Orca AI is developing a marine platform that combines its own sensors with the ships’ existing safety systems.

The software must deliver safe autonomous navigation even in heavily trafficked routes. The Israeli company has been training its algorithms using data collected by the Japanese cargo ship Suzaka for more than a year. It writes Electrek.

The startup already has a collaboration with Nippon Yusen Kabushiki Kaisha (NYK Group), one of the world’s largest shipping companies that is part of the Mitsubishi group. Now Orca AI has tested its “Automatic Ship Target Recognition System” in sharp position on board Suzaka, which belongs to NYK Line.

The test took place within the project DFFAS (designing the future of full autonomous ship), which brings together 30 Japanese companies. The autonomous transport began in the Gulf of Tokyo, one of the busiest waters in the world – and an absolute majority of the journey was carried out without human intervention.

According to Orca AI, their system acts as a human lookout, around the clock.

The ship’s 18 cameras detected other boats in real time, writes Interesting Engineering. The distance was measured, they were classified and their movements were followed.

During 40 hours and 790 km, the Suzuka system was able to perform 107 evasive maneuvers, and 400-500 other sea vessels were avoided before the ship docked at Tsumatsuaka.