SORS: Scalable methods for Verifying Autonomous Cyber-Physical Systems

Digital Twins, a virtual representation of reality, have been effectively used successfully for a number of applications such as training, visualization, and collaboration. For the task of validation and verification (V&V), digital twins combined with predictive models of reality (simulation) offer an attractive method for V&V of large complex systems. Indeed, this process has succeeded spectacularly in the field of semiconductors and electronics, where multi-trillion dollar industries are built on a complex interconnected web of mathematical abstractions, each with processes for abstraction, transformation, and synthesis. However, in the field of autonomous cyber-physical systems (ground, air, marine), V&V is largely an unsolved problem. The required abstractions, separation of concerns, characterization methods, and integrated V&V processes have not been discovered yet. This talk will discuss the underlying structure of the semiconductor V&V process, its mappings to the cyber-physical problem, and introduce PolyVerif(www.avvc.net), an open-source digital twin autonomous vehicle validation framework, whose objective is to accelerate the discoveries of scalable methodologies for cyber-physical V&V. While PolyVerif is focused on ground vehicles, the underlying structures map to the airborne and marine applications as well.