Abstract: In some Internet of Things（IoT） scenarios that require high information freshness, UAV can be used to assist in collecting device data due to their high mobility and flexibility. The existing evaluation of information freshness has certain limitations in terms of age and cannot be accurately defined from a content perspective. This article uses the age of outdated information to characterize the freshness of information, and proposes a UAV-assisted IoT data collection scheme based on the curiosity driven DQN algorithm. By optimizing the trajectory of the UAV, the age of outdated information is minimized while meeting the constraints of UAV energy consumption. The simulation results show that compared to the traditional DQN algorithm, the proposed algorithm enhances the exploration ability of the intelligent agent, accelerates the convergence speed by about 45%, and achieves a reward value of about 67% higher.