An analytic network process model for ranking exoskeleton evaluation criteria

In nearly 60 years of research on exoskeleton design, manufacturing, testing, and application, few researchers study what goes into the process of designing these devices that augment human performance. Exoskeleton evaluation and selection, however, remains under-­researched and ultimately unclear given the large number of decision criteria with possible interdependencies. In this paper, an analytic network process (ANP) model was structured and used due to its ability to consider multi-levels of interdependencies between decision criteria while encapsulating the advantages of expert opinion. A panel of experts categorized 55 engineering design metrics into five categories. Discussion sessions were conducted to identify the metrics' interdependencies yielding an unweighted, weighted, and limit priority super-matrix from the ANP model. The results of the ANP model provide an analysis of how interdependencies affect the importance of design metrics in exoskeleton design. The results of the proposed model were compared to a previous study where a panel of 40 experts individually ranked the importance of engineering design metrics without considering interdependencies. The use of the ANP can provide a stronger, more holistic approach to the inherently multi-criteria decision-making involved in exoskeleton design.