Grabbing pose generation method based on convolutional neural network

The grabbing pose generation method based on the convolutional neural network comprises the steps of determining a representation method of a grabbing position, performing data enhancement processing on an original data set, constructing an optimal grabbing pose generation network model based on the convolutional neural network RS-ConvNet, and generating the grabbing pose based on the optimal grabbing pose generation network model. And the enhanced data set is adopted to train an optimal grabbing pose generation network model based on a convolutional neural network RS-ConvNet, an image obtained by a camera is input into the trained optimal grabbing pose generation network model based on the RS-ConvNet to be predicted, and after the optimal grabbing pose is obtained, a mechanical arm is made to execute the grabbing action through coordinate transformation. The method has good accuracy and high detection speed, meets the requirements for the grabbing precision and grabbing speed of irregular commodities in a re