Taiwan Develops AI Drone Swarms to Counter China's Gray-Zone Pressure

Taiwan is developing an artificial-intelligence system to coordinate swarms of drones for continuous maritime surveillance and to counter China's "gray-zone" operations, the South China Morning Post reported, citing Taiwan's Ocean Affairs Council (OAC).

A research team led by National Cheng Kung University has already completed a simulation of two swarms of 20 drones flying simultaneously. The project marks a shift from controlling individual aircraft to autonomous coordination of whole groups. Gray-zone operations are coercive actions by China that stay below the threshold of open war.

How the system works

The project uses three US-made fixed-wing vertical take-off and landing (VTOL) aircraft: JUMP 20, V-BAT and Flexrotor. With AI and multi-agent reinforcement learning, the drones can distribute missions among themselves, plan routes and keep formation. For flight safety, specialists integrated leader-follower collision-avoidance algorithms and a dynamic obstacle-avoidance system. The algorithms provide spatial separation and keep a mandatory distance of at least 30 metres vertically between drones.

"The next stage will include hardware-in-the-loop simulation, integration of prototypes and flight tests over real sea areas to assess reliability in various maritime and weather conditions," the OAC said. The swarms are intended to become additional "eyes in the sky" for the Taiwanese coast guard when it responds to "various forms of provocation in the gray zone" by Beijing.

Part of a wider programme

The programme is part of an OAC initiative to build an integrated "sea-air" coast guard that combines unmanned platforms with upgraded intelligence, surveillance and reconnaissance (ISR) systems. The OAC noted that wars in Ukraine and the Middle East have shown drones evolving from individually piloted platforms to coordinated swarms of different types.

In July the Taiwanese coast guard received 45 multirotor drones and began a large basic training programme for front-line personnel. It later allocated NT$1.2 billion (US$39 million) to buy 12 US-made V-BAT reconnaissance systems from Shield AI and six control stations; the first delivered V-BAT was shown publicly at Taipei port on 7 October. More in Technology.