Trust Considerations in the Coordination of Computational Design Teams

Abstract

Design of complex systems requires a team and a significant coordination effort to produce a coherent whole that can achieve the desired system-level characteristics. Trust plays a critical role in group interactions with a significant impact on the efficacy of teamwork. Methods to make effective team coordination decisions based on proper trust relationships are not well studied in the literature. Here, we present a notional case study to quantitatively analyze the impact of trust in team coordination on the overall team performance when the team members make design decisions using uncertain information sources. The analysis is based on computational teams where a system-level problem is decomposed into subproblems solved by individual computational agents. In this case study, we model the design decisions of the individual agents with deterministic or Bayesian optimization depending on whether the models used for decision-making contains uncertainty. We use analytical target cascading to coordinate the team decisions where trust is a simple static quantity that determines the importance of the response coming from individual subsystems in the system-level problem. The results show that maximum trust in a team does not always yield the best teaming performance and that trust indeed should be a design variable to be determined based on the quality of the decisions that a team member makes. This study demonstrates these results on a notional problem for simplicity and interpretability while the approach can be generalized to more practical design problems.