A robot vision navigation method using deep learning in edge computing environment

Abstract In the development of modern agriculture, intelligent use mechanical equipment is one main signs for agricultural modernization. Navigation technology key machinery to control autonomously in operating environment, and it a hotspot field research on machinery. Facing accuracy requirements autonomous navigation robots, this paper proposes visual algorithm robots based deep learning image understanding. The method first uses cascaded convolutional network hybrid dilated convolution fusion process images collected by vision system. Then, extracts route processed improved Hough transform algorithm. At same time, posture adjusted realize navigation. Finally, our proposed verified using non-interference experimental scenes noisy scenes. Experimental results show that can perform complex environments has good practicability applicability.