Road Scene Recognition of Forklift AGV Equipment Based on Deep Learning

The application of scene recognition in intelligent robots to forklift AGV equipment is great significance order improve the automation and intelligence level distribution centers. At present, using camera collect image information obtain environmental can break through limitation traditional guideway positioning equipment, beneficial path planning system expansion later stage warehouse construction. Taking center as research object, this paper explores based on a deep convolution neural network. On basis characteristics environment, semantic segmentation network applied environment established, method suitable for proposed, so that use learning learn features achieve accurate large-scale without adding landmarks, which provides an effective model environment. activation function layer studied by with better gradient performance. results show performance H-Swish than ReLU accuracy computational complexity, it save costs calculation form mobile terminal.