Matthew I. Campbell, Jonathan Cagan
Computational Design Laboratory
Department of Mechanical Engineering
Carnegie Mellon University, Pittsburgh, Pennsylvania, USA
AND
Kenneth Kotovsky
Department of Psychology
Carnegie Mellon University, Pittsburgh, Pennsylvania, USA
This paper introduces a new learning algorithm to be used within a larger automated design synthesis technique. This learning method, known as TODO/TABOO learning, is based on the idea that experience gained from past design activities greatly improves the efficiency of the design process and the value of future designs. While computational search techniques have the ability to search many design alternatives quickly, the human engineer can often arrive at a more elegant and robust solution by applying heuristics learned from past experiences. The TODO/TABOO learning method extracts common “trends” from previous designs and sorts them as either positive or negative learning criteria for use in future design activity. Results show how such a technique improves the quality of designs and efficiency of an existing automated search process.