Abstract: To address the challenge that classical evolutionary algorithms struggle to balance convergence speed and global search capability in 3D path planning for unmanned aerial vehicles (UAVs), this paper proposes a Q-learning guided hybrid continuous ant colony optimization (ACOAGR-GA) path planning method.First, a hybrid population initialization strategy combining Q-Learning and random initial solutions is designed to optimize the initial pheromone distribution of the ant colony, providing high-quality initial paths for UAVs. Then, a Gaussian-Lévy hybrid walk strategy is introduced into the ant colony algorithm to achieve dynamic step size during path planning, which helps the algorithm escape from local optima. Finally, the adaptive node selection of the ant colony algorithm is optimized, and a fitness function is constructed via a genetic elite selection mechanism to fully utilize all ant individuals, thus accelerating convergence and enhancing global search performance.Comparative experiments are conducted on 4 task scenarios built based on the open-source digital elevation model from Phung et al., with 5 typical algorithms as benchmarks. The results demonstrate that the proposed method achieves the optimal total path cost. Compared with contrast algorithms, its convergence accuracy is significantly improved, and it possesses excellent local optimum escape ability, with only a slight reduction in convergence speed.