With explosive attention on data-sensing services, the unmanned aerial vehicle (UAV) has been applied to accelerate the information collection. Particularly, as for the deadline-sensitive data, the age of information (AoI) has become a significant metric to avoid messages expiration. In this letter, we propose a UAV trajectory planning model for data collection with minimizing expired data packets in the whole sensor system, and then relax the obscure original problem into a min-max-AoI-optimal path scheme due to complex constraints. We utilize a reinforcement learning-based strategy for the solution and validate its effectiveness with comparison of other two benchmarks in simulations.