Researchers developed a diffusion-based artificial intelligence system that learns ship navigation directly from real-world voyage data rather than relying on optimization algorithms. Trained on a vessel in Japan's busy and geographically complex Seto Inland Sea, the AI learned to balance multiple objectives simultaneously, including collision avoidance, efficient routing, and speed control. The AI also learned local navigation practices, such as staying within designated traffic lanes, without being explicitly programmed to do so. The study demonstrates that diffusion-based AI can capture human-like navigation behavior from real-world data, offering a promising new approach to autonomous ship navigation in complex maritime environments.

New framework enables researchers to model how light and electrons interact in metals, semiconductors and two-dimensional materials, paving the way for faster development of advanced chips, optical technologies and quantum devices.

A new mathematical and computational model developed by researchers at Politecnico di Milano studies the link between the accumulation of toxic proteins and damage to the brain’s blood vessels

New research shows that even in the famous Prisoner's Dilemma, where selfishness is supposed to prevail, cooperation can naturally emerge and persist, offering fresh insight into how complex life and human societies evolved.