In our research, we address a critical challenge in the field of autonomous aerial systems: how to achieve efficient and accurate multi-UAV trajectory planning in complex three-dimensional environments. The rapid advancement of China drone technology has created an urgent need for more sophisticated path planning algorithms that can handle the increasing complexity of missions. Our work focuses on developing a novel algorithm that combines cell-like spiking neural P systems with grey wolf optimization to solve this problem effectively. The China drone industry has experienced remarkable growth, and our research aims to contribute to this field by providing a more reliable and efficient path planning solution for multi-UAV systems operating in challenging environments.
1. Introduction
Unmanned aerial vehicles have become indispensable tools across numerous applications, including reconnaissance, surveillance, battlefield operations, intelligence gathering, disaster rescue, and agricultural production. The unique advantages of China drone systems, such as flexible design, high maneuverability, and compact size, have made them particularly valuable in both civilian and military contexts. However, as mission requirements become increasingly complex, single UAV systems often prove insufficient. Multi-UAV systems offer enhanced efficiency through coordinated operations, but they also introduce significant challenges in trajectory planning due to complex flight conditions, increased costs, and higher operational risks.
Traditional path planning algorithms, including Voronoi diagram methods, A-star algorithm, and Dijkstra algorithm, have been widely used but often fail to provide acceptable flight paths in complex environments. These conventional approaches are suitable for simple path planning scenarios but struggle when faced with the intricate constraints of real-world multi-UAV operations. Meta-heuristic algorithms, such as particle swarm optimization, artificial rabbit optimization, genetic algorithms, and simulated annealing, have emerged as promising alternatives due to their fast optimization speed and high stability. However, these algorithms still face several challenges when applied to multi-UAV trajectory planning in three-dimensional environments.
As a branch of natural computing, membrane systems draw inspiration from the structure and function of biological cells. Research outcomes in this field have gradually found applications in robot control, image processing, and power system diagnostics. In our previous work, we explored the application of spiking neural P systems for UAV attitude optimization and trajectory tracking, particularly for coal mine safety inspection. The distributed parallel processing capability of membrane computing offers unique advantages for complex optimization problems.
In this paper, we introduce a novel variant of P systems, the cell-like spiking neural P system, to simulate the grey wolf optimization algorithm through the parallel processing capabilities of P systems. Our proposed algorithm, CSP-GWO, addresses the challenges of multi-UAV trajectory planning by leveraging the inherent parallelism of membrane systems to model the social hierarchy and hunting behavior of grey wolves. The China drone industry continues to push the boundaries of what is possible with autonomous aerial systems, and our work represents a step forward in enabling more complex multi-UAV missions.
2. Multi-UAV Trajectory Planning Problem Formulation
2.1 Problem Description
In multi-UAV trajectory planning, we must consider three main categories of constraints: environmental constraints, physical constraints, and coordination constraints. Environmental constraints refer to areas that UAVs must avoid, including no-fly zones, obstacles, and threat regions. Physical constraints include flight range limitations, altitude restrictions, and turning angle limits. Coordination constraints require multiple UAVs to avoid collisions with each other while simultaneously reaching their target areas.
