Especificação de preferências de planos usando metas estendidas na lógica alpha-CTL

Abstract

One of the most important aspects of human behavior is to plan tasks, whatever they may be, in order to achieve some goal. Knowing this, developing algorithms capable of autonomous planning is one of the main objectives of several researches done in Artificial Intelligence, in particular, in subarea called Automated Planning. In this subarea, the task we want the algorithm to autonomously plan for is called the planning problem. Most planners in the literature are able to devise plans for simple reachability goals specified through desired end state sets. Planners able to handle more expressive goal specifications (i.e., extended goals) have increasingly aroused interest in the planning community. In (PEREIRA; BARROS, 2007), the α-CTL logic was defined to express planning issues with extended reachability goals. Another approach that has gained prominence in the literature is Planning with Preferences (BAIER; MCILRAITH, 2008; SMITH, 2004; BRIEL et al., 2004; NIGENDA; KAMBHAMPATI, 2005; BENTON et al., 2007; BOUTILIER et al., 2004; BRAFMAN et al., 2006). This approach makes it possible to specify properties of a plan, which distinguish it as ”high quality” or ”preferable”. Although the areas of planning with extended goals and planning with preferences over plans are distinct approaches, they seem to work on similar issues and both use theoretical frameworks in their formulations such as temporal logics. As far as we know, in the literature the relationship between these areas is not well established, but this work aimed to establish a relationship between them and find a way to use the α-CTL planning framework to solve planning problems with Preferences.