Fly, Track, Land: Infrastructure-less Magnetic Localization for Heterogeneous UAV-UGV Teaming

A research team from the University of Bologna and ETH Zurich has developed a breakthrough infrastructure-less localization system that enables drones to autonomously track and land on moving ground robots with centimeter precision. Their paper, "Fly, Track, Land: Infrastructure-less Magnetic Localization for Heterogeneous UAV-UGV Teaming," presents a magneto-inductive (MI) approach that achieves 5 cm 3D position accuracy at 20 Hz refresh rates using only a single lightweight receive coil on the drone. This system requires no GPS, cameras, or external anchors, making it ideal for GPS-denied environments like disaster zones or indoor facilities.

The technology represents a significant advance in heterogeneous robot collaboration, where ultra-lightweight flying robots (typically under 100g) can serve as mobile perception agents for ground-based Unmanned Ground Vehicles (UGVs). In real-world experiments, the system demonstrated successful autonomous landings on a moving quadruped robot while maintaining 7.2 cm RMSE tracking accuracy. The embedded estimation pipeline fuses magnetic measurements with inertial and optical-flow data in an onboard extended Kalman filter, providing robust performance even in dynamic scenarios where the docking platform is in motion.

This infrastructure-free approach bridges the gap between coarse navigation and precise docking, enabling practical applications where traditional localization methods fail. The system's compact, embedded design makes it suitable for resource-constrained platforms while delivering the precision needed for reliable physical interaction between aerial and ground robots. By extending the sensing horizon and providing complementary viewpoints, this technology could revolutionize how robot teams explore and map large-scale unknown environments.