SEA.AI Competition 320 - AI collision avoidance system for offshore racing

The SEA.AI Competition 320 is an AI-based vision system designed to assist in collision avoidance. Its purpose is to help racing and offshore sailing crews detect floating objects and other vessels, day and night, to prevent collisions.

This system is specifically designed for offshore racing and competitive sailboats. It is built for teams seeking lightweight watch assistance with offshore performance focus, offering night thermal detection and decision support without overloading the masthead.

The SEA.AI Competition 320 system consists of an ultra-light masthead unit integrating cameras and an external processing unit. The masthead unit includes two thermal cameras and a daylight camera, while the external unit is dedicated to AI analysis, alerts, and onboard integration.

The SEA.AI Competition 320 is equipped with two thermal cameras, each with a resolution of 256 x 320 pixels. These two streams are combined to provide a global thermal image of 512 x 320 pixels, enabling effective object detection in low light or complete darkness.

Yes, in addition to thermal cameras, the SEA.AI Competition 320 system integrates a daylight camera. This camera is used for visible-light monitoring, complementing the system's detection capability in varying lighting conditions.

The SEA.AI Competition 320 system integrates advanced electronic stabilization. This technology compensates for boat movements (pitch, roll, yaw) to maintain a stable and clear image, essential for reliable object detection at sea.

Yes, the SEA.AI Competition 320 is specifically designed to be suitable for racing conditions, including on rotating masts. It features automatic compensation for mast rotation, ensuring continuous and accurate monitoring regardless of mast orientation.

The external processing unit is the brain of the system. It is responsible for analyzing video data via artificial intelligence, generating alerts when objects are detected, and integrating the system with other electronic equipment on board the vessel.

The SEA.AI Competition 320’s artificial intelligence analyzes images captured by the cameras to identify and classify different types of objects. It provides reliable object recognition, aiding decision-making to avoid collisions.

The SEA.AI Competition 320’s external processing unit, equipped with an NVIDIA® Jetson AGX Xavier™ processor, can analyze images at a high frequency of three times per second. This processing speed ensures near real-time detection of objects.

Yes, the SEA.AI Competition 320 system is designed to generate alerts. When the AI detects a potentially hazardous object, alerts are issued to inform the crew, enabling rapid decision-making to avoid a collision.

The masthead unit must be installed at the top of the mast. Its ultra-light design is specifically adapted to minimize weight and balance impact at the masthead, which is crucial for racing sailboats.

The external processing unit is connected to the masthead unit via an Ethernet cable. This connection enables the transfer of video streams from the cameras to the processing unit for AI analysis, as well as the transmission of commands.

Yes, the system allows manual scanning of the masthead unit with a ±50° range. This feature gives the user the ability to manually orient the monitoring area based on specific navigation or observation needs.

Yes, the SEA.AI Competition 320 is designed to integrate with existing onboard navigation systems. It can communicate with multifunction displays (MFD) from major manufacturers via standard interfaces, enabling data and alert display.

The SEA.AI Competition 320 primarily uses Ethernet for transmitting video and control data, and NMEA 2000 for receiving navigation data (heading, speed, position). These standard interfaces facilitate its integration with existing marine electronics.

Yes, the SEA.AI Competition 320 is compatible with multifunction displays from major manufacturers such as Garmin, Raymarine, B&G, Simrad, and Furuno. Integration is typically via Ethernet, allowing video streams and detection data to be displayed directly on the MFDs.