Simple Default Reasoning in Theories of Action

We extend a recent approach to integrate action formalisms and non-monotonic reasoning. The resulting framework allows an agent employing an action theory as internal world model to make useful default assumptions. While the previous approach only allowed for modeling static defaults, that are independent of state properties, our extension allows for the expression of dynamic defaults. Problems that arise due to the interaction of defaults with the solution of the frame problem are dealt with accordingly: we devise a general method of integrating defaults into the formal representation of action effects and show that the method prevents counter-intuitive conclusions.

• Type: Conference
• Year: 2009
• Where: AUSAI
• Authors: Hannes Strass, Michael Thielscher
• Keywords: Artificial Intelligence, AUSAI 2009, Dynamic Defaults, Static Defaults, Useful Default Assumptions