Page - 62 - in The Future of Software Quality Assurance

62 T.Linz • By 2025, more than 12% of newly produced vehicles will have autonomous driving hardware capability of Level 3 orhigher of the SAE International Standard J3016.1 • By 2022, 40 of the world’s 50 largest economies will permit routinely operated autonomous drone flights, up from none in 2018. It can be assumed that within the next10 yearsmobile systems will conquer the publicspaceandbeautonomously(orat leastpartiallyautonomously)“ontheway” there. The degree of autonomy of these systems dependson whether and how quickly manufacturers succeed in equipping their respective products with the sensors and artificial intelligencerequiredforautonomousbehavior. The major challenge here is to ensure that these systems are sufficiently safe and that they are designed in such a way that they can be approved for use in public spaces (road traffic, airspace,waterways).Theadmissibilityof the emerging systems and their fundamental social acceptance depend on whether the potential hazards tohumans,animals, andpropertyposedbysuchsystemscanbeminimized and limited to anacceptable level. Consensus must be reached on suitable approval criteria and existing approval proceduresmust be supplemented or new ones developed and adopted. Regardless of what the approvalprocedureswill look like in detail, manufacturerswill have to provethat theirownproductsmeet theapprovalcriteria. Thesystematicandrisk-adequatetestingofsuchproductswillplayan important rolein thiscontext.BoththeExpertGrouponArtificial Intelligenceof theEuropean Commission and the Ethics Commission “Automated and Networked Driving” set upbytheGermanFederalMinisterofTransportandDigital Infrastructureexplicitly formulatecorrespondingrequirementsfor testing in theirguidelines [3,4]. This chapter therefore discusses the question in which aspects the testing of future autonomous systems will differ from the testing of software-based systems of today’s character and gives some suggestions for the corresponding further developmentof the test procedure. 2 AutonomousSystems We understand the term “Autonomous System” in this chapter as a generic term for the most diverse forms of vehicles, means of transport, robots, or devices that arecapableofmovinginspace inaself-controllingmanner–withoutdirecthuman intervention. An older term for such systems is “Unmanned System (UMS)” [5]. The term emphasizes the contrast with conventional systems that require a driver or pilot on boardandalso includesnonautonomous,remote-controlledsystems. The modern term is “Autonomous Things (AuT)” [6]. This term is based on the term “Internet of Things (IoT)” and thus conveys the aspects that Autonomous 1See [2].