
Hangzhou AI competition tests underwater robot autonomy
Teams were required to guide self-developed underwater robots through a full sequence of search, positioning, retrieval and return tasks across the lake’s thousand-acre natural water area.

Teams were required to guide self-developed underwater robots through a full sequence of search, positioning, retrieval and return tasks across the lake’s thousand-acre natural water area.

Galileo’s “Tiejia Tianbao,” or Iron Leopard, team outperformed rivals with its self-developed quadruped robots, excelling in mobility, motion control and all-terrain obstacle crossing.

Under the agreement, the two sides will jointly develop infrastructure spanning data centers, AI model services and computing platforms.
在杭揭牌-1-600x400.jpg)
The launch comes as embodied AI startups face mounting bottlenecks around fragmented computing resources, limited access to training data and a lack of standardized testing environments.

The partnership will combine Rokid’s AR and AI hardware ecosystem with Seichitech’s expertise in Micro LED inspection technology.

The company said the data reflects how intelligent driving systems are increasingly becoming standard features for everyday commuting rather than experimental add-ons.

This year’s competition focuses on industrial, service and special-use applications, with all tasks derived from practical pain points identified by leading companies rather than abstract benchmarks.

KeenCool said its immersion cooling systems can reduce data center power usage effectiveness, or PUE, to below 1.08, cutting overall energy consumption by nearly 40%.

The methanol-powered system is also positioned as an alternative to battery-only agricultural machinery in northern China’s cold climate.

The sector aims to reach an annual output of 8 billion yuan by 2027, with 30 takeoff and landing sites, 40 drone routes, and low-altitude flight hours climbing from 5,000 to 75,000 annually.