To improve the order picking efficiency of the robotic mobile fulfillment systems (RMFS), this paper investigates the RMFS replenishment items storage assignment problem. A mixed integer programming model is developed with the objective of maximizing the correlation degree among items in all pods. To solve the problem, a large neighborhood search algorithm is designed for the solution. The initial feasible solution is generated by a greedy algorithm. Combining the characteristics of the problem, the destroy operator and repair operator are proposed to improve the solution. Numerical experiments show that the large neighborhood search algorithm has good performance. In medium- and large-scale instances, the large neighborhood search algorithm can effectively improve the solution quality by 37.4% and 21.5% respectively based on the initial solution generated by the greedy algorithm, and it outperforms the variable neighborhood search algorithm with an average improvement of 8.9% and 10.3%. Moreover, the sensitivity analysis experiments analyze the effects of the number of pods, the number of slots, and the scattered degree of items on the objective value.