A hierarchical system modelling mechanism for planning in transportation and logistics

Abstract

Planning and scheduling (P&S) is an important area of artificial intelligence (AI). In the late decade, application-oriented research receives much more attention in AI planning community. Hierarchical Task Network (HTN) is an effective AI planning approach and has been applied in many realistic domains. However, current modeling mechanisms of HTN planning are mainly based on logical and/or functional mechanism and not efficient for large and complex system modeling. In this paper, an object-oriented mechanism known as Model-based Integrated Planning Language (MIPL) is proposed, which can describe physical systems with hierarchical, logical and numerical relations, and can also describe HTN planning domain such as operator, method and task. Based on MIPL, planning process is provided concerning compiling schemes and integration of reasoning and computing. Transportation and logistics (T&L) is an attractive application domain of AI planning. A T&L system model is built by this mechanism with hierarchical structure and hybrid component attributes, and has been used in HTN planning effectively.