This dissertation presents a model of the knowledge a person has about the spatial structure of a large-scale environment: the "cognitive map." The functions of the cognitive map are to assimilate new information about the environment, to represent the current position, and to answer route-finding and relative-position problems. This model (called the TOUR model) analyzes the cognitive map in terms of symbolic descriptions of the environment and operations on those descriptions. Knowledge about a particular environment is represented in terms of route descriptions, a topological network of paths and places, multiple frames of reference for relative positions, dividing boundaries, and a structure of containing regions. The current position is described by the "You Are Here" pointer, which acts as a working memory and a focus of attention. Operations on the cognitive map are performed by inference rules which act to transfer information among different descriptions and the "You Are Here"...

Abstract: Natural language requests involving vague spatial concepts are not easily communicated to a GIS because the meaning of spatial concepts depends largely on the contexts (such as task, spatial contexts, and user’s personal background) that may or may not be available or specified in the system. To address such problems, we developed a collaborative dialogue approach that enables the system and the user to construct shared knowledge about relevant contexts. The system is able to anticipate what contextual knowledge must be shared, and to form a plan to exchange contextual information based on the system’s belief on who knows what. To account those user contexts that are not easily communicated by language, direct feedback approach is used to refine the system’s belief so that the intended meaning is properly grounded. The approach is implemented as a dialogue agent, GeoDialogue, and is illustrated through an example dialogue involving the communication of the vague spatial concept near. 1

Interaction between robots and humans should be at a level which is accessible and natural for human operators. A lexicon template is proposed for specifying commands for 2D mobile robot navigational tasks. The language lexicon is a minimal spanning semantic set for human 2D navigational tasks. The task command lexicon consists of a verb, destination, direction and a speed. The destination is a location in the environment defined by a geometric model positioned at a particular spatial location in a globally-referenced Cartesian coordinate space. The task command lexicon has been used as a language template for specifying commands for the SPOTT mobile robot control architecture. It is not the intent that the template serve as the only commands that the robot recognizes, but rather as an internal language that gets mapped onto planning and control constructs. A speech recognition system maps spoken commands onto the proposed internal natural language command template...