In order to realize that air traffic controllers can directly issue instructions to control unmanned aerial vehicles (UAVs), combined with the application of deep learning in natural language processing (NLP), a method of unmanned aerial vehicle (UAV) command intention recognition based on deep learning is proposed. The improved Skip-Gram model is used to generate the word vector of the instruction text, which is input into the convolutional neural network to classify the instruction text, and the intention of the air traffic controller to issue the instruction is obtained. The experimental results show that the method can accurately identify the command intention, which is helpful for the realization of the subsequent command understanding technology and for the further direct interaction between air traffic controllers and UAVs.