Abstract: To address the issues of low exploration efficiency, slow convergence speed and poor path smoothness in unmanned aerial vehicle (UAV) path planning under complex environments, an improved path planning algorithm based on the deep deterministic policy gradient (DDPG) framework is proposed, which integrates curriculum learning and an embedded prioritized replay mechanism. First, a curriculum of obstacle environments ranging from easy to difficult is designed to guide the UAV to gradually learn path planning tasks from simple to complex, thereby improving training efficiency and stability. Then, the prioritized replay mechanism is embedded into the algorithm to ensure rapid extraction of effective experiences from the complex experience replay buffer, further enhancing convergence speed and stability, and ensuring higher path efficiency and better smoothness. Simulation results demonstrate that the reinforcement learning method integrating curriculum learning and the embedded prioritized replay mechanism can effectively improve the autonomous obstacle avoidance and path planning capabilities of UAVs in unknown complex environments. Compared with the traditional DDPG algorithm, the proposed method increases path planning efficiency by 26.84% and improves path smoothness by 66.19%, while achieving faster convergence speed during training and better stability in the later stage of training.