@article{Stach_Koch_Constapel_Portier_Schmid_2024,
	author = {Stach, Thomas and Koch, Paul and Constapel, Manfred and Portier, Martin and Schmid, Helmut},
	title = {VerifAI: Framework for Functional Verification of AI based Systems in the Maritime Domain},
	journal = {TransNav, the International Journal on Marine Navigation and Safety of Sea Transportation},
	volume = {18},
	number = {3},
	pages = {585-591},
	year = {2024},
	url = {./Article_VerifAI_Framework_for_Functional_Stach,71,1432.html},
	abstract = {With the continuous emergence and steady development of new technologies the way for Maritime Autonomous Surface Ship (MASS) is being paved. However, this manifold of available and imminent technologies challenges regulatory bodies and auditing authorities. Technologies which make use of Artificial Intelligence (AI), in particular Machine Learning (ML), play a special role. On one hand, they are not covered by current regulations or audit processes and, on the other hand, they may represent black boxes whose behaviours are not readily explainable and thus impede audit processes even further. In an upcoming study titled VerifAI the authors focus on this gap within European and German regulatory bodies and auditing authorities. The technological scope lies on MASS-related products which rely on partially or fully AI based systems. In the present article the original authors summarize the outlined study. The authors review the current regulatory status concerning audit processes and the market situation concerning available and imminent (partially) AI based systems of MASS-related products. To close the gap a conceptual, integrated framework consisting of a Safety Guideline for the manufacturers and a Verification Guideline for the auditing authorities is presented. The framework aims to give regulatory bodies and auditing authorities an overview of necessary steps for robust verification of safe products without hindering innovation or requiring in-depth knowledge about the (black box-like) systems. The results are condensed into recommendations for actions, listing the most important results, and proposing entry points as well as future research in the field of verifying (partially) AI-based MASS-related products.},
	doi = {10.12716/1001.18.03.12},
	issn = {2083-6473},
	publisher = {Gdynia Maritime University, Faculty of Navigation},
	keywords = {Methodological Framework, Safety of Life at Sea (SOLAS) Convention, Maritime Domain, Maritime Autonomous Surface Ships (MASS), Artificial Intelligence (AI), Machine Learning, AI-based Systems, Artificial Intelligence Act}
}