AI cameras give ship crews another view of nearby hazards

A small boat can be visible near a merchant ship while producing little useful radar information. During a Lloyd’s Register assessment of Orca AI’s SeaPod, the camera system detected a small wooden boat that was absent from the ship’s radar display. It gave the bridge team another observation to work with.[2]

SeaPod uses day and thermal cameras to detect and track surrounding vessels and objects. It can show contacts without reliable Automatic Identification System broadcasts, alongside their movement and developing risk. Navigational decisions remain with the qualified bridge team under the master’s authority.[1]

A lookout that organises the view

The system connects an image with questions that matter on a bridge: which contact is approaching, how close might it pass, and how soon? Its display includes closest-point-of-approach information and alerts adapted to conditions such as ports, canals and busy traffic lanes. This helps officers focus on the contacts requiring attention.[1]

Detecting an object starts the task. Maintaining its track and placing it in the traffic picture helps a crew judge whether the situation is changing. An alert suitable near a port may differ from one needed in open water.[1]

What a voyage can establish

In the Lloyd’s Register report, precision concerns how many detections were correct, while recall concerns how many relevant reference targets were detected. Most observations were in daylight. The assessment excluded fully obscured objects from its missed-target count and did not evaluate classification accuracy. Its results therefore describe a specific detection test, with a useful example of cameras complementing radar.[2]

The larger NorthStandard study followed each ship through its first year using the platform, spanning over 10.8 million nautical miles. The reduction was measured between earlier and later use, with no untreated comparison fleet. Training, adaptation and operating changes could contribute; the result supports an operational improvement during adoption without isolating AI’s causal share.[3]

The voyage becomes material for the next crew

Orca’s FleetView gives shore teams access to the events recorded on board. They can examine video, vessel tracks and encounter geometry together, ask the crew for context, and use an actual incident for a debrief or training session. The same records can show whether a pattern recurs on a particular vessel or route.[5]

That adds a second use for continuous observation. A bridge team needs to understand what is happening now; a fleet manager needs to understand what keeps happening. Keeping the visual record attached to the event makes the discussion more specific than a monthly total of near misses.[5]

Eastern Pacific Shipping’s August 2026 expansion planned to add the platform to dozens more vessels after its five-ship pilot. The practical ambition is earlier, better-informed seamanship: notice the contact, understand its motion, choose a response and learn from the encounter afterwards.[4]

Report an error or suggest a correction