An approach to high-level interaction with autonomous robots by means of schematic maps is outlined. Schematic maps are knowledge representation structures to encode qualitative spatial information about a physical environment. A scenario is presented in which robots rely on high-level knowledge from perception and instruction to perform navigation tasks in a physical environment. The general problem of formally representing a physical environment for acting in it is discussed. A hybrid approach to knowledge and perception driven navigation is proposed. Different requirements for local and global spatial information are noted. Different types of spatial representations for spatial knowledge are contrasted. The advantages of high-level / low-resolution knowledge are pointed out. Creation and use of schematic maps are discussed. A navigation example is presented.