To address challenges such as dynamic obstacles, temporary no-fly zones, multi-drone collaborative conflicts, and energy consumption constraints in complex urban environments, a dynamic risk-averse route planning method for multi-drone collaborative delivery is proposed. First, a collaborative delivery model for multi-distribution centers is constructed, integrating a 3D grid-based method to quantify obstacle risks and energy constraints. The urban airspace is stratified into three flight levels, and a dynamic risk update mechanism is designed. Second, a Hybrid A*-Whale Optimization Algorithm (Hybrid A*-WOA) is proposed, combining the improved A* algorithm’s local route search capability with the global optimization characteristics of the Whale Optimization Algorithm. This hybrid approach achieves collaborative optimization of task allocation and dynamic obstacle avoidance for multi-drone systems. Furthermore, a flight-level comprehensive evaluation function is introduced to balance the risk and time cost through weight coefficients and dynamically select the optimal flight altitude.