It was developed by scientists from Westlake University in China the FlyingToolbox system, enabling micro aircraft (MAV) to replace tools during flight with subcentimeter accuracy – despite strong airflow disturbances. The innovation promises a complete transformation of hazardous tasks such as skyscraper maintenance, disaster response and industrial construction that are risky or impossible for human access.
Flying drones in tight formation has long been thwarted by “jet flow” – the strong downward flow of air from the propellers. When one drone hovers directly above another, speeds can exceed 13 meters per second, generating forces of up to 25 Newtons (40% of the drone's weight). In the past, this turbulence limited docking precision to 6-8 cm in similar systems, making mid-air tool changes unreliable.
FlyingToolbox breaks down these barriers. The system combines two specialized MAVs: a lower “toolbox” drone carrying multiple tools on flexible cables, and an upper “manipulator” drone with a robotic arm. Inspired by the work of a surgical team in which a nurse hands instruments to a doctor, the configuration enables uninterrupted work in the air.
The heart of the system is an estimator based on a neural network, which predicts and compensates for disturbances caused by precipitation in real time. Thanks to visual tracking via QR code, drones achieve precise alignment even in turbulent air conditions. Docking is ensured by electromagnetic connectors with flexible cables that absorb positioning errors and ensure reliable transfer of the tool.
During extensive testing, the system achieved a docking accuracy of 0.80 ± 0.33 cm, even at rinsing speeds up to 13.18 m/s. FlyingToolbox successfully completed 20 consecutive docking attempts, maintaining accuracy and repeatability throughout. This represents a significant improvement over previous air docking systems.
Experiments ranged from stationary tool changing to dynamic sequences with moving drones carrying tools. The multi-step tasks simulated real-world scenarios such as sequential repairs with 100% success. The system's robustness was evident in the airborne formations, in which the drones shared the work like a coordinated airborne workforce.
Although the experiments were conducted in controlled laboratory conditions, the researchers believe the technology could be adapted to real-world environments, allowing drones to autonomously refuel, replace tools or deliver materials while in flight. Such advances could change the way drones are used in industrial maintenance, construction and disaster response, enabling multiple drones to work together as a workforce capable of performing complex, coordinated missions.
The Westlake University team continues to refine the FlyingToolbox system. Future improvements include multi-tool compatibility, improved robotic arms with more degrees of freedom, and wind-resistant algorithms.
Projects like FlyingToolbox and QuData's UAV AI development platform reflect a broader movement in modern robotics – a shift toward adaptive, multi-functional systems that can seamlessly adapt to changing conditions and mission requirements. This evolution redefines automation by combining precision engineering, intelligent control and real-time decision-making to deliver greater efficiency and resiliency across industries.
QuData engineers are improving navigation and coordination technologies for GPS-free environments, ensuring drones will continue to operate even when satellite signals are lost – for example, in natural disaster zones, urban canyons, or underground areas. These systems play a key role in crisis response, infrastructure monitoring and air support missions where reliable autonomy and situational awareness can be critical. Collectively, such achievements expand the boundaries of what autonomous aerial robotics can achieve in real-world conditions.
