Reconfigurable system design is known to be a strategy that offers vast performance improvement over traditional designs. Extending from previous work, this paper presents a strategy for assessing the domination of a reconfigurable concept. By considering the performance demands placed upon a system at different segments of a mission, reconfigurable systems are shown in a multi-objective space as a collection of points. This is different than static designs which are typically represented as a single point. The principles of Pareto dominance are extended to describe the comparisons between systems in this space. A surrogate point is introduced to reduce calculation burden and provide a foundation for the development of a necessary condition for dominance. This approach is then demonstrated on a case study of Mars exploration rovers where a traditional rocker-bogie architecture is compared to the two-state Transforming Rolling Roving Explorer architecture. These principles lay the groundwork for choosing between reconfigurable and static systems when multiple objectives are considered.
I.
IntroductionR econfigurability and flexibility are concepts associated with complex system design that offer large improvements in system performance and robustness. Reconfigurability is the ability of a system to change "configurations … repeatedly and reversibly," 1 and benefits include multi-ability (operating at multiple performance points in the design space non-simultaneously), evolvability (changes to meet new demands), and survivability (changes to maintain functionality despite component failures). The goal of this paper is to explore the advantages of multi-ability when initially selecting system architecture. Specifically, this paper explores the architecture selection for a Mars rover.Tradespace visualization is a powerful tool for navigating the trade-offs inherent in system design, especially when multiple objectives are considered. This paper explores the challenges with, and establishes initial groundwork for, visualizing the complexity associated with reconfigurable systems in a multidimensional environment. Key challenges of this task relate to understanding how performance domination in a multi-objective space can be extended to a reconfigurable system, and how this domination may be explored visually. Solving these challenges will allow for the development of fundamental rules for evaluating reconfigurable architectures that can then be used to facilitate the application of multi-objective optimization techniques toward reconfigurable designs.Research over the last decade has seen an increased interest in the optimization of reconfigurable systems. Unmanned aerial vehicles have been a popular topic in particular. Work in this area has conducted sensitivity studies 2 , explored concept embodiment 3 , and developed analyses engines capable of spanning multiple disciplines 4-6 . Reconfigurable UAVs have also been envisioned with offline reconfigurations where the UAV is changed between missions by swapping out wi...