Abstract: To address the key technical challenges of limited route-planning efficiency, insufficient path smoothness, and inadequate real-time dynamic conflict resolution during multi-UAV cooperative operations within virtual tubes in complex urban low-altitude environments, this paper proposes a systematic multi-UAV route-planning and conflict resolution framework specifically designed for virtual tube-constrained operations. With the rapid commercialization of urban low-altitude logistics and urban air mobility (UAM), the uncoordinated operation of large numbers of unmanned aerial vehicles (UAVs) in densely built environments has become a major contributor to flight safety incidents. Virtual tube technology is widely regarded as a promising approach for structured low-altitude airspace management. However, existing methods cannot simultaneously meet the dual requirements of high-efficiency route planning under strict boundary constraints and real-time dynamic conflict resolution in multi-UAV systems, thereby severely restricting the large-scale application of virtual tube technology. To overcome these challenges, this study develops a structured virtual tube-based airspace model for urban low-altitude environments characterized by dense buildings, complex obstacle distributions, and diverse operational entities. Specifically, a three-dimensional urban building threat model incorporating a safety expansion mechanism is established to mitigate collision risks arising from UAV positioning errors and airflow disturbances. The model is mathematically defined and divided into a core constraint zone for normal steady flight and an elastic adjustment zone for emergency maneuvers. On this basis, a multi-objective route optimization model with strict inequality constraints was formulated, with minimization of total path length, cumulative heading angle change (to characterize path smoothness), and safety risk cost as the optimization objectives. This formulation transforms the unstructured low-altitude environment into a navigable space with well-defined boundary constraints and sufficient maneuvering margins. To overcome the limitations of traditional Bidirectional RRT algorithms, including massive invalid sampling, slow convergence, and poor path feasibility under virtual tube constraints, this study incorporates four core improvement mechanisms into the Bidirectional RRT^* framework: ellipsoidal confined sampling, artificial potential field (APF)-based target guidance, alternating exploration expansion, and real-time tube boundary verification. These enhancements collectively constitute the Bi–APF–RRT^* global route-planning algorithm. This algorithm substantially improves route search efficiency, path feasibility, and trajectory quality in complex, constrained environments while markedly reducing the computational burden associated with invalid sampling and redundant expansion. Furthermore, to address dynamic conflict issues in multi-UAV cooperative operations, this paper proposes an adaptive large-range sampling-based conflict-detection method. The method dynamically adjusts the prediction time window and sampling step size based on the relative distance and speed between UAVs, thereby balancing detection accuracy and computational efficiency. Conflicts are categorized into long-, medium-, and short-range levels based on the predicted collision time. Building on this framework, a three-layer hierarchical conflict-resolution strategy consisting of a global route layer, local replanning layer, and maneuver control layer was constructed, adopting speed adjustment, heading fine-tuning, and height adjustment strategies for different conflict levels to realize dynamic safety separation maintenance and local maneuver coordination of UAVs without breaking the virtual tube constraints. To rigorously validate the proposed approach, extensive simulation experiments were conducted using MATLAB 2024a. The evaluation included static route-planning tests in three representative urban environments (simple, medium, and complex), as well as dynamic conflict-resolution tests involving multi-UAV scenarios. The results demonstrate that the proposed algorithm consistently generates feasible routes that meet virtual tube constraints across all environments, exhibiting strong environmental adaptability, path continuity, and operational robustness. Compared with conventional algorithms, the proposed method reduced the average path length by approximately 13% and the average cumulative turning angle by approximately 25.7% in moderately complex environments, demonstrating superior performance in path compactness and smoothness. In a 10-UAV cooperative scenario, the proposed framework achieved a 98.2% conflict-resolution success rate, with an average replanning time of only 126 ms and a minimum safety distance of 6.2 m, thereby satisfying low-altitude UAV flight safety requirements. Overall, this research provides a robust technical support for UAV route organization, operational safety guarantees, and high-density cooperative flight in low-altitude logistics and UAM scenarios.