Shield AI's Hivemind Links Two Thunder Tiger SeaShark USVs in Six Weeks

From simulation to sea trials

Shield AI describes Hivemind as mission autonomy software that enables unmanned systems to see, think, and act with minimal human input. According to the company, most autonomous unmanned surface vehicles (USVs) today focus on individual platform functions, such as following a predefined route, while Hivemind is built to manage the mission itself, including how multiple vessels work together as conditions change.

Shield AI partnered with Thunder Tiger Corp., a Taiwan-based manufacturer of unmanned systems, to integrate Hivemind aboard the SeaShark 600 and SeaShark 800. Shield AI says the software's modular, open architecture let the teams build on work from dozens of earlier platform integrations.

The work began in simulation. Shield AI engineers used digital representations of both vessels to develop mission behaviors before operating the physical USVs. In parallel, the teams prepared Hivemind's interfaces with the vessels' systems and sensors, then moved to vehicle-in-the-loop testing with the actual hardware. According to Shield AI, the demonstration showed the same autonomy architecture could manage assignments across two different SeaShark models, and the effort created a repeatable path for integrating Hivemind aboard other USVs.

Bryan L., Thunder Tiger's USV Division Lead, said the company chose Shield AI because "they have already integrated and tested on 30+ platforms."

Building a shared picture in Dapeng Bay

The intelligence, surveillance and reconnaissance (ISR) demonstration was a shared patrol mission. Mission planners divided the patrol area into two sections and assigned one SeaShark to each. Both vessels used maritime radar and Automatic Identification System (AIS) data to monitor their surroundings. Planners then sent a third boat into the area as the "vessel of interest."

To respond together, the SeaSharks needed a consistent view of that boat. Radar estimated its position, while AIS broadcasts supplied identity and location. Because the two sources update at different rates, they usually report slightly different positions. Hivemind compared the reported position and movement over time, determined which reports described the same boat, and fused them into a single track. It then shared that track across both SeaSharks. Meanwhile, the SeaShark 600's electro-optical/infrared sensor sent live video of the vessel to operators ashore.

Switching from Patrol to Escort

Once the shared track was established, Hivemind temporarily suspended the patrol assignments and started a higher-priority Escort behavior. In this mission, escorting meant accompanying the vessel of interest to keep continuous watch on its movement and activity. One SeaShark took the port side and the other the starboard side, which gave operators complementary full-motion video views and placed the USVs in position to support interdiction if directed.

After the vessel left the patrol area, Hivemind ended the escort and sent both SeaSharks back to their sectors. According to Shield AI, the operator did not have to rebuild the patrol plan or manually reposition each USV.

Simulation points to wider coverage

Shield AI also modeled a larger deployment. In the simulation, five USVs using adaptive mission autonomy covered a 50-nautical-mile stretch, compared with 6 nautical miles for five USVs using fixed patrol patterns. Both maintained a 95% vessel identification rate, which Shield AI puts at roughly 8.3 times the coverage with the same number of vessels. The company cautions that actual coverage depends on sensor performance, vessel speed, and operating conditions.

The pitch is aimed at a practical problem: Taiwan's coastline stretches more than 1,500 kilometers across the main island and outlying islands, and adding USVs increases the time operators spend directing each vessel. Shield AI argues that mission autonomy lets a fleet grow without operators steering every movement, leaving them to decide which vessels need attention and how to respond.

Thunder Tiger Chairman William Chen said the company is "very optimistic about Hivemind, which brings autonomy to USVs for area search, target tracking, and the execution of these missions."

What it means for maritime surveillance

The Dapeng Bay trial is a single demonstration, and the larger coverage figure comes from simulation rather than sea trials. Still, Shield AI says it shows a foundation for fleets that carry out more of the mission themselves. Paired with Taiwan's domestic USV manufacturing, the company says Hivemind offers a way to get more surveillance capability from vessels already fielded, and it sees the same approach applying to other countries with extensive coastlines.