Research
Research Focus
My research concerns search-based optimization for testing cyber-physical systems. I work with black-box systems, where the tester selects input signals and evaluates the resulting output signals without access to the system's implementation.
I formulate test generation as an optimization problem. Signal Temporal Logic (STL) formalizes the requirements and provides a robustness degree that can be used as a quantitative measure to guide the search toward requirement violations and specified output behavior.
Current Topics
- Multi-requirement falsification: searching for violations of several STL requirements in one search.
- Continuous requirement falsification: transferring knowledge between system revisions instead of restarting the search after each change.
- Explicit output coverage: finding a set of inputs whose outputs satisfy a set of STL test requirements.
The aim is to find useful test inputs with fewer executions of the system under test, especially when each software simulation is computationally expensive to run.
PhD thesis: Requirement Conformance Testing of Cyber-Physical Systems Across Multiple Requirements and Revisions.
Research Software
I maintain stgem (system testing using generative models), an open-source Python library for developing black-box falsification and test-generation methods. It supports time-varying inputs and outputs, STL requirements, and search methods based on generative models.