A Doxastic Characterisation of Autonomous Decisive Systems

A highly autonomous system (HAS) has to assess the situation it is in and derive beliefs, based on which, decides what do next. The beliefs are not solely observations HAS made so far, but also general insights about world, which operates. These have either been built during design or provided by trusted sources its mission. Although may be imprecise might bear flaws, will extrapolate possible futures order evaluate consequences of actions then take decisions autonomously. In this paper, we formalize an decisive as a that always chooses currently believes best. We show can checked whether given application domain, dynamically changing knowledge base list LTL mission goals. moreover synthesize belief formation for system. For formal characterization, use doxastic framework safety-critical HASs where supports HAS's extrapolation.