Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Assisting maritime search and rescue (SAR) personnel with AI-based speech recognition and smart direction finding

 
: Gözalan, Aylin; John, Ole; Lübcke, Thomas; Maier, Andreas; Reimann, Maximilian; Richter, Jan-Gerrit; Zverev, Ivan

:
Fulltext urn:nbn:de:0011-n-6152195 (1.8 MByte PDF)
MD5 Fingerprint: 5e1ed0ef856dd0b8a1aafe39deaf4d9b
(CC) by
Created on: 17.12.2020


Journal of marine science and engineering 8 (2020), No.10, Art. 818, 13 pp.
ISSN: 2077-1312
Bundesministerium für Bildung und Forschung BMBF (Deutschland)
13N15018; Anwender–Innovativ: Forschung für die zivile Sicherheit
Bundesministerium für Bildung und Forschung BMBF (Deutschland)
13N15019; Anwender–Innovativ: Forschung für die zivile Sicherheit
Bundesministerium für Bildung und Forschung BMBF (Deutschland)
13N15020; Anwender–Innovativ: Forschung für die zivile Sicherheit
English
Journal Article, Electronic Publication
Fraunhofer IML ()
Fraunhofer IAIS ()
maritime search and rescue; VHF maritime radio; coastal radio station; speech recognition system; direction finding; AIS; maritime simulation; training; GMDSS

Abstract
Communication for processing relevant information plays a paramount role in developing a comprehensive understanding of Search and Rescue (SAR) situations and conducting operations in a successful and reliable manner. Nevertheless, communication systems have not changed considerably in the context of simplifying very high frequency (VHF) maritime communication and enhancing the value of SAR practices. The Automated Transcription of Maritime VHF Radio Communication for SAR Mission Coordination (ARTUS) project approaches this problem with the development of an assistance system which employs AI-based speech recognition and smart direction finding. First, ideas and specified needs of end users for designing the user interface are presented in this paper. Further, preliminary accomplishments of domain specific language training for maritime speech recognition, and the direction-finding algorithms for localizing senders are sketched out. While the preliminary results build a solid ground, additional field experiments will be conducted in order to enhance the accuracy and reliability of speech recognition and direction finding. The identified end user requirements across different personnel groups show commonalities, but call for a differentiated approach in order to meet the challenges and peculiar needs of the various working contexts.

: http://publica.fraunhofer.de/documents/N-615219.html