Goal model analysis of autonomy requirements for Unmanned Aircraft Systems

Abstract

Designing Unmanned Aircraft Systems (UASs) for optimal autonomy while meeting user requirements is quite challenging. Researchers have focused on improving autonomy algorithms and verification methods to ensure safe and reliable autonomous behavior in UASs, but little research has been conducted on requirements engineering for UASs to answer design questions and explore the trade space for using autonomy to satisfy user requirements. This paper introduces a method to determine an optimal set of autonomous capabilities that satisfies UAS user requirements in the early stages of conceptual design. The method uses a modified Autonomy Requirements Engineering (ARE) process that applies quantitative measures and statistical analysis to Goal-Oriented Requirements Engineering (GORE). We demonstrate this method in a case study of a “disaster robot,” i.e., a hazard response UAS for which the autonomy requirements were optimized using a goal model developed in the Goal-oriented Requirement Language (GRL), as implemented in the modeling tool jUCMNav. The high-level goals of the hazard response UAS—system performance, cost, and safety—were evaluated using the formula-based GRL strategy evaluation algorithm resident in jUCMNav version 6.0. An autonomy trade space study was conducted through a Design and Analysis of Simulation Experiments (DASE). Our designed simulation experiment inserted the number of trials (evaluation strategies) and inputs into the goal model, and evaluation data were analyzed to optimize design factors based on user weightings of the response variables. This paper presents a structured method of ARE for UASs, which could be adopted more broadly across other domains, demonstrating how to optimize autonomous capabilities for different design conditions.