Path planning strategies for multi-UAV swarm systems: algorithms, challenges and future directions

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Multi-drone systems demonstrate efficient coverage and task parallelism in fields such as disaster assessment, public safety, and environmental monitoring. However, path planning often involves multiple objectives, constraints, and environmental dynamics, and its complexity significantly limits the collective performance of the swarm. In this paper, within the unified MDP/PO-MDP framework, we systematically review the five main approaches for multi-drone path planning: classical algorithms, heuristic algorithms, meta-heuristic algorithms, machine learning algorithms, and hybrid algorithms. For each algorithm category, an eight-dimensional evaluation system was developed, covering time efficiency, computational cost, system complexity, convergence speed, adaptability to complex environments, scalability over large areas, trajectory smoothness, and obstacle avoidance capability, followed by a cross-comparison of representative algorithms. Based on the comparative analysis, this paper proposes the “method hybridization + directed optimization” paradigm, which enhances global-local coordination through strategies such as learning priors, swarm intelligence parallel search, and geometric pruning, and anticipates future research directions like learning-enhanced provable optimization, native 4D conflict-free trajectory planning, and decentralized collaborative decision-making, offering references for advancing multi-drone path planning from “feasible” to “usable.”

This publication has 71 references indexed in Scilit: