On Computational Synthesis of Complex Electromechanical Systems Position paper

Developing new approaches to aiding computational synthesis of modern electromechanical systems is a ma-jor need. Current techniques use product representa-tions that reason with single abstractions such as either geometry or physical dynamics. Further, these tech-niques are utilized in the context of static design pro-cesses. This article proposes the development of com-putational frameworks wherein both the process of de-sign along with the product being designed are reasoned with in an integrated manner. Developing such a frame-work would require advances in product models that in-tegrate geometric and behavioral abstractions. Further, development of new process models would require in-tegration of planning and machine learning techniques that reason with these new product representations. An integrated framework would aid in the development of better cost-effective synthesis tools and allow for assim-ilating and reusing many kinds of design knowledge. Potential approaches towards developing such a syn-thetic framework are outlined