Probabilistic epistemic reasoning about actions

Modelling agents that are able to reason about actions in an ever-changing environment continues to be a central challenge in Artificial Intelligence, and many technical frameworks that tackle it have been proposed over the past few decades. This thesis deals with this problem in the case in which the envi- ronment and its evolution is incompletely known, and agents can seek to gain further information about it and act accordingly. Two languages are proposed, namely PEC+ and EPEC, which extend a standard logical language for reasoning about actions known as the Event Calculus, and use Probability Theory as a measure of the agent’s degree of belief about aspects of the domain. These languages are then shown to satisfy some essential properties. PEC+ is implemented and tested against a number of real world scenarios