Researchers at Pusan National University in South Korea have developed an AI-driven ship navigation system that uses real-time weather data to cut harmful coastal air pollution without sacrificing fuel efficiency. The framework combines environmental sensor data, physics-informed deep learning and multi-objective optimisation to recommend routes and speed profiles that reduce pollution exposure for communities near ports.

Rather than relying on blanket slow steaming, the system predicts how exhaust plumes will disperse under changing wind and weather conditions, then adjusts a vessel’s route and speed to take advantage of more favourable meteorological windows. The team argues the greatest public health risk is often not how much a ship emits, but where and when those emissions are carried.

In simulations including scenarios around Busan Port, the framework improved optimisation performance by 20% to 35% over conventional navigation methods while cutting peak pollutant exposure by 34% to 78%. The study was led by assistant professor Dowon Kim alongside PhD student Seongbeom Park and professor Jinhyeok Yun.

Select the link to learn more.