be seen, the paths generated by the traditional GWO algorithm do not completely avoid the threat area, and the track points are distributed discretely, causing drone flight jitter. In contrast, the drone paths generated by EGWO circumvent radar threat areas in three-dimensional space, minimizing the threat cost. Track points are connected via cubic spline interpolation, resulting in a smooth path that ensures flight stability. The spacing between multiple drone paths is even, meeting safety distance constraints.

[0090] contrast Figure 5 and Figure 6 The fitness curve of the traditional GWO algorithm shows that it lacks an annealing mechanism and exhibits significant curve jitter. The improved GWO algorithm's fitness curve shows a smoother decline, with slightly faster convergence and improved premature convergence. This demonstrates the efficiency and stability of the improved algorithm, making it suitable for real-time path planning scenarios.

[0091] The present invention provides an efficient and robust collaborative path solution for UAV swarm missions, and is suitable for high-dynamic scenarios such as military penetration and disaster relief.

[0092] A simulation experiment was conducted on a computer. The simulation scene included a drone swarm consisting of 10 drones, 5 high-altitude and low-altitude radars of different sizes, and 8 threatening obstacles of different sizes. The experimental results are shown in Table 1:

[0093] Table 1 Simulation results of different algorithms

[0094]

[0095] in conclusion:

[0096] Simulation experimental results show that in a complex scenario containing 10 drones, 5 radar threats of different sizes and 8 dynamic obstacles, the improved Grey Wolf Optimization Algorithm (EGWO) exhibits significant advantages over the traditional Grey Wolf Algorithm (GWO): the solution time of EGWO is shortened by 2.5% (686.93 seconds vs. 704.49 seconds), the objective function convergence value is optimized by 4.1% (28.58 vs. 29.80), the total flight distance and duration are reduced by 6.8% (257.09 km vs. 275.95 km) and 8.1% (1181.18 seconds vs. 1284.66 seconds) respectively, and the convergence generations are reduced by 3.8% (152 generations vs. 158 generations), while achieving zero collision risk. By integrating the simulated annealing mechanism with the adaptive mutation strategy, the algorithm significantly improves the path planning efficiency, global optimization capability and adaptability to complex environments, providing efficient and reliable technical support for UAV cluster collaborative tasks under dynamic threat scenarios (such as military penetration and disaster relief).

Claims

1. A UAV swarm collaborative path planning method based on an improved grey wolf algorithm, characterized in that: The following steps are included: Step 1: Construct the environment model of the UAV swarm mission scenario and the multi-machine collaboration constraints, and initialize the UAV motion parameters; Step 2: Based on the environment model and constraints in step 1, a greedy random initialization strategy is used to generate a high-quality initial feasible solution, and prior knowledge is used to reduce invalid searches. Step 3: Starting from the initial feasible solution in step 2, global optimization is performed by combining dynamic weight adjustment with the multi-objective fitness function to balance multi-objective conflicts. During the iteration process, the multi-machine collaborative constraints in step 1 are called in real time to adjust the search direction and output the optimized path. Step 4: Based on the optimized path output in step 3, conflict detection is performed in combination with the coordination constraints in step 1. The path is repaired through local path replanning or speed adjustment. If the repair fails, the conflicting individual is fed back to step 3 for re-optimization, forming a closed-loop coordination mechanism. Step 5: Verify the convergence of the algorithm based on the multi-objective fitness function data from step 3. Combine the repaired path from step 4 with the mission scenario and multi-machine collaboration constraints from step 1. Generate a 3D path planning diagram and fitness curve through the drawing function to verify the path safety and algorithm convergence.

2. The method for collaborative path planning of UAV clusters based on the improved grey wolf algorithm according to claim 1 is characterized in that: The step 1 is specifically as follows: Step 1.1: Initialize the UAV cluster parameters, set the starting point (UAV.S), end point (UAV.G), number N, communication radius R, and maximum speed V_max of the UAV; Step 1.2: Build a 3D threat field model, define obstacles and radar threat areas (UAV, Menace) and constraints.

