- 3
3
SEA.AI uses its machine vision technology to complement radar and AIS capabilities. Combining EO/IR sensors with embedded AI, its systems detect and classify objects on the water’s surface and estimate their distance in real time.
- The Sentry and Watchkeeper systems adapt SEA.AI’s machine vision technology to different types of platforms.
- 3D rendering of the Sentry Black, SEA.AI’s machine vision system.
© SEA.AI
Small boats, floating objects, people overboard and unmanned surface vessels can be difficult to detect using conventional systems. Radar can struggle to detect small targets sitting very low in the water, while the Automatic Identification System (AIS) only provides information about vessels transmitting their position. SEA.AI adds another capability: visual perception. Using edge computing, its systems analyse images from EO/IR sensors in real time, directly on board and without relying on an external connection.
The development of unmanned surface vessels (USVs) has made this visual perception capability increasingly important. A USV must be able to detect and interpret its surroundings, including objects that transmit no information. Machine vision therefore helps improve both autonomous operation and situational awareness.
SEA.AI’s machine vision technology is incorporated into several systems designed for different types of platforms. Its product range extends from the Sentry, weighing approximately 6.2kg, to the Watchkeeper, weighing less than 1kg. This allows the technology to be deployed on a wide variety of crewed and uncrewed platforms.
In defence applications, the technology has another key advantage: its passive operation. Unlike radar, optical and infrared sensors do not emit signals to observe their environment. The system’s visual perception capabilities are also independent of GNSS. Even if GPS signals are jammed or spoofed, the sensors can continue to provide visual information about their surroundings.
“As tensions rise, technology is evolving fast. Sensors that emit signals can be detected or jammed, and the risk is growing as jamming equipment becomes more widely available. Our maritime machine vision technology, developed in Europe, requires neither active emissions nor satellite positioning signals. It gives operators another source of information to improve maritime situational awareness, day and night. Already proven at sea and easy to deploy, it is increasingly being integrated into existing surveillance systems,” explains Marcus Warrelmann, CEO of SEA.AI.
Proven technology, from search and rescue to surveillance
The technology has a wide range of applications, from detecting people overboard, small boats and floating hazards to coastal surveillance and harbour and naval base protection. According to SEA.AI, its systems can detect a person overboard at distances of up to 700 metres, depending on conditions and system configuration. In military applications, the technology can also help detect, track and classify small, potentially hostile USVs. It provides operators with additional information, while leaving operational decisions in human hands.
SEA.AI also stands out because its technology is already in production and operational use. In 2024, the company secured a multimillion-euro European contract to equip a fleet of autonomous vessels for defence applications. It does not aim to replace existing sensors, but rather to complement them by giving ships and autonomous platforms an additional visual perception capability.
Founded in 2018 as Oscar and renamed SEA.AI in 2022, the Franco-Austrian company stems from the combination of Brittany’s maritime expertise and machine vision technology developed particularly in the automotive industry. The technology was initially developed for offshore racing, where detecting floating obstacles in time can be crucial to both performance and safety.
“We started out trying to solve a very practical problem: enabling an offshore racing skipper to see at night what their instruments couldn’t detect. We needed technology that was reliable, compact and energy-efficient, capable of operating for long periods in challenging conditions. The data also had to be processed directly on board, using edge computing, without relying on a continuous connection. Defence applications weren’t even on our minds when we wrote the first algorithms. But at sea, whatever the platform, the requirements are the same: autonomy, robustness and low power consumption. So our technology quite naturally attracted interest for other applications and platforms, well beyond sailing,” explains Solenn Gouerou, Head of Marketing and founding member of SEA.AI.
Today, more than 1,300 SEA.AI systems are in operation, serving recreational boating, commercial shipping, maritime search and rescue (SAR), and autonomous platforms.
- Linda Verhaeghe
