计算机科学
数学优化
运动规划
粒子群优化
路径(计算)
过程(计算)
人口
遗传算法
加速度
元启发式
蒙特卡罗方法
适应度函数
局部最优
惯性
能量(信号处理)
无人地面车辆
群体行为
算法
功能(生物学)
弹道
混合算法(约束满足)
局部搜索(优化)
转弯半径
进化算法
模拟
实时计算
健身景观
搜索算法
群体智能
路径长度
作者
Pamarthi Venkatasivarambabu,P. Ravi Sankar,Thulasi Bikku,Arepalli Tirumala
标识
DOI:10.1007/s44196-026-01329-5
摘要
Path planning for multiple unmanned aerial vehicles (UAVs) in dynamic environments remains challenging due to the simultaneous satisfaction of multiple critical constraints. Each UAV must keep at least 5 m of space between it and other UAVs while avoiding both static and moving obstacles. Also, the resulting paths should be short, smooth, and easy to calculate so that ground stations or high-performance onboard systems can frequently replan in near-real-time. In such scenarios, traditional metaheuristic algorithms frequently underperform. The problem’s high dimensionality, non-convexity, and time-varying nature often cause it to converge too soon to suboptimal solutions or require too much computing power, making it hard to use in practice. This study proposes a sequential hybrid optimisation framework to tackle these issues. The process starts with an Improved Particle Swarm Optimisation (IPSO) that uses adaptive inertia weight and linearly changing acceleration coefficients. This strikes a good balance between exploration and exploitation, quickly generating high-quality initial path candidates. Then comes the Whale Optimisation Algorithm (WOA) refinement stage, which uses its encircling, spiral bubble-net feeding, and random search operators to make the population more diverse and avoid getting stuck in local optima. A centralised fitness function reduces the total length of the path while adding penalties for things such as excessive curvature, proximity to obstacles, or high energy consumption. In 100 × 100 × 50 m³ environments involving 4–8 UAVs and 6–12 dynamic obstacles, Monte Carlo simulations (50 runs per scenario) demonstrate that the IPSO–WOA hybrid yields path lengths 42–52% shorter than those produced by Genetic Algorithm (GA), reduces average computation time by 14–18% (from 8.55 s to 7.12 s), achieves success rates up to 88%, and improves minimum inter-vehicle clearance (6.1 ± 0.7 m vs. 4.2 ± 1.1 m for GA). Wilcoxon signed-rank tests ( p < 0.01) confirm statistical significance. These results highlight the effectiveness of complementary metaheuristic hybridisation for real-time multi-UAV coordination in applications such as search-and-rescue, precision agriculture, surveillance, and infrastructure inspection.
科研通智能强力驱动
Strongly Powered by AbleSci AI