Organizations: Department of Engineering, University of Perugia, Perugia, 06125, Italy · Multi-Robot Systems Group, Faculty of Electrical Engineering, Czech Technical University in Prague, Prague, 16627, Czech Republic · Department of Information Technology and Electrical Engineering (D-ITET), ETH Zurich, Zurich, 8092, Switzerland
Persistent air ground robot teams require nano UAVs that can leave a mobile UGV, perform short scouting or inspection tasks, and reliably return to a compact landing interface. This final docking phase remains difficult: it demands centimeter-scale relative localization on a moving platform, while nano UAVs provide only a few grams of payload and severe onboard sensing, power, and computation constraints. We address this problem with an infrastructure-less magneto inductive anchor-tag localization system for heterogeneous UAV-UGV teaming. Unlike passive magnetic docking aids or approaches exploiting environmental magnetic features, the proposed system actively generates a frequency multiplexed AC magnetic field on the UGV and uses it as an onboard localization reference for closed loop flight. The nano UAV carries only a lightweight passive receive coil and estimates its 3D position directly in the UGV frame, providing a local close range reference for hovering, tracking, and landing without external anchors, visual fiducials, GNSS, or motion capture online. The magnetic estimate is fused with the native onboard sensing stack, enabling operation under nano UAV SWaP constraints. Experiments with a Unitree A1 quadruped and a Crazyflie nano UAV, evaluated against motion capture ground truth, show centimeter level accuracy in static hovering and landing and approximately 8-11 cm RMSE during UGV motion, while the flow only baseline frequently violates the safety bound and fails the task.
Figures & tables
Figure 1: Representative picture of a nano-size UAV autonomously landing on a moving legged robot and GNSS-denied environment. The infrastructure-free localization systems relies on onboard sensors, combining IMUs, optical-flow camera, UWB, and the MI system. The four reference coils are represented by the red lines with a magnetic dipole shape. The landing deck can be used to recharge the UAV’s batteries.
Paper
Localization Method
Infrastructure-less
Moving Landing
Landing metric
UAV-UGV Teaming
Field Test
nano UAV
Liu et al. [2025]
UWB+IMU+Vision
✗
✓
-
✗
Outdoor
✗
De Silva and Katupitiya [2024]
UWB ranging (2 UAVs)
✓
✗
14cm
✓
Outdoor
✗
Polonelli et al. [2022a]
GNSS + UWB
✗
✗
±10cm
✗ (static sensors)
Outdoor
✗
Nguyen et al. [2024]
Vision + motion
✓
✓
± 5 °
✗
Simulation
✗
Ladosz et al. [2024]
Vision
✓
✓
80% SC
✗
Sim. + Indoor
✓
Zheng et al. [2025]
SLAM (shared map.) + manual
✗
✗
-
✓
Outdoor
✗
Table 1: Comparison of UAV-UGV landing and cooperative systems in recent literature.
Figure 2: Custom Magnetic Hardware. (a) The Unitree A1 legged robot UGV. (b) One of the four lightweight MI coils used for precise localization. (c) The Crazyflie 2.1 nano UAV. (d) The landing pad with a diameter of 22cm mounted on UGV. (e) The wiring for the AnchorDeck and the four coils, plus the connections for the wireless battery charger. (f) The ultra-lightweight MI coil mounted on the Crazyflie 2.1 nano UAV and the MagneticDeck. (g) Motion caption system markers used to assess the system performances.
Parameter
Value
Platform
UGV / UAV
Unitree A1 / Crazyflie 2.1
UAV mass / MI payload
47 g / 9 g
Landing pad / Target
25 × 44 cm / ∅ 22 cm
Coils
TX anchors / RX tag
4 / 1 passive coil
Table 2: Main parameters of the MI anchor-tag localization system.
Figure 4: Visual overview of the experimental validation scenarios using the proposed magnetic localization system. (a)-(c) S1 (Static Hovering & Landing): The nano UAV performs an autonomous sequence of takeoff, hovering, and precision landing on the stationary UGV. (d)-(f) S2 (Linear Tracking & Landing): The nano UAV tracks the UGV moving along a linear trajectory. Visualization includes (d) 3D perspective, (e) lateral view showing the relative distance maintenance ( T0 to T1 ), and (f) top-down view of the alignment. (g)-(h) S3 (Composite Motion Tracking): The UAV tracks and follows the UGV performing complex planar maneuvers with varying velocity and direction, maintaining the relative position within the magnetic workspace.
Test
Hovering
in-out
Mag+Flow
Flow
Mag+Flow
Flow
1
1.14
4.56
6.93
FAIL
2
0.80
4.55
6.21
FAIL
3
1.79
2.56
FAIL
FAIL
4
1.76
5.36
6.94
FAIL
5
4.01
5.46
6.07
FAIL
Table 3: Scenario S1: The UGV is static. For Hovering, the UAV performs the sequence in Figure 4(a) , Figure 4(b) , Figure 4(c) . For the test in-out , the sequence includes also a movement ( 0.6m ) outside the landing pad. In summary, the sequence is: takeoff, fly 0.6m ahead, fly back, landing. Values represent RMSE in centimeter, success rate (SC) in percentage.
