Runtime Assumption Verification Using Formal Methods and Machine Learning

Canelas et al.'s (2024) study of ROS misconfigurations revealed a critical gap: most tools focus on static analysis, while many problematic assumptions only manifest at runtime. While formal methods like those mentioned in Ghimis et al.'s (2020) RIVER framework offer precision, they struggle with the scale and complexity of modern component systems. This research proposes a novel hybrid approach that uses formal methods to explicitly specify component assumptions as verifiable contracts, then employs ML techniques (like anomaly detection) to monitor these contracts at runtime. When the system detects an assumption violation, it could either prevent the problematic interaction or roll back to a safe state. The innovation lies in bridging the gap between static assumption specification and dynamic verification. Unlike traditional testing approaches that try to find all possible misconfigurations beforehand (impractical for complex systems), this approach continuously monitors for assumption violations in production. This directly addresses the "unstated assumptions" problem identified by Canelas et al. (2024) while leveraging the automation capabilities demonstrated in ML testing frameworks like those by Akila et al. (2023).

References:

1. Enhancing Software Testing with Machine Learning Techniques. V. Akila, A. Vasuki, J. Christaline, R. Sathiya, Priti Rishi, A. Edward (2023). 2023 International Conference on Sustainable Computing and Data Communication Systems (ICSCDS).
2. Understanding Misconfigurations in ROS: An Empirical Study and Current Approaches. Paulo Canelas, Bradley Schmerl, Alcides Fonseca, C. Timperley (2024). International Symposium on Software Testing and Analysis.
3. RIVER 2.0: an open-source testing framework using AI techniques. Bogdan Ghimis, Miruna Paduraru, Alin Stefanescu (2020). Proceedings of the 1st ACM SIGSOFT International Workshop on Languages and Tools for Next-Generation Testing.