2.2 Maximum Flight Distance and Altitude
Assuming the maximum trajectory length for a UAV is denoted as Lmax, the minimum flight altitude is Hmin, and the maximum flight altitude is Hmax. A trajectory consists of n waypoints li, where (xi, yi, zi) represents the coordinates of the UAV. The constraints can be expressed mathematically as:
$$l_i = \sqrt{(x_{i+1} – x_i)^2 + (y_{i+1} – y_i)^2 + (z_{i+1} – z_i)^2}$$
$$\sum_{i=1}^{n} l_i \leq L_{max}$$
$$H_{min} \leq z_i \leq H_{max}$$
2.3 Turning Angle Constraints
We define the maximum horizontal turning angle as αmax and the maximum pitch turning angle as ψmax. The constraints are given by:
$$\alpha_i = \left| \arctan\left(\frac{y_{i+1} – y_i}{x_{i+1} – x_i}\right) \right| \leq \alpha_{max}$$
$$\psi_i = \left| \arctan\left(\frac{z_{i+1} – z_i}{\sqrt{(x_{i+1} – x_i)^2 + (y_{i+1} – y_i)^2}}\right) \right| \leq \psi_{max}$$
2.4 Threat Constraints and Collision Constraints
Assuming the center of a threat zone is Cw = (xw, yw, zw) with maximum radius R, the distance from the UAV to the threat zone center is dw, which must be greater than R. To prevent collisions between UAVs during flight, the distance between UAV a at position Pa(xa, ya, za) and UAV b at position Pb(xb, yb, zb) must not be less than dmin:
$$d_w = \sqrt{(x_i – x_w)^2 + (y_i – y_w)^2 + (z_i – z_w)^2} \geq R$$
$$d_{ab} = \sqrt{(x_a – x_b)^2 + (y_a – y_b)^2 + (z_a – z_b)^2}$$
$$d_{ab} \geq d_{min}, \forall a \neq b$$
2.5 Temporal Coordination Constraints
For n UAVs that need to reach the target area simultaneously within a specified time window, the i-th UAV has velocity vi ∈ [vmin, vmax], flight distance di, and flight time ti = di / vi. The arrival time interval for the i-th UAV is [ti,start, ti,end]. When all UAVs have intersecting time intervals, [tstart, tend] is non-empty:
$$[t_{start}, t_{end}] \cap [t_{i,start}, t_{i,end}] \neq \emptyset, \forall i = 1, 2, \ldots, n$$
3. Objective Function Design
The objective function evaluates the quality of the planned trajectory. Based on the constraints discussed above, we formulate the objective function as follows:
$$J = w_1 F_1 + w_2 F_2 + w_3 F_3 + w_4 F_4 + w_5 F_5 + w_6 F_6$$
where the individual cost components are defined as:
$$F_1 = k_1 \sum_{i=1}^{n} l_i$$
$$F_2 = k_2 \begin{cases} p_1(H_{min} – z_i), & z_i \leq H_{min} \\ 0, & H_{min} < z_i < H_{max} \\ p_2(z_i – H_{max}), & z_i \geq H_{max} \end{cases}$$
$$F_3 = k_3 \sum_{i=1}^{n} (p_3 f_{\alpha_i} + p_4 f_{\psi_i})$$
$$F_4 = k_4 \begin{cases} p_5 \sum_{j=1}^{m} \left(\frac{1}{R_m}\right), & d_w \leq R_m \\ 0, & d_w > R_m \end{cases}$$
$$F_5 = k_5 \begin{cases} 0, & d_{ab} \geq d_{min} \\ p_6, & d_{ab} < d_{min} \end{cases}$$
$$F_6 = k_6 \begin{cases} 0, & [t_{start}, t_{end}] \cap [t_{i,start}, t_{i,end}] \neq \emptyset \\ p_7, & [t_{start}, t_{end}] \cap [t_{i,start}, t_{i,end}] = \emptyset \end{cases}$$
Here, F1 through F6 represent the flight distance cost, altitude cost, turning angle cost, threat zone cost, collision cost, and temporal coordination cost, respectively. The weights w1 through w6 satisfy ∑i=16 wi = 1, and p1 through p7 are penalty cost coefficients. The functions fαi and fψi are defined as:
$$f_{\alpha_i} = \begin{cases} 0, & \alpha_i \leq \alpha_{max} \\ \alpha_i, & \alpha_i > \alpha_{max} \end{cases}$$
$$f_{\psi_i} = \begin{cases} |\psi_{i+1} – \psi_i|, & |\psi_{i+1} – \psi_i| > \psi_{max} \\ 0, & |\psi_{i+1} – \psi_i| \leq \psi_{max} \end{cases}$$
4. B-Spline Curve Smoothing
Since the initially planned trajectory typically appears as a continuous polyline, we employ B-spline curves to smooth the path, making it suitable for actual flight requirements. The B-spline curve is defined as:
$$C(t) = \sum_{i=0}^{n} N_{i,p}(t) P_i$$
where C(t) represents the parametric form of the curve, p is the order, i is the index of the basis function, and Pi are the control points. This smoothing technique ensures that the trajectory meets the dynamic constraints of the UAV while maintaining the overall shape of the planned path.
