The safety and performance of cyber-physical systems (CPS) depend crucially on control and scheduling decisions that often are fixed at design time, which significantly restricts the conditions under which a system can operate both safely and with suitable performance. Going beyond prior work that has explored different control and scheduling adaptations in individual system designs, this project will conduct more general and in-depth investigations, into how cyber-physical systems? control and scheduling can be co-designed to adapt jointly, automatically, dynamically, safely, and effectively even in response to rapid, large, and diverse changes in: (1) the system?s controlled behavior; (2) its environment; (3) its physical components; and (4) its platform software and hardware. Our project will immerse multiple graduate students in cross-disciplinary research, with extensive education, training, and mentoring spanning computer science, control theory, natural hazards engineering, structural engineering, mechanical engineering, and computer engineering. We will also involve undergraduate students via summer REU supplements and in-semester mentored independent study projects for academic credit, and will leverage our existing initiatives and relationships with partner organizations for K-12 outreach. As we have done in each of our previous collaborations, our multi-university team will recruit, mentor, and retain participants from groups traditionally under-represented in science and technology fields, leveraging effective and established outreach programs at our institutions.

In this cross-disciplinary research project we will develop new formal models, analyses, system infrastructure, and evaluation metrics, to explicitly represent, respect, and even exploit control and scheduling inter-dependencies, to ensure that systems? behaviors remain safe while enabling significant improvements in performance. The novel co-design approach we propose will enable radically improved cyber-physical system performance capabilities while respecting safety constraints that may cross-cut cyber and physical components and the system?s environment. For example, it will enable more extreme (but safely realizable) stress testing and adaptive management of mechanical systems and civil structures, to gauge and maintain resilience to significant (potentially adverse) changes to conditions in a system and its environment, and to enact adaptive mitigating responses accordingly.