The invention discloses a method for realizing visual motion control by using a deep convolutional neural network, and the method comprises the steps: obtaining a current image of a mobile platform only by using a monocular camera, and only analyzing a currently obtained single-frame image, thereby obtaining a reasonable motion control plan under a current visual angle. Positioning information ofthe mobile platform does not need to be obtained through multi-image association of multiple frames of images, and meanwhile feature point data matching and plane mapping do not need to be carried out. In addition, different from an existing visual operation control scheme based on a neural network, road sign recognition and feature point extraction are not needed. Furthermore, semantic segmentation is performed on the currently acquired image by utilizing the semantic segmentation capability of the deep neural network. The segmentation result is directly used for motion target point planning, the single-frame image is input into the motion control target point planning, and the whole process does not need the input of human priori knowledge. The method has the advantages of low cost, simplicity and rapidness in operation control and the like.