Research
Research Interests
Ensuring safety in cyber-physical systems (CPS), such as autonomous vehicles, require rigorous validation against complex temporal requirements. My PhD research develops automated testing techniques for validating black-box cyber-physical systems against multiple Signal Temporal Logic (STL) requirements using search-based software engineering, machine learning, and formal methods.
Key contributions include:
- Explicit output coverage: Identifying inputs that produce safety-critical or boundary-case behaviors through predetermined output analysis.
- Multi-requirement falsification: Simultaneously validating systems against interacting STL formulae to comprehensively test requirement dependencies.
- Continuous falsification: Extending validation efficacy across evolving system revisions while maintaining testing efficiency.
These techniques address critical gaps in black-box CPS testing, where search-based methods like generative adversarial networks (GANs) efficiently discover requirement violations without system internal knowledge. Consequently, this research advances scalable CPS validation frameworks directly supporting safety certification in automotive and aerospace industries, where requirement interactions pose significant risks.
PhD Thesis: Scalable Falsification of Multi-Formula STL Requirements for Cyber-Physical System Validation (title subject to change).
Current Work
I am a PhD Researcher at Åbo Akademi University, soon finishing my degree in Computer Science. I am a maintainer of stgem (system testing using generative models), an open-source Python framework for automated test generation for CPS. STGEM leverages machine learning techniques, such as GANs and search-based algorithms to efficiently discover requirement violations and corner cases in safety-critical systems. The framework supports various testing approaches such as falsification, corner case generation, and requirement-driven testing, making it particularly well-suited for safety-critical systems in domains like automotive, robotics, and industrial automation.