REPORT OF THE MARKET SURVEY OF MARITIME ELEC- 
TRO-OPTICAL SENSORS AND AI ASSISTANCE FOR NAUTI- 
CAL CREW 
Jelmut Schmid, Martin Portier, Hans Herrmann © 
Abstract 
This report examines the recent evolution of maritime night vision systems and their potential impact 
an situation awareness tasks for highly automated vessels. On one hand, it is based on a study that 
avaluates algorithms for nautical applications and includes a market survey involving 26 products (Koch 
at al., 2024). On the other hand, it describes the comparison between technical specifications and the 
performance observed in field trials conducted in the German Bight under real-life conditions. 
Machine Learning and more affordable sensor technology resulted in product enhancements and ad: 
ditional features of maritime night vision systems. Machine Learning features are currently not speci- 
fied in the ISO Standard 16273 Night vision equipment for high-speed craft. This report outlines possi- 
ble next steps whilst focusing on the definition of performance metrics and specific maritime datasets 
for training and validation purposes. The benefits of harmonisation on classification is highlighted. 
Introduction 
In 2000, the International Maritime Organisation (IMO) adopted the performance standard 
MSC.94(72) for night vision systems for high-speed crafts (IMO, 2000). The aim of this performance 
standard is to extend the capabilities of collision avoidance and safe navigation beyond established 
systems such as Radar and AIlS. According to this performance standard, night vision equipment should 
anable the crew to detect small objects such as unlit boats, floating logs, oil drums, containers, buoys, 
and whales. The night vision technologies specified in ISO 16273 are for the regulated market and only 
a subset of electro-optical (EO) sensors. In order to fulfil the task of object detection and because of 
the different visibility conditions on the open sea, it is a technically sensible option to consider the 
simultaneous use of various EO sensors. 
Figure 1 illustrates the sensitivity range of sensor technologies currently used for maritime night vision 
systems in the non-regulated market. It includes the human eye (black curve) in order to compare it 
with a short-wave infra-red (SWIR, orange) sensor and a mid-wave infra-red (MWIR, red) sensor. In 
addition to a light amplifier tube in combination with low lux camera, long-wave infra-red (LWIR, 
brown) is the most widespread sensor technology. 
Progress in image processing and object detection yield various EO sensor systems for decision support 
and situational awareness, which are offered by traditional manufactures of navigation systems as well 
as start-ups. Integrated functions such as object classification are powered by Machine Learning com: 
ponents and offer entirely new possibilities for perception tasks. Such EO systems with Al components 
are currently not regulated. Several EO sensor systems on the market use multiple sensors instead of 
only long-wave or light amplifier technology as indicated by ISO standard 16273 (ISO, 2020). 
Most scientific publications on object detection using EO systems focus on the fundamental technical 
aspects of ship identification (Chen et al., 2022)(lancu et al., 2021)(Kim et al., 2022)(“Oscar,” 2024).