- Research Staff
- Join the lab!
- Contact Us
Matthew I. Campbell
Department of Mechanical Engineering
Carnegie Mellon University
The A-Design approach to automated design has been developed for a general class of conceptual design problems. The methodology is founded on the notion that engineering design occurs in interaction with an ever-changing environment and therefore computer tools developed to aid the design process should be adaptive to these changes. A-Design invents solutions to open-ended design problems through the interactions of a multitude of agents folded into a stochastic iterative process capable ofadapting to changes in user preference.
The motivation for A-Design is to integrate qualities of the human design process into a computational algorithm. This has been accomplished by creating four subsystems that each embody a different characteristic of human conceptual design: an open-ended representation of the design problem, an adaptability to changes in problem specifications, a collaborative involvement of different ideas and preferences, and an iterative yet guided search for successful solutions.
A design problem is presented to A-Design by a description of the desired functionality in the form of system inputs and outputs, a set of objectives to be optimized, and a library of electromechanical components. Two test problems (an electromechanical weighing machine and a MEMS accelerometer) test A-Design’s ability to invent novel configurations. Several experimental results validate the development of A-Design as a successful model of human design process characteristics. The results show that ADesign is independently capable of invention, has the ability to aid the human designer in developing new conceptual designs, and provides an experimental framework to model how to achieve conceptual design as well as how humans might achieve conceptual design.