The Golog family of programming language, based on the situation calculus, has proven to be successful in the area of high level
autonomous agent control. However, the current Golog systems are quite slow, scribing to the Prolog-based implementation and blind search for
look-ahead. Additionally, agents always run in some restricted environments where exists some particular knowledge that can help to
enlighten agents' reasoning process. Since this kind of knowledge only refers to some specific domain, it is called domain-dependent
knowledge(DDK). This topic focuses on incorporating DDK into a Golog-based system for the purpose of make it faster. Three different
approaches are studied and implemented into one Golog-based system, followed by an empirical study to evaluate their pros and cons within
different domains.