Verification of Autonomous Systems

This book is a comprehensive guide to current practical and theoretical understanding of verification of autonomous systems, helping users find the tools and techniques they need to address this challenging problem. Autonomous systems are transitioning out of the lab and into the commercial and industrial space in ever increasing numbers. Verification provides the assurance case necessary for deployment of autonomous systems, but we lack guidance, certifications, and standards to help a business determine whether their autonomous system is safe, secure, or reliable. A community has formed to develop tools, techniques, and processes to solve this problem, but the existing literature largely focuses on autonomy design tools and techniques rather than on tools and techniques that can be used to verify an existing autonomy design. This book fills that gap. It directly connects specific challenges faced by verification agents and autonomy developers with the research and topics most relevant to their concerns: Specific challenges faced by stakeholders and researchers are cross-referenced to chapters addressing the relevant tools and techniques. Individual aspects of the verification process like requirements and specification development, model tests, and integration tests are mapped into technical challenges related to modeling, testing, abstraction and verification tools. Each chapter is tied to both the stakeholder challenges and the technical challenges it addresses. Verification processes, industrial needs, and issues with verification of learning systems are addressed across all domains, encompassing platforms from aircraft and spacecraft to marine systems and ground vehicles operating in industrial, military, commercial and household applications. Broad in scope, this reference ties advances in formal analysis, hardware testing, verification process modifications, and design and evaluation tools to the needs of verification researchers and professionals.