X-Plore: Combining Model-Driven Engineering, Bio-Inspiration and Formal Analysis To Mitigate Uncertainty in High Assurance Software Systems
Abstract
Increasingly, cyber-physical systems are expected to deliver acceptable and trusted behavior despite highly dynamic and uncertain operating conditions. The X-PLORE project investigated the integration of evolutionary search algorithms with formal analysis methods in order to enhance system robustness and resiliency, while identifying corner cases that might lead to system failure under certain conditions. Evolutionary search algorithms operate in an open-ended manner, unconstrained by human bias and preconceptions. Combining this capability with formal analysis enables discovery of unintuitive solutions to design problems as well as situations that might cause the system to behave in an unintended manner after deployment. This report describes the main capabilities developed in the project along with the results of studies in applying those methods to autonomous vehicles of different scales. A list of publications and presentations resulting from this research is also provided.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 29, 2020
- Accession Number
- AD1090421
Entities
People
- Betty Cheng
- Philip Mckinley
Organizations
- University of Michigan