This article proposes an improved potential function based-RRT* (P-RRT*) and modified dynamic window approach (DWA) for unmanned aerial vehicle (UAV) dynamic trajectory planning. Firstly, an optimized sampling region is designed to increase sampling point effectiveness in P-RRT*, thereby speeding up the generation of global paths. Simultaneously, the greedy algorithm and B-spline curves are utilized to optimize the smoothness of the global route. Then, the modified DWA enhances the primitive evaluation function and designs new distance evaluation terms for different obstacle categories, accompanied by corresponding velocity evaluation terms, with the goal of better aligning with dynamic environmental conditions. Finally, this hybrid algorithm ensures efficiency and safety during flight.