Industry

Yuhang’s AI-powered drones can fly without GPS signals

2025-03-13

By Jin Yingying

Recently, at Yuhang Differential Robotics, researchers demonstrated the complete process of the autonomous exploration drone P300 taking off, flying according to preset design parameters, and smoothly navigating obstacles. It also returned flight video and high-definition scanning data.

The breakthrough in this drone’s core technology lies in its fully autonomous exploration algorithm, which allows it to fly independently without GPS signals or human intervention.

Differential Robotics, located in China Artificial Intelligence Town, is a high-tech enterprise founded by Gao Fei, an associate professor and Ph.D. supervisor at Zhejiang University. The company focuses on research and innovation in robotics and artificial intelligence, offering fully autonomous drone mapping and inspection solutions integrated for industrial users.

Currently, over 90% of consumer and commercial drones rely on GPS systems, which are critical for core functions such as automatic return, precise hovering, and route planning.

With the integration of AI algorithms, the drone is equipped with an “intelligent brain”, enabling it to make autonomous decisions. Its response time can reach milliseconds or even faster, allowing it to avoid obstacles effortlessly during flight.

The P300 is capable of performing exploration and mapping tasks in unknown or restricted environments, with applications across mining, forestry, chemical industries, construction, urban management, and emergency response. It has already been deployed in regions such as Beijing, Zhejiang, Jiangsu, Tianjin, Jiangxi, Sichuan, and Yunnan, reducing the labor intensity of manual inspections and significantly improving inspection efficiency.

One notable application of the P300 was its exploration of operational mine shafts and long-abandoned mine tunnels. Replacing human exploration of unknown risks, the drone collected data in a complex mining area nearly 20 stories high, with a volume of over 10,000 cubic meters, to create a high-precision 3D model. This provided valuable support for safety assessments and future planning in the mining sector.