Document is current

Article History

Received: 1 May 2023

Revised: 1 August 2023

Accepted: 25 August 2023

First Online: 20 September 2023

Declarations

:

: On behalf of all authors, the corresponding author states that there is no conflict of interest.

: All numeric parameters used in the models used within this paper are either given in this paper, or follow those of the papers they originated from (where their values can be found). All numerical calculations in this paper were conducted using NumPy (Harris et al. ) (version 1.13.3). SciPy (Virtanen et al. ) (version 1.1.0) was used to solve optimization problems determining the minimum and maximum lift possible for the purposes of evaluating the constraint requiring airfoils to be capable of control (see Rudnick-Cohen et al. (Rudnick-Cohen et al. ) for details), using the SLSQP solver. The genetic algorithm implementation used in this paper was the inspyred (Garrett ) library, using default settings and operators except where previously noted.The authors caution the reader that the optimization problem solved in the paper contains a number of non-deterministic elements, such as the random sampling used to construct the motion planning graph, the genetic algorithm used to solve the optimization problem and the random samples used by the robust optimization approach. Thus replications of this study will likely produce different trajectories and worst-case scenarios than observed in this paper. Furthermore, because of the wide range of topologies that SPIDRS (Bielefeldt et al. ) can generate, it is likely that different (but similar performing) designs might be found if the experiments in this paper were repeated. However, repetitions of the study conducted in this paper will still produce the result observed where the multi-objective optimization approach’s solutions are overspecialized for the set of flight states its designs are optimized for, while the design found by the robust optimization approach has optimal worst-case performance.