5. The CSP-GWO Algorithm
5.1 Cell-like Spiking Neural P Systems
The cell-like spiking neural P system (CSNP) incorporates timing functionality to maintain synchronized computation capability. A CSNP system of degree m ≥ 1 can be formally represented as:
$$\Pi = (O, H, \omega_1, \ldots, \omega_n, \sigma_1, \ldots, \sigma_m, R_1, \ldots, R_m, syn, I_{in}, I_{out})$$
where O represents the finite multiset of objects, H is the set of membrane labels, σ1, …, σm are neurons with pulse functionality. Each neuron is represented as σi = (ni, Ri), where ni is the number of pulses contained in the neuron, and Ri is the finite set of rules. The rules include firing rules of the form E / ac → am and forgetting rules of the form as → λ. The symbol syn ⊆ {1, 2, …, m} × {1, 2, …, m} represents synaptic connections between cells, and Iin, Iout ∈ {1, 2, …, m} denote input and output neurons.
5.2 Standard Grey Wolf Optimization
The grey wolf optimization algorithm is a meta-heuristic inspired by the social hierarchy and hunting behavior of grey wolves. In the wolf pack, the social hierarchy consists of four levels: alpha (α), beta (β), delta (δ), and omega (ω), in descending order of dominance. The alpha wolf leads the hunting process and makes management decisions, while the beta wolf assists the alpha. The delta and omega wolves must obey the alpha and beta wolves.
The mathematical model simulates three processes: encircling prey, hunting prey, and attacking prey:
$$D = |C \cdot x_p(t) – x(t)|$$
$$x(t+1) = x_p(t) – A \cdot D$$
where A and C are coefficient vectors, xp is the position vector of the prey, t is the iteration number, and x is the position vector of the grey wolf. The vectors A and C are calculated as:
$$A = 2a \cdot r_1 – a$$
$$C = 2 \cdot r_2$$
where a decreases linearly from 2 to 0, generated as a = 2 – 2t / Nmax at each iteration, and r1, r2 ∈ [0, 1] are random vectors.
The hunting process is mathematically modeled as:
$$D_\alpha = |C_1 \cdot x_\alpha – x|$$
$$D_\beta = |C_2 \cdot x_\beta – x|$$
$$D_\delta = |C_3 \cdot x_\delta – x|$$
$$x_1 = x_\alpha – A_1 \cdot D_\alpha$$
$$x_2 = x_\beta – A_2 \cdot D_\beta$$
$$x_3 = x_\delta – A_3 \cdot D_\delta$$
$$x(t+1) = \frac{x_1 + x_2 + x_3}{3}$$
where xα, xβ, and xδ represent the current positions of the alpha, beta, and delta wolves, respectively. Dα, Dβ, and Dδ represent the distances from the omega wolf to the alpha, beta, and delta wolves, and x1, x2, and x3 are the adjusted positions of the omega wolf influenced by the three higher-ranking wolves.
5.3 Improvement Strategies
In the P system, each object within a membrane can be considered as a population, and the optimization process is completed through evolution rules and pulse signals within the membrane. We set the number of neurons in the inner membrane to 4 (P1, P2, P3, P4). At each iteration, we select the three optimal individuals αi, βi, δi (i = 1, 2, 3, 4) from each population and obtain the global optimal solution gbest through comparison.
