Building and refining abstract planning cases by change of representation language
Journal of Artificial Intelligence Research
| Hi-index | 0.00 |
Most of the work to date on automated control-knowledge acquisition has been aimed at improving the efficiency of planning; this work has been termed "speed-up learning". The work presented here focuses on the automated acquisition of control knowledge to guide a planner towards better solutions, i.e. to improve the quality of plans produced by the planner, as its problem solving experience increases. To date no work has focused on automatically acquiring knowledge to improve plan quality in planning systems. We present a taxonomy of plan quality metrics and a first prototype that partially automates the task of acquiring quality-enhancing control knowledge for the Prodigy nonlinear planner. We are working on testing the effect of such control knowledge in plan quality, and developing methods to learn such control knowledge. Two complex domains, namely a transportation logistics domain, and a machining process planning domain, are being used to evaluate these ideas.