Figure 5: Touchdown accuracy analysis for scenario S1 -Hovering. The plot shows the planar 2D position of the UAV relative to the docking pad center at the moment of landing for the proposed Mag+Flow method (circles).
Figure 6: Scenario S2 : time-history qualitative results for a representative linear docking sequence. The colored bands highlight the mission phases (pre-takeoff, takeoff/approach, hovering, return over the UGV, landing). Top: x position in the world frame. Middle: altitude z . Bottom: magnitude of the FFT peaks of the coil voltage, which provides a proxy of the received magnetic signal strength over time.
1
2
3
4
5
6
7
8
9
10
Mean
SC
S2
6.04
14.7
7.11
6.09
10.7
5.56
13.5
FAIL
6.51
FAIL
8.77
80%
S3
9.39
8.14
27.1
7.14
11.3
7.09
7.34
8.52
9.53
9.46
10.5
100%
Table 4: Linear tracking and composite motion results. Performance comparison during UGV planar forward-backward motion ( S2 ) and during complex UGV trajectories ( S3 ). RMSE calculated on the whole path. Values represent RMSE in centimeter, success rate (SC) in percentage.
Figure 7: Scenario S3 (Test 1): qualitative time-aligned analysis of dynamic tracking. The numbered markers (1–7) identify key flight phases via synchronized video snapshots and the corresponding dashed vertical lines in the time traces. The trajectory plots (3D and top-down) compare Ref. (World frame) , UGV (World frame) , and EKF (World frame) . Markers 1–2 (early takeoff/settling) highlight small transient oscillations.
Figure 8: Model-based sensitivity of the magnetic localization pipeline to uncompensated UGV attitude. The comparison shows that uncompensated UGV yaw, roll, and pitch are mapped by the current position-only estimator into systematic position biases.
Motion
Angle
Final projected error
Final 3D error
Mean 3D error
[deg]
[cm]
[cm]
[cm]
Yaw
0
2.28
3.45
4.51
Yaw
10
14.05
14.15
14.76
Yaw
40
66.33
66.38
67.38
Roll
0
3.39
3.45
4.51
Roll
5
10.74
10.81
10.76
Table 5: model based sensitivity of the position only magnetic estimator to uncompensated UGV attitude. Errors are computed in the world frame. The projected error is reported in the dominant affected plane: xy for yaw, yz for roll, and xz for pitch.
Nano unmanned aerial vehicles (nano-UAVs) can navigate confined spaces that larger robots cannot, but their payload capacity severely limits the sensors and compute available for self-localization in global navigation satellite system (GNSS)-denied environments. We present a vision-based system that localizes a nano-UAV from a quadruped robot with an arm-mounted camera. The quadruped tracks the drone, estimates its position in its own coordinate system using segmentation masks and depth, and transmits that position over a real-time radio link. To ensure continuous tracking, we developed a perception-aware nonlinear Model Predictive Controller (NMPC) that dynamically adjusts the quadruped's body and arm to maximize the drone's visibility, treating observation reliability as a primary control objective. Relying solely on this external estimate, the nano-UAV executed predefined trajectories from takeoff to landing with a median 3D localization error of 59 mm. The system demonstrates robustness against short visual occlusions by smoothly transitioning to onboard inertial flight when line-of-sight is temporarily lost. Ultimately, this framework allows a quadruped to offload the localization burden of a nano-UAV, enabling inspection of complex spaces that neither robot could navigate alone.
Alejandro Lorite Mora, Dimitrios Arapis, Andrés Faíña
IT University of Copenhagen, Denmark · Novo Nordisk A/S, Hillerød, Denmark
Micro unmanned aerial vehicles (micro-UAVs) are small enough to reach confined spaces that larger robots cannot access, but too small to carry the sensing and computing power required for autonomous flight. We move the localization stack entirely off the aerial platform onto a quadruped robot with a 7-degree-of-freedom (DOF) arm, which supplies the micro-UAV (27 g bare, 42 g with fiducial markers) its full 6-DOF pose. A camera at the arm's end-effector detects AprilTag fiducial markers on the drone and composes that observation with the quadruped's own self-localization to place the drone in a shared map frame, so the ground robot localizes its partner, rather than only tracking it relative to the camera. The arm acts as an actively-controlled observer, repositioning to keep the drone in view as both robots move; the drone carries only an inertial measurement unit and fuses the external pose to fly commanded setpoints. In lab flights the external pose is accurate to 12-16 mm, enough to fly the drone autonomously within 2-5 cm of motion-capture-fed control. Having the quadruped actively follow the drone reduces the tracking error from 11.0 cm to 6.9 cm by holding the camera in the close range, where the markers are most accurate.
Alejandro Lorite Mora, Andrés Faíña
IT University of Copenhagen, Rued Langgaards Vej 7, Copenhagen 2300, Denmark · Helix Lab, Campus Kalundborg 3, 4400 Kalundborg, Denmark · Novo Nordisk A/S, Hallas Alle 1, 4400 Kalundborg, Denmark
Accurate localization of unmanned aerial vehicles (UAVs) is essential for applications such as structural health monitoring, especially in environments where Global Positioning System (GPS) signals are denied or unreliable, like indoor spaces, tunnels, urban canyons, or areas beneath large structures. To address this challenge, we propose Cross-Fusion, a novel method for real-time UAV localization that integrates data from