nau-planning - AI Magazine Volume 28 Number 4(2007 AAAI...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
Automated planning technology has become mature enough to be useful in applications that range from game playing to control of space vehicles. In this article, Dana Nau discusses where automated-planning research has been, where it is likely to go, where he thinks it should go, and some major challenges in getting there. The article is an updated version of Nau’s invit- ed talk at AAAI-05 in Pittsburgh, Pennsylvania. I n ordinary English, plans can be many dif- ferent kinds of things, such as project plans, pension plans, urban plans, and floor plans. In automated-planning research, the word refers specifically to plans of action : representations of future behavior . .. usually a set of actions, with temporal and other con- straints on them, for execution by some agent or agents. 1 One motivation for automated-planning research is theoretical: planning is an impor- tant component of rational behavior—so if one objective of artificial intelligence is to grasp the computational aspects of intelligence, then cer- tainly planning plays a critical role. Another motivation is very practical: plans are needed in many different fields of human endeavor, and in some cases it is desirable to create these plans automatically. In this regard, automated- planning research has recently achieved sever- al notable successes such as the Mars Rovers, software to plan sheet-metal bending opera- tions, and Bridge Baron. The Mars Rovers (see figure 1) were controlled by planning and scheduling software that was developed joint- ly by NASA’s Jet Propulsion Laboratory and Ames Research Center (Estlin et al. 2003). Soft- ware to plan sheet-metal bending operations (Gupta et al. 1998) is bundled with Amada Cor- poration’s sheet-metal bending machines such as the one shown in figure 2. Finally, software to plan declarer play in the game of bridge helped Bridge Baron to win the 1997 world championship of computer bridge (Smith, Nau, and Throop 1998). The purpose of this article is to summarize the current status of automated-planning research, and some important trends and future directions. The next section includes a conceptual model for automated planning, classifies planning systems into several differ- ent types, and compares their capabilities and limitations; and the Trends section discusses directions and trends. Conceptual Model for Planning A conceptual model is a simple theoretical device for describing the main elements of a problem. It may fail to address several of the practical details but still can be very useful for getting a basic understanding of the problem. In this article, I’ll use a conceptual model for planning that includes three primary parts (see figure 4a and 4b, which are discussed in the fol- lowing sections: a state-transition system , which is a formal model of the real-world system for which we want to create plans; a controller , which performs actions that change the state of the system; and a planner , which produces the plans or policies that drive the controller.
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

Page1 / 16

nau-planning - AI Magazine Volume 28 Number 4(2007 AAAI...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online