The booming development of the low-altitude economy is rapidly promoting the large-scale application of unmanned aerial vehicle (UAV) delivery. To address the challenges of cost and safety in multi-UAV coordinated delivery within three-dimensional low-altitude airspace, this paper proposes a bi-level optimization model. This model achieves the joint optimization of flight trajectory planning and mission scheduling. The upper-level model aims to minimize a comprehensive cost function that integrates flight distance, altitude changes, obstacle avoidance, and turn penalties. An Improved Gold Rush Optimizer (IGRO) is designed to solve for the optimal trajectories and the resultant distance matrix. The lower-level model formulates the objective of minimizing total delivery cost under constraints including UAV payload, range, and soft time windows. An Improved Adaptive Large Neighborhood Search (IALNS) algorithm is developed to determine the optimal cooperative delivery schedule for the UAV fleet. Simulation results demonstrate the effectiveness of the proposed approach. In trajectory planning, the IGRO algorithm achieves an average reduction of 7.99% in total cost compared to the original method while exhibiting faster convergence. For mission scheduling, the IALNS algorithm outperforms other benchmark algorithms across varying task scales. Furthermore, comparison experiments involving diverse UAV types, varying numbers of dispatch centers, and different scheduling modes validate the robustness and generalizability of the proposed bi-level planning method.