To address the challenges of collaborative navigation for cluster systems within cross-domain scenarios, a novel adaptive clustering method of collaborative navigation for cross-domain swarm system is proposed. By partitioning cross-domain clusters into distinct spatial domains, an adaptive clustering method for cross-domain systems, which combines centralized and distributed clustering algorithms, is designed to optimize clustering problems for large-scale cluster system in low-altitude domain. Targeting inter-domain as well as intra-cluster navigation scenarios, specific collaborative navigation models are formulated based on the relative distance measurement between cluster nodes, which enhances collaborative navigation performance for cluster nodes in low-altitude domain. Simulation results demonstrate that the proposed scheme outperforms traditional collaborative navigation methods, which improves the performance of collaborative navigation for cluster systems in cross-domain. Furthermore, its resilience to environmental challenges in low-altitude domain underscores its significant practical utility and contributions.