Interaction Models and Automated Control under Partial Observable Environments

Abstract

The problem of automatically constructing a software component such that when executed in a given environment satisfies a goal, is recurrent in software engineering. Controller synthesis is a field which fits into this vision. In this paper we study controller synthesis for partially observable LTS models. We exploit the link between partially observable control and non-determinism and show that, unlike fully observable LTS or Kripke structure control problems, in this setting the existence of a solution depends on the interaction model between the controller-to-be and its environment. We identify two interaction models, namely Interface Automata and Weak Interface Automata, define appropriate control problems and describe synthesis algorithms for each of them.