Aiming at the problems of slow search speed and many redundant nodes of traditional A* algorithms in UAV trajectory planning when facing complex environments and high-dynamic tasks, a 3D UAV trajectory planning method based on the combination of optimized bi-directional A* algorithm and artificial potential field is proposed. By adopting a bidirectional search mechanism and setting weight coefficients to optimize the heuristic function, introducing an adaptive step-size strategy, and dynamically adjusting the bidirectional A* search step-size by using an adjustment factor, the Bi-A*PF algorithm is proposed by comprehensively considering the global planning and real-time obstacle avoidance requirements. Simulation experiments show that compared with traditional trajectory planning methods, the Bi-A*PF algorithm not only enables the UAV to efficiently plan a desired trajectory in a 3D environment, but also effectively avoids unexpected threats.