Abstract: Considering the uneven distribution of sensor nodes, an unmanned aerial vehicle(UAV) formation assisted sensor node localization scheme(UFSNL) is proposed. A UAV formation hovering point selection algorithm based on Q-learning was proposed. The ant colony optimization was used to plan the shortest flight path through all hovering points. The maximum likelihood estimation method was used to estimate the coordinates of unknown nodes. Simulation results show that when the level of uneven node distribution increases, the proposed scheme can ensure localization accuracy and localization ratio, reduce the energy consumption of UAV formation and localization time.