JSPS Kakenhi Project

Surrogate Model-based Integration Testing of CPS with Complex Black Box Components

My first grant application was accepted!

I applied to the JSPS Kakenhi Research Startup Grant, which supports researchers in picking up new research activities. In my case, I am interested in applying my knowledge of systems modelling and verification to the automotive domain. In particular, I am looking into autonomous driving systems and their use of black-box components such as machine learning and AI units.

I am very excited that my project “Surrogate Model-based Integration Testing of CPS with Complex Black Box Components” was accepted and that it received a total funding of 2,860,000 JPY (at that time ~ 27k USD / 23k EUR) over two fiscal years (2020/2021). According to the project agreement, I will be spending some 15% of my work effort on this project.

Feel free to contact me know if you have questions about this project and its details.

Here’s the project’s summary that I submitted as part of my project application:

Testing and verification of cyber-physical systems (CPSs) often suffer from problems such as systems complexity and many-actor environments. Recently, however, a new complexity dimension arose, namely, the use of machine learning (ML) and artificial intelligence (AI) components, whose internal behaviour cannot be fully analysed. Their use in CPSs such as autonomous driving systems (ADSs) poses a significant hindrance to guaranteeing safety and security. Existing black-box component verification techniques (e.g. search-based testing) currently do not scale to the system level and therefore focus on the components’ verification in isolation. The growing number of ADSs and other safety-critical applications in everyday life renders this situation unacceptable. This project describes an effective approach to lifting existing verification methods to the integration level and thereby coming one vital step closer to the ultimate goal of full CPS verification at the system level.

Link to the official grant page: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-20K23334/