This thesis describes a program called KIP (Knowledge Intensinve Planner). KIP is a general, commonsense planner that can reason about planning situations in the real world for which it is provided information. KIP is the planning component of t he UC (UNIX consultant) system. KIP is used to solve the problems the user poses to the UC. KIP has knowledge about UNIX commands, including the effects of those commands and under what conditions those commands can and should be issued. The best plan is reported to the user after (1) determining the goals of the user, (2) selecting and specifying a plan that fulfills the goals of the user (plan determination), (3) testing if the plan will work in this particular problem situation without causing unacceptable consequences, and (4) modifying this plan if necessary. A major problem in commonsense planning is the focus of attention on relevant knowledge. In particular, the problem of identifying potential plan failures in a plan is difficult, since there are often many sources of plan failure, both for failures due to an unsatisfied condition of a plan and failures due to goal conflict. This problem is further complicated because many values of conditions in a particular planning problem may be unknown. In order to address the problem of identifying potential plan failures, a new idea, called a concern, has been introduced. Concerns identify which aspects of a plan are most likely to fail.
Cited By
- Wilensky R, Chin D, Luria M, Martin J, Mayfield J and Wu D (2019). The Berkeley UNIX Consultant Project, Artificial Intelligence Review, 14:1-2, (43-88), Online publication date: 1-Apr-2000.
- Virvou M and Du Boulay B (1999). Human Plausible Reasoning for Intelligent Help, User Modeling and User-Adapted Interaction, 9:4, (321-375), Online publication date: 1-Apr-1999.
- Weld D and Etzioni O The first law of robotics (a call to arms) Proceedings of the Twelfth AAAI National Conference on Artificial Intelligence, (1042-1047)
- Wilensky R, Chin D, Luria M, Martin J, Mayfield J and Wu D (1988). The berkeley UNIX consultant project, Computational Linguistics, 14:4, (35-84), Online publication date: 1-Dec-1988.
Recommendations
Control knowledge in planning: benefits and tradeoffs
AAAI '99/IAAI '99: Proceedings of the sixteenth national conference on Artificial intelligence and the eleventh Innovative applications of artificial intelligence conference innovative applications of artificial intelligenceRecent new planning paradigms, such as Graphplan and Satplan, have been shown to outperform more traditional domain-independent planners. An interesting aspect of these planners is that they do not incorporate domain specific control knowledge, but ...