Dominic II: Meta-Level Control in Iterative Redesign

This paper describes the meta-level control system of a program (Dominic) for parametric design of mechanical components by iterative redesign. We view parametric design as search, and thus Dominic is a hill climbing algorithm. However, from experience with Dominic we concluded that modeling engineering design as hill climbing has several limitations. Therefore, a need for meta-level control knowledge exists. To implement meta-level control, we have taken the approach of dynamically modifying the way hill climbing is performed for this task, rather than requiring the addition of domain-specific control knowledge. We have identified the limitations of hill climbing, constructed various generic hill climbing strategies, and developed a meta-strategy to guide the application of the strategies. The program monitors its own performance for unproductive efforts and selects among different strategies to improve its performance as it designs. This meta-level control significantly improves the performance of the program over the performance of an earlier version.

[1]  Paul R. Cohen,et al.  Design for Acquisition: Principles of Knowledge-System Design to Facilitate Knowledge Acquisition , 1987, Int. J. Man Mach. Stud..

[2]  B. Chandrasekaran,et al.  Generic Tasks in Knowledge-Based Reasoning: High-Level Building Blocks for Expert System Design , 1986, IEEE Expert.

[3]  William J. Clancey,et al.  Heuristic Classification , 1986, Artif. Intell..

[4]  Tianran Wang,et al.  Knowledge Acquisition for Constructive Systems , 1985, IJCAI.

[5]  Victor R. Lesser,et al.  The Distributed Vehicle Monitoring Testbed: A Tool for Investigating Distributed Problem Solving Networks , 1983, AI Mag..

[6]  Craig Cornelius,et al.  Computational Costs versus Benefits of Control Reasoning , 1987, AAAI.

[7]  Paul R. Cohen,et al.  Knowledge Engineering Tools at the Architecture Level , 1987, IJCAI.

[8]  William J. Clancey,et al.  From Guidon to Neomycin and Heracles in Twenty Short Lessons: ORN Final Report 1979-1985 , 1986, AI Mag..

[9]  Victor R. Lesser,et al.  Meta-Level Control Through Fault Detection and Diagnosis , 1984, AAAI.

[10]  Michael P. Georgeff.  Strategies in Heuristic Search , 1983, Artif. Intell..

[11]  William A. Beckman,et al.  Solar heating design, by the f-chart method , 1977 .

[12]  Allen Newell,et al.  SOAR: An Architecture for General Intelligence , 1987, Artif. Intell..

[13]  Paul R. Cohen,et al.  An Architecture for Application of Artificial Intelligence to Design , 1984, 21st Design Automation Conference Proceedings.

[14]  Paul R. Cohen,et al.  MU: A Development Environment for Prospective Reasoning Systems , 1987, AAAI.

[15]  Herbert A. Simon,et al.  The Sciences of the Artificial , 1970 .

[16]  Barbara Hayes-Roth,et al.  A Blackboard Architecture for Control , 1985, Artif. Intell..