Agile Vetting of Autonomous Systems

Abstract

Autonomous systems are already coping with complex, unpredictable situations. Some of these are raising red flags about unanticipated behaviors and unintended consequences (Miley 2015). Systems engineering practitioners must respond with new ways and means for creating and vetting autonomous systems. Systems have become so complex that most of their caretakers do not understand them. This means the system must be protected not only from cyber attacks by others but also from its own caretakers. Eventually every new line of code introduced into a system must be considered a cyber attack until proven otherwise. This article explores the challenges in vetting autonomous systems for sustainable system life, and offers a checklist for agile vetting throughout the lifecycle.