Hypothesising Object Relations from Image Transitions
Paulo Santos and Murray Shanahan
Abstract
Abstract. This paper describes the construction of a qualitative spatial
reasoning system based on the sensor data of a mobile robot. The spatial
knowledge of the robot is formalised in three sets of axioms. First of all,
axioms for relations between pairs of spatial regions are presented. Assuming
the distance between regions as a primitive function in the language, the
main purpose of this initial axiom set is the classification of relations
between images of objects (from the robot’s vision system) according to their
degree of displacement. Changes in the sensor data, due to the movement either
of objects in the robot’s environment or of the robot itself, are represented
by transitions between the displacement relations. These transitions are
formalised by the second set of axioms. The predicates defining the transitions
between image relations are connected to possible interpretations for the
sensor data in terms of object-observer relations, this issue is handled
by the third set of axioms. These three axiom sets constitute three layers
of logic-based image interpretation via abduction on transitions in the sensor
data.