3. The UAV swarm collaborative path planning method based on the improved gray wolf algorithm according to claim 1 is characterized in that: The mission scenario is three-dimensional terrain, obstacles, and threat area distribution; The multi-machine collaboration constraints are speed limit and obstacle avoidance safety distance.

4. The method for collaborative path planning of a UAV cluster based on an improved gray wolf algorithm according to claim 1, characterized in that: The step 2 is specifically as follows: Step 2.1: Use a greedy random initialization strategy and combine historical task data to generate the initial path; Step 2.2: Insert random track points between the start and end points to ensure path continuity and obstacle avoidance feasibility; Step 2.3: Assign an initial velocity to each UAV to satisfy the constraints, ensure the quality of the initial solution, and reduce invalid searches.

5. The method for collaborative path planning of UAV clusters based on the improved grey wolf algorithm according to claim 1 is characterized in that: The step 3 is specifically as follows: Step 3.1: Dynamic position update: Update the positions of α, β, and δ wolves according to the standard GWO rule. Introduce the linear attenuation coefficient a to balance exploration and development. The linear attenuation coefficient formula is: a: The dynamic attenuation coefficient of the Grey Wolf Algorithm, which is used to balance global exploration and local development capabilities and control the UAV path search range (global exploration → local optimization); iter: the current number of iterations; Max_iter: the maximum number of algorithm iterations; Step 3.2: Simulated annealing mechanism, in each iteration, calculate the fitness of the new solution f new , if it is worse than the current solution f current , accept inferior solutions with probability P to avoid premature convergence. The probability P acceptance formula of the simulated annealing mechanism is F new : The fitness value of the newly generated solution, reflecting the comprehensive performance indicators of the path such as fuel consumption, threat cost, and number of collisions; the smaller the value, the better the path; f current : The fitness value of the current solution, that is, the performance index corresponding to the optimal path currently retained by the algorithm; T: Temperature parameter, which controls the probability of accepting an inferior solution and allows the drone path to temporarily degrade to explore a better solution. The initial value is set to T0 and decays with the number of iterations. The formula is: T=T0·γ iter Where: T0: initial temperature (100-1000), which needs to be adjusted according to the complexity of the task; γ: cooling coefficient (0<γ<1), which controls the temperature drop rate (usually 0.9-0.99); iter: current iteration number; Step 3.3: Adaptive mutation strategy. If the optimal solution does not improve after K consecutive iterations, apply Gaussian perturbations to the wolf positions α, β, and δ. The Gaussian mutation formula is: X: The current position vector of the gray wolf (i.e., path coordinates or speed parameters), the dimension is consistent with the drone state variable; The mean is 0 and the variance is σ 2 Gaussian distributed random number is used to perturb the position; σ: Perturbation intensity, which is dynamically adjusted with the number of iterations. The position of the track point is dynamically adjusted to avoid threats. The formula is: Where: σ0: initial perturbation intensity (0.1-1.0), controlling the initial amplitude of the mutation; Max_iter: maximum number of algorithm iterations (100-500); trigger condition: when the global optimal solution does not improve in K consecutive iterations, the mutation operation is triggered (K is set to 5-10).

6. The method for collaborative path planning of UAV clusters based on the improved grey wolf algorithm according to claim 1, characterized in that: The step 4 is specifically as follows: Step 4.1: Call the BoundAdjust function to delete track points that violate the safety distance or altitude constraints and use cubic spline interpolation to generate a smooth path; Step 4.2: Integrate the collision penalty term f into the objective function c and the time synchronization error term f m ,ensure cluster coordination, the coordination penalty term in the objective function; f c =p5·col_times,f m =p4·|t i -UAV.tc| f c : Collision penalty item, reflecting the number of collisions between drones; p5: Collision penalty weight (1.0-2.0), which needs to be adjusted according to the mission safety requirements to ensure a safe interval between paths and avoid drone collisions; col_times: The total number of collisions, counted by the path detection module; f m : time synchronization error term, reflecting the deviation between the actual flight time of the UAV and the coordinated time; p4: time synchronization weight (1.0-1.5), controlling the UAV to reach the target point according to the coordinated time; t i : The actual flight time of the i-th UAV; UAN.tc: The preset collaborative time threshold.

Images