A novel unmanned combat aerial vehicle path planning method: multi-strategy improved salp swarm algorithm with bi-stage model

• Published in: Cluster computing Vol. 28; no. 12; p. 807
• Main Authors: Fan, Jingkun, Jia, Yingjuan, Qu, Liangdong, Li, Xiaoqin
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.11.2025; Springer Nature B.V
• Subjects: Algorithms; Combat aircraft; Computer Communication Networks; Computer Science; Convergence; Energy consumption; Learning; Military operations; Operating Systems; Optimization; Path planning; Processor Architectures; Strategy; Unmanned aerial vehicles; Waypoints; Salp swarm algorithm; Global optimization; UCAV path planning; Bi-stage path planning model
• ISSN: 1386-7857; 1573-7543
• DOI: 10.1007/s10586-025-05528-1

The increasing deployment of unmanned combat aerial vehicles (UCAVs) in modern military operations demands secure and efficient path planning, yet the complexity of aerial environments and the strategic distribution of waypoints pose significant challenges. This study proposes a novel bi-stage model to address the waypoint distribution problem, where the first stage generates an initial path and the second stage refines it by adjusting the waypoints. Additionally, a multi-strategy improved salp swarm algorithm (MSSSA) is designed to address the UCAV path planning problem, incorporating three enhanced strategies. Firstly, a consensus optimization mechanism is employed to mitigate the risk of premature convergence in the MSSSA. Secondly, a fitness-weighted learning strategy is devised to equilibrate the algorithm’s exploration and exploitation capabilities. Thirdly, a quasi-opposition-based learning strategy is utilized to expedite the convergence process of the MSSSA. Numerical experiment results demonstrate that MSSSA exhibits superior optimization performance and robustness compared to ten state-of-the-art algorithms. By integrating the bi-stage model with MSSSA to form bi_MSSSA, experimental results on UCAV path planning clearly show that the paths generated by bi_MSSSA significantly outperform those produced by other algorithms.