Switching tasks and flexible reasoning in the Situation Calculus

Abstract

In this paper we present a new framework for modelling switching tasks and adaptive, flexible behaviours for cognitive robots. The framework is constructed on a suitable extension of the Situation Calculus, the Temporal Flexible Situation Calculus (TFSC), accommodating Allen temporal intervals, multiple timelines and concurrent situations. We introduce a constructive method to define pattern rules for temporal constraint, in a language of macros. The language of macros intermediates between Situation Calculus formulae and temporal constraint Networks. The programming language for the TFSC is TFGolog, a new Golog interpreter in the Golog family languages, that models concurrent plans with flexible and adaptive behaviours with switching modes. Finally, we show an implementation of a cognitive robot performing di erent tasks while attentively exploring a rescue environment