UAV autonomous inspection plays an important role in high-speed rail inspection. In order to improve the efficiency and scope of UAV work, the obstacle avoidance problem of UAV in the process of autonomous inspection must be solved. The obstacle avoidance algorithm was mainly studied, and an obstacle avoidance scheme based on the combination of IRIS (iterative recombination) and APF (artificial potential field) was proposed. According to the requirements and constraints of UAV autonomous inspection, the IRIS and APF were optimized and effectively combined to achieve the application of IRIS algorithm in the UAV obstacle avoidance, which ensured the accurate direction of the UAV toward the target point. Through the simulation and experimental verification, the proposed IRIS-APF obstacle avoidance scheme not only achieves the goal of avoiding obstacles, but also achieves a higher algorithm speed and accuracy.