Session: DAC-03-01-Novel AI or ML Frameworks for Design or Systems Science

Paper Number: 89318

89318 - Does Design Proficiency Matter in Engineering Design Teams? A Computational Model and Experiments 

A cognitively inspired, agent-based model of engineering design proficiency is introduced in this work. Proficiency is modeled using move selection heuristics and different problem space search strategies, both of which were extracted from a prior human subjects study. Individual agent behavior in the Proficient Simulated Annealing Design Agents (PSADA) Model is validated against human designer behavior and performance. A multi-agent model of design teams is then produced with Monte Carlo simulation methods and it confirmed the previous finding that the most proficient member of a configuration design team has the largest impact (positive or negative) on team performance. Further experiments are conducted with different team characteristics. It is found that the effect of individual team member proficiency is reduced with greater team size, but the highest proficiency team member likely remains the most influential. Nominal teams, where team members do not interact and where the best design is taken as the team output, were found to outperform interacting teams. This was due to high proficiency agents making poor decisions during interactions by switching away from their own designs to the designs of their lower proficiency teammates. Conclusions about the role that individual team member proficiency in engineering design teams is discussed.

Presenting Author: Ethan Brownell Carnegie Mellon University

Presenting Author Biography: Ethan Brownell is a Ph.D. Candidate in Mechanical Engineering. He studies design, teaming, and modeling.

Authors:

Ethan Brownell Carnegie Mellon University
Jonathan Cagan Carnegie Mellon University
Kenneth Kotovsky Carnegie Mellon University