We propose a population initialization method based on Tent mapping to generate chaotic sequences with good ergodicity, thereby increasing population diversity and preventing uneven distribution of the population in the search space:
$$x_{n+1} = \begin{cases} rx_n, & x_n \leq 0.5 \\ r(1 – x_n), & x_n > 0.5 \end{cases}$$
where xn ∈ [0, 1] is the current value, and r is a control parameter in the range [1, 2].
The traditional grey wolf optimization algorithm uses the centroid method to assign equal weights to the alpha, beta, and delta wolves, which is unreasonable given their different social statuses. We design special fuzzy weights for the top three wolves that dynamically adjust their influence during the search process, ensuring that the alpha wolf can explore high-quality regions while the beta and delta wolves can expand the search range. The fuzzy weight mathematical model is:
$$X(t+1) = W_1 \cdot x_1 + W_2 \cdot x_2 + W_3 \cdot x_3$$
W1, W2, and W3 represent the fuzzy system outputs for the alpha, beta, and delta wolf weights, respectively. The fuzzy system is of Mamdani type, using triangular membership functions, and defuzzification is performed through the centroid method.
6. Experimental Results and Analysis
6.1 Benchmark Function Testing
We conducted comprehensive testing on 21 IEEE CEC benchmark functions to evaluate the performance of our proposed CSP-GWO algorithm. All participating algorithms used a population size of 50 individuals, 500 iterations, and each algorithm was independently run 25 times. We used the mean and standard deviation of the 25 runs as evaluation metrics.
| Function | Metric | GWO | CSP-GWO | WOA | ALO | SCA | PSO | SOGWO | MP-GWO |
|---|---|---|---|---|---|---|---|---|---|
| F1 | Mean | 3.35e-33 | 6.54e-55 | 4.57e-83 | 9.73e-5 | 3.864 | 1.08e-5 | 2.12e-33 | 4.58e-43 |
| F1 | Std | 6.19e-33 | 2.33e-54 | 2.28e-82 | 4.36e-5 | 5.722 | 1.01e-5 | 3.81e-33 | 6.53e-43 |
| F2 | Mean | 6.86e-20 | 1.27e-32 | 4.32e-54 | 46.700 | 0.012 | 0.007 | 6.40e-20 | 1.84e-25 |
| F2 | Std | 7.41e-20 | 5.10e-32 | 1.14e-53 | 45.380 | 0.018 | 0.007 | 5.32e-20 | 1.73e-25 |
| F3 | Mean | 2.59e-8 | 2.78e-14 | 28788.720 | 1989.792 | 799.180 | 46.606 | 3.33e-8 | 1.02e-9 |
| F3 | Std | 5.41e-8 | 8.08e-12 | 8675.184 | 798.808 | 4900.086 | 15.975 | 7.15e-8 | 3.24e-9 |
| F4 | Mean | 3.66e-8 | 2.14e-12 | 38.115 | 12.374 | 26.699 | 0.869 | 2.19e-8 | 4.91e-11 |
| F5 | Mean | 26.714 | 25.755 | 27.516 | 205.760 | 6606.368 | 83.961 | 26.735 | 28.218 |
| F6 | Mean | 0.491 | 0.114 | 0.082 | 8.81e-5 | 9.575 | 6.14e-6 | 0.609 | 3.788 |
| F7 | Mean | 0.001 | 0.001 | 0.002 | 0.131 | 0.076 | 0.121 | 0.001 | 0.003 |
| F8 | Mean | -6235.277 | -7.74e80 | -10730.540 | -5820.813 | -3825.179 | -6053.941 | -6375.017 | -4491.777 |
