In view of the inaccurate calculation of the optimal path scheduling parameters for transportation vehicles as well as the subsequent longer transportation time with the conventional transportation vehicle partition scheduling path planning method adopted, a partition scheduling path planning has thus been proposed for large scale automatic sorting warehouse transport vehicles. By using neural network approximation algorithm to fit the Q-values in large-scale states, a transportation vehicle partition scheduling path planning model is established, with the population size calculated, and with the individuals with the highest crowding degree selected. By using immune genetic algorithm to calculate the optimal path scheduling parameters, the order of goods pickup is exchanged according to the parameters so as to obtain approximate output action instructions in the near state, with the transport vehicle scheduled according to this action instruction for a completion of the planning of the transport vehicle partition scheduling path. The experimental results show that under the scheduling path planned by the currently proposed method, the completion time of transportation vehicle tasks is relatively short, indicating that the method is characterized with a good application effect.