Testing Cyber-Physical Systems via Evolutionary Algorithms and Machine Learning

Cyber-Physical Systems (CPS) are systems or systems of systems made up of collaborating computational elements that control physical entities. CPS are developed in diverse domains ranging from automotive and aerospace to medical systems. This keynote argues that search-based techniques are a suitable match for testing CPS as they can handle complex continuous behaviors, scale to large test input spaces and are applicable to black-box systems such as physics-based simulators used in CPS development. In addition, the keynote demonstrates how search-based techniques can be flexibly combined with machine learning to improve search effectiveness and extend test results with explanatory information.

Paper

Full text

PDF

Testing Cyber-Physical Systems via Evolutionary Algorithms and Machine Learning

Semantic Scholar · Computer Science · 2019

Abstract

Cyber-Physical Systems (CPS) are systems or systems of systems made up of collaborating computational elements that control physical entities. CPS are developed in diverse domains ranging from automotive and aerospace to medical systems. This keynote argues that search-based techniques are a suitable match for testing CPS as they can handle complex continuous behaviors, scale to large test input spaces and are applicable to black-box systems such as physics-based simulators used in CPS development. In addition, the keynote demonstrates how search-based techniques can be flexibly combined with machine learning to improve search effectiveness and extend test results with explanatory information.

Similar papers

© 2026 NYSGPT2525 LLC