Representation of Local Space in Perception/Action Systems: Behaving Appropriately in Difficult Situations

The effective field of view paradigm: adding representation to a reactive system

“…The agent decides if a given object should be in view, based on the object's location (its where component), the azimuth of the optic axis of the agent's current absolute field of view, and knowledge of the size of the absolute field of view. Ego-motion transformations are applied to the coordinates of all objects which the agent does not expect to find in the absolute field of view [9]. In figure 1b, as the agent moves toward the rhino, the rhino's position is determined from the absolute field of view, while the jeep's location is calculated from ego-motion transforms.…”

“…The effective field of view is a unified, task-oriented description of the agent's environment built from current sensor data, the agent's expectations and a carefully maintained collection of recent sensor values [9].…”