Aiming at the complex environment of indoor environment where a lot of sundries are stored or outdoor urban scene with many high-rise buildings, this paper proposes a method, which uses Theta* algorithm for global path planning and improved Dynamic Window Approach (DWA) for tracking path nodes, to realize the path planning of the quad-rotor UAVs in three-dimensional environment.Compared with the method of making global static path planning first and then making path C2 continuity optimization, this method can realize the avoidance of unknown obstacle information in the map, and the generated curve conforms to the dynamic constraints of UAV, which ensures that the UAV can follow the path without attitude disorder caused by the drastic change of speed.Finally, the algorithm not only generates the planned obstacle-avoidance path, but also generates the speed change of the motor.The proposed method can realize trajectory tracking only by tracking the change of motor speed without interference, which simplifies the control of UAV.