In Simulations, the AI Correctly Identified Vessel Intentions 89.6% of the Time, Compared With 68.5% for a Conventional Rule-based System.

Chinese researchers have developed an AI system that reduced the simulated risk of a coastguard vessel using lethal weapons during disputed-water encounters from 0.3% to zero, according to a study published this month in the Chinese journal Command Control Simulation. The research, led by Sun Shengzhi, a professor at the China Coast Guard Academy, was designed to help commanders distinguish between routine maritime activity and deliberate probing or harassment in the so-called 'grey zone' between peace and open conflict. In simulations, the AI correctly identified vessel intentions 89.6% of the time, compared with 68.5% for a conventional rule-based system.

The system combines radar and other sensor information with a causal model to determine why a vessel might be manoeuvring in a particular way. The researchers built a computer war game around a scenario involving a Chinese coastguard vessel protecting fishing boats from simulated foreign vessels carrying out low-intensity harassment and reconnaissance, the South China Morning Post reported. The scenario reflects encounters that frequently occur in disputed waters, where vessels can operate aggressively without crossing the threshold into conventional military conflict.

Simulated vessels were required to maintain a minimum separation of 0.5 nautical miles, or about 3,000 feet, while following restrictions governing weapons use. Under the existing rule-based system, the simulated coastguard vessel used lethal weapons in 0.3% of scenarios — a figure the researchers argued could be unacceptable in sensitive maritime confrontations. In one simulation, the AI initially assigned a 72% probability that a nearby vessel was simply passing through.

When three vessels formed a wedge and approached from different directions, the system raised the probability of coordinated probing to 68% and ordered the use of a water cannon rather than escalating immediately. The AI never triggered a weapons-use violation across the simulated scenarios and reduced other rule violations, such as vessels moving too close to one another, from 1.2% to 0.8%. It was also tested against deceptive behaviour: in one scenario, fishing boats appeared to retreat while vessels from another direction accelerated toward the coastguard ship, and the AI interpreted the manoeuvre as a potential deception, maintaining its position rather than reacting to the apparent retreat. While the system is not intended to remove human commanders from the decision-making process, the researchers envision AI supporting human operators by combining intention recognition, constrained decision-making and adaptive simulations.