A Planning Model for Agents in Dynamic and Uncertain Real-Time Environments*
Alexander Nareyek · 1998
The management of an agent in a dynamic and uncertain real-time environment is fairly intractable with traditional planning methods. Such methods cannot be computed efficiently, are not capable of replanning, and mostly lack explicit temporal formalisms. The planning model proposed in this paper overcomes these weaknesses by applying iterative repair methods, constraintprogramming techniques, and an explicit timeline approach. For the management of incomplete knowledge, a single-plan approach is proposed in which only one plan possibility is considered. The application of Unknown values, Known values, Not values, and continuous domain fluents enables to cope with incomplete knowledge. Introduction The work originated as part of the Excalibur project. The goal of this project is to develop a generic architecture for autonomously operating agents, like computerguided characters/mobiles/items, within a complex computer-game environment. These agents have to be able to find the right ac...