Acquiring planning knowledge from experts is a rather difficult knowledge engineering task. This work provides a novel method for accumulating domain and control planning knowledge by learning from the observation of expert planning agents and from one’s own practice. The observations of an agent consist of: 1) the sequence of actions being executed, 2) the state in which each action is executed, and 3) the state resulting from the execution of each action. Planning operators are learned from these observation sequences in an incremental fashion utilizing a conservative specific-to-general inductive generalization process. Operators are refined and new ones are created extending related previous work such as [?] and [?]. In order to refine the new operators to make them correct and complete, the system uses the new operators to solve practice problems, analyzing and learning from the execution traces of the resulting solutions or execution failures. Acquiring control knowledge that effectively copes with the incremental changes in the domain knowledge is a challenging problem. We discuss our approach to address this problem by the accumulation of complete problem solving episodes or macros from observation, and their flexible reuse and refinement at practice time.
Mendeley saves you time finding and organizing research
There are no full text links
Choose a citation style from the tabs below