Visual Navigation Based on Language Assistance and Memory

In order to solve outdoor mobile robots’ dependence on geographic information systems, and realize automatic navigation in the face of complex changeable scenes, we propose a method that selects landmark adds prompt guidance so robot can navigate relying visual-language memory. Visual-language guide direction robot’s movement, obeying annotation people according its memory scene, which refers strategy selecting passed-by landmarks for route remembering scene features. When passing it, agent ascertain position match it carry out action. Experiments showed our proposed achieve purpose independent without GIS, is superior existing methods.