Abstract: Objectives: Autonomous underwater vehicles (AUV) play an increasingly vital role in ocean exploration, resource surveying, and national defense. Effective gravity-aided navigation can only be ensured when the submersible operates within adaptive regions characterized by significant gravity field variations. Consequently, during the gravity matching navigation phase, path planning that ensures the submersible remains within adaptive areas while avoiding non-adaptive areas is of critical importance. Existing path planning approaches exhibit inherent limitations. Although the A^* algorithm can rapidly obtain a geometrically shortest path, the resulting paths often contain excessive turning points and exhibit poor smoothness. In contrast, Traditional Ant Colony Optimization (T-ACO) demonstrates strong global search capability but suffers from slow convergence and initial search blindness due to the uniform distribution of pheromones at algorithm initialization. To overcome these deficiencies, a hybrid algorithm is proposed to exploit the complementary advantages of both methods, aiming to achieve faster convergence, higher path quality, and improved computational efficiency for gravity matching navigation path planning. Methods: A hierarchical and progressively enhanced algorithm, termed A^*-Guided Improved Elite Ant Colony Optimization (A^*-ACO), is proposed. Built upon a baseline algorithm that already integrates a dynamic parameter strategy, the approach first introduces an elite strategy to optimize the pheromone update criterion to accelerate convergence and combines a path-smoothing factor to improve the transfer probability model to optimize the geometric quality of the paths. Furthermore, to address the initial search blindness of the ant colony algorithm, the geometrically shortest path pre-planned by the A^* algorithm is incorporated as prior heuristic information, guiding the search direction of ants by reconstructing the state transition probability model. This effectively overcomes the blindness of early-stage search and significantly enhances both the global optimization capability and convergence efficiency of the algorithm. Results: To evaluate the effectiveness of the algorithm, ten repeated experiments were conducted in three typical local regions of different scales. Four algorithms were compared: T-ACO, Improved Elite Ant Colony Optimization (I-ACO), A^*, and A^*-ACO. In Region 1, A^*-ACO converged in only 4 iterations, while achieving the same optimal path length as A^* (30.08) with markedly smoother geometry. Compared with I-ACO, computational efficiency improved by approximately 20.4%, and the path length was reduced by 4.3%. In Region 2, A^*-ACO converged in 3 iterations with a path length of 30.38.Compared with I-ACO, the number of iterations to convergence decreased by approximately 90.6%, and computational efficiency improved by approximately 28.4%.In the large-scale Region 3, A^*-ACO achieved an average path length of 63.47, which was shorter than those of T-ACO, I-ACO, and A^* by 8.99%, 3.53%, and 1.72%, respectively, while converging in only 8 iterations. Across all three regions, compared with T-ACO, the proposed A^*-ACO reduced the average optimal path length by 5.9%–9.0%, decreased the average number of iterations to convergence by over 90%, and improved computational efficiency by approximately 45%–67%.Conclusions: By leveraging the rapid convergence characteristics of the A^* algorithm and the multi-objective optimization capability of ant colony optimization, the A^*-ACO algorithm achieves a synergistic integration of efficient heuristic search and robust swarm-intelligence-based exploration, providing a solution with faster convergence and superior path quality for underwater gravity matching navigation path planning.