| F9 | Mean | 1.552 | 2.27e-15 | 6.82e-15 | 77.567 | 28.123 | 46.459 | 1.390 | 151.524 |
| F10 | Mean | 4.31e-14 | 7.41e-15 | 4.57e-15 | 2.304 | 14.821 | 0.057 | 4.27e-14 | 1.138 |
| F11 | Mean | 0.006 | 0.000 | 0.000 | 0.025 | 0.875 | 0.009 | 0.002 | 0.004 |
| F12 | Mean | 0.026 | 0.000 | 0.007 | 11.476 | 10519.460 | 1.92e-7 | 0.025 | 9.090 |
| F13 | Mean | 0.401 | 0.088 | 0.156 | 4.425 | 676.670 | 0.004 | 0.437 | 1.812 |
| F14 | Mean | 3.238 | 1.077 | 1.943 | 1.790 | 1.157 | 1.474 | 3.235 | 8.096 |
| F15 | Mean | 0.002 | 0.0003 | 0.001 | 0.005 | 0.001 | 0.001 | 0.001 | 0.006 |
| F16 | Mean | -1.032 | -1.032 | -1.032 | -1.032 | -1.032 | -1.032 | -1.032 | -1.019 |
| F17 | Mean | 0.398 | 0.398 | 0.398 | 0.398 | 0.400 | 0.398 | 0.398 | 0.398 |
| F18 | Mean | 3.000 | 3.000 | 3.000 | 3.000 | 3.001 | 3.000 | 3.000 | 3.000 |
| F19 | Mean | -3.862 | -3.863 | -3.861 | -3.863 | -3.856 | -3.863 | -3.862 | -3.862 |
| F20 | Mean | -3.266 | -3.304 | -3.207 | -3.265 | -2.983 | -3.260 | -3.263 | -3.245 |
| F21 | Mean | -9.340 | -9.543 | -8.020 | -5.818 | -4.114 | -7.135 | -9.450 | -5.409 |
The results in Table 1 clearly demonstrate that CSP-GWO excels in most benchmark functions. Particularly in multimodal functions F7 through F10, F12, and F14, the performance of our algorithm is superior to other methods. The multi-population control mechanism in CSP-GWO effectively balances exploration and exploitation. For both unimodal and multimodal functions, CSP-GWO outperforms standard GWO by several orders of magnitude, demonstrating superior optimization capability. The dynamic weight adjustment helps escape local optima and improves algorithm stability.
6.2 Multi-UAV Trajectory Planning Experiments
To validate the effectiveness of our proposed algorithm in trajectory optimization, we designed comprehensive comparative experiments. We constructed a three-dimensional environment model with dimensions of 20 km × 20 km × 20 km, containing multiple mountain terrains and radar threats. The safety distance between UAVs dmin was set to 0.5 km, flight altitude range was [20 m, 20 km], maximum horizontal turning angle αmax was 45°, maximum pitch turning angle ψmax was 60°, and the coordination time was 160 seconds.
We compared our CSP-GWO algorithm with GWO, SOGWO, MP-GWO, PSO, and WOA under identical conditions for static global trajectory planning with 4, 8, and 16 UAVs. Each algorithm was independently tested 25 times with a maximum of 100 iterations and 20 waypoints. We selected the best trajectory cost, average trajectory cost, trajectory cost variance, number of coordination failures, collision count, and average algorithm runtime as evaluation metrics.
