The offshore energy sector is beginning to adopt artificial intelligence–driven detection systems to boost safety and reduce the risk of man overboard incidents.
One such system, ZOE, developed by Edinburgh-based firm Zelim, has now been installed on a second jack-up rig after proving its capabilities during more than a year of service on a North Sea installation.
ZOE combines camera hardware with AI software to automatically detect when a person falls overboard, track their position in real time, and provide data to coordinate rescue. Its deployment across multiple rigs signals growing interest in using machine learning to improve personnel safety and asset monitoring.
Zelim founder and CEO Sam Mayall said the technology was designed specifically for maritime conditions, which present unique challenges compared with land-based detection. “The first challenge in a man overboard incident is knowing when and where it happened. The second is keeping track of the person in the water, particularly in variable sea states. ZOE provides real-time visual tracking and geo-location data to coordinate an effective response,” he explained.
The company has built what it describes as the most extensive maritime search-and-rescue dataset in the world, containing more than seven million annotated images of people in water captured under varied conditions. Mayall said this was essential to ensure the system can recognise human figures in real-world situations, whether partially submerged, obscured by spray, or dressed in dark clothing.
ZOE is designed to integrate with a rig’s existing navigation, surveillance and emergency systems. It operates locally through onboard processing, removing reliance on cloud connectivity. Zelim has also developed modules based on the same AI core, including Watchkeeper, which assists bridge crews by spotting navigational hazards or nearby vessels, and Shield, which supports security by identifying suspicious activity around offshore sites.
The technology has been tested with the US Coast Guard, which has noted the limitations of human visual spotting during searches, where detection rates can fall below 20 percent. Mayall said AI adds a layer of reliability: “AI doesn’t fatigue, doesn’t blink, and doesn’t overlook what’s in plain sight.”
While not a replacement for human judgment, the system reduces detection times from minutes to seconds, significantly improving the chances of a successful rescue. Offshore operators are now beginning to view AI-powered detection as part of their standard safety toolkit, with Zelim’s work pointing to a future in which intelligent monitoring becomes a routine feature of marine operations.
