Fujitsu’s “Zinrai AI” Beats Humans in Port Ship Guidance and Collision Avoidance

By: |

Fujitsu has developed an AI (artificial intelligence) named “Zinrai” which specializes in ship management on ports. Large numbers of cargo ships come and go in busy ports, and managing these behemoths is a taxing task that strains the nerves of even the most experienced experts.

Ships need to move in and out the port without colliding with other vessels, reach loading or unloading docks quickly, and pass through customs for clearance. Delaying any of that or not doing it efficiently results in loss of time, and in turn, loss of money.

Fujitsu thought that with machine learning, they could take the guesswork out of the equation, and let the computer decide how to handle large vessel traffic in the most efficient and totally safe manner.

During the trials of the system that lasted from December 2019 until March 2020, “Zinrai” demonstrated superior ability to manage ship movement in the Tokyo Bay area compared to even the most capable officials. In fact, the AI could predict risks of collisions two minutes quicker than conventional port management systems which can make a huge difference in practice.

Each year in Japan, 280 maritime collisions occur. Sometimes, supply chains are interrupted, the crew’s safety is jeopardized, and the environment could suffer from catastrophic pollution. With machine-learning systems like the Zinrai, the Japanese port authorities hope that the number of these collisions will be dramatically reduced. During the trial period, Fujitsu’s AI sent twice the number of alerts to ship captains compared to what the existing systems would send, and they were all justified (no false alarms).

For the next couple of years, operation controllers will stay at the helm of managing port traffic, with systems like Zinrai assisting them by generating consulting collision alerts. In the future though, we may see these systems fully undertake the role of maritime traffic management, eliminating human errors, delays, and negligence.

Image by Frauke Feind from Pixabay