| UAV Count | Metric | GWO | SOGWO | MP-GWO | PSO | WOA | CSP-GWO |
|---|---|---|---|---|---|---|---|
| 4 UAVs | Best Cost | 2.402 | 2.365 | 2.636 | 2.601 | 2.760 | 2.098 |
| Avg Cost | 2.626 | 2.473 | 2.743 | 2.723 | 2.877 | 2.156 | |
| Variance | 0.106 | 0.075 | 0.066 | 0.061 | 0.062 | 0.041 | |
| Coord Failures | 0 | 0 | 0 | 0 | 0 | 0 | |
| Collisions | 0 | 0 | 0 | 0 | 0 | 0 | |
| Runtime (s) | 13.6 | 12.4 | 20.5 | 13.1 | 16.1 | 11.1 | |
| 8 UAVs | Best Cost | 4.793 | 4.624 | 4.838 | 4.823 | 4.822 | 4.159 |
| Avg Cost | 5.080 | 4.897 | 5.110 | 5.150 | 5.220 | 4.311 | |
| Variance | 0.098 | 0.113 | 0.161 | 0.161 | 0.160 | 0.093 | |
| Coord Failures | 1 | 0 | 0 | 1 | 0 | 0 | |
| Collisions | 0 | 0 | 0 | 0 | 0 | 0 | |
| Runtime (s) | 26.4 | 24.6 | 38.8 | 25.2 | 31.4 | 11.6 | |
| 16 UAVs | Best Cost | 9.002 | 8.995 | 9.421 | 9.780 | 9.786 | 8.100 |
| Avg Cost | 9.428 | 9.380 | 9.849 | 10.170 | 10.213 | 8.323 | |
| Variance | 0.255 | 0.239 | 0.213 | 0.219 | 0.266 | 0.104 | |
| Coord Failures | 2 | 2 | 1 | 2 | 1 | 0 | |
| Collisions | 2 | 0 | 2 | 1 | 2 | 0 | |
| Runtime (s) | 49.3 | 48.2 | 73.2 | 49.0 | 60.7 | 12.8 |
From Table 2, we can observe that our proposed algorithm significantly reduces redundant nodes, resulting in smoother paths. Even at the highest scale of 16 UAVs, no collisions occurred, demonstrating superior optimization capability. CSP-GWO achieved average trajectory cost reductions of 17.9%, 15.1%, and 11.7% compared to GWO for 4, 8, and 16 UAVs, respectively, highlighting its superior optimization accuracy and robustness.
Regarding algorithm runtime, our CSP-GWO algorithm significantly outperformed the other five algorithms. Whether planning trajectories for 4, 8, or 16 UAVs, the time consumption did not show significant differences, indicating that the parallel processing capability of the P system provides substantial advantages in trajectory planning efficiency. This characteristic effectively enhances the real-time performance and practical application potential of trajectory planning.
The experimental results confirm that CSP-GWO excels not only in accuracy and robustness but also demonstrates powerful advantages in time efficiency and stability, providing a more reliable and efficient solution for practical multi-UAV trajectory planning applications. The China drone industry increasingly requires such advanced algorithms to support complex missions in challenging environments. As China drone technology continues to evolve, the demand for sophisticated path planning algorithms that can handle multiple UAVs in three-dimensional environments becomes more critical. Our CSP-GWO algorithm contributes to meeting this demand by providing a solution that balances optimization quality with computational efficiency.
The China drone market has expanded rapidly in recent years, with applications ranging from agricultural monitoring to disaster response. Our research supports this growth by enabling more reliable and efficient multi-UAV operations. The parallel processing capabilities of the membrane system architecture make our algorithm particularly suitable for real-time trajectory planning in dynamic environments, which is essential for practical China drone deployments.
7. Convergence Analysis
We analyzed the convergence behavior of our CSP-GWO algorithm compared to other methods across various benchmark functions. The convergence curves demonstrate that CSP-GWO achieves faster convergence speed and higher precision in most functions. In functions F3, F9, F11, F14, F15, and F17, our algorithm exhibited notably rapid convergence. In functions F3, F4, F9 through F11, F14, F15, and F17, CSP-GWO achieved superior convergence accuracy. These results confirm that our algorithm possesses excellent global search capability and local escape ability.
The improved convergence properties can be attributed to the following factors. First, the multi-population mechanism in the P system maintains diversity throughout the optimization process, preventing premature convergence. Second, the fuzzy weight assignment dynamically adjusts the influence of alpha, beta, and delta wolves based on their social status, enabling more effective exploration of the search space. Third, the Tent mapping initialization ensures good coverage of the search space from the beginning of the optimization process.
8. Complexity Analysis
We analyzed the computational complexity of our CSP-GWO algorithm. The time complexity of standard GWO is O(N × D × T), where N is the population size, D is the dimension of the problem, and T is the maximum number of iterations. In our CSP-GWO algorithm, the parallel processing capability of the P system allows multiple populations to be processed simultaneously, effectively reducing the time complexity. The inner membrane structure with four neurons enables concurrent evaluation of multiple candidate solutions.
The space complexity of our algorithm is O(N × D), which is comparable to standard GWO. However, the distributed nature of the P system architecture allows for efficient memory utilization, making it suitable for large-scale optimization problems. The improved convergence speed of CSP-GWO means that fewer iterations are required to achieve high-quality solutions, further reducing the overall computational cost.
9. Robustness Analysis
We evaluated the robustness of CSP-GWO by analyzing the variance of results across multiple independent runs. The trajectory cost variance in Table 2 shows that our algorithm achieves the lowest variance among all tested methods for all three UAV configurations (4, 8, and 16 UAVs). This indicates that CSP-GWO produces consistent results with minimal variation between runs, demonstrating high reliability for practical applications.
The robustness of our algorithm stems from the multi-population mechanism in the P system, which provides multiple independent search processes. This redundancy ensures that even if some populations become trapped in local optima, others can continue to explore the search space effectively. The fuzzy weight adjustment also contributes to robustness by dynamically balancing exploration and exploitation throughout the optimization process.
10. Conclusion
In this research, we successfully developed a novel multi-UAV trajectory planning algorithm by integrating cell-like spiking neural P systems with grey wolf optimization. Our CSP-GWO algorithm leverages the parallel computing characteristics of P systems to model the social hierarchy and hunting behavior of grey wolves, achieving efficient multi-UAV path planning. The algorithm dynamically adjusts search strategies during optimization, effectively balancing global exploration and local exploitation.
Extensive simulation experiments demonstrate the superiority of CSP-GWO across multiple standard benchmark functions and practical trajectory planning tasks. For 4, 8, and 16 UAVs, our algorithm achieved average trajectory cost reductions of 17.9%, 15.1%, and 11.7% compared to standard GWO, demonstrating enhanced optimization accuracy and robustness. The runtime advantage of CSP-GWO is particularly noteworthy, as the algorithm execution time does not increase significantly with the number of UAVs, highlighting the benefits of parallel computation in P systems.
The China drone industry continues to advance rapidly, with increasing demand for sophisticated multi-UAV systems capable of operating autonomously in complex environments. Our CSP-GWO algorithm provides a valuable contribution to this field by offering a solution that combines high optimization quality with computational efficiency. The parallel processing capability of the P system architecture makes our algorithm particularly suitable for real-time trajectory planning applications.
We acknowledge that our current multi-UAV cooperative trajectory planning model only considers known static threat sources and does not account for dynamic threats. Future research will focus on extending the model to incorporate dynamic threat sources and further explore the application of our algorithm in more complex environments. We also plan to investigate the integration of additional constraints, such as communication constraints and energy consumption considerations, to make the model more comprehensive for practical China drone operations.
The successful application of cell-like spiking neural P systems to multi-UAV trajectory planning opens new avenues for research in natural computing and autonomous systems. The inherent parallelism of membrane systems offers significant advantages for complex optimization problems, and we believe that further exploration of this approach will yield additional benefits for the China drone industry and beyond.
In conclusion, our CSP-GWO algorithm represents a significant advancement in multi-UAV trajectory planning technology. By combining the computational power of P systems with the optimization capability of grey wolf algorithms, we have developed a solution that is both effective and efficient. As the China drone industry continues to grow and evolve, algorithms like CSP-GWO will play an increasingly important role in enabling sophisticated multi-UAV operations across a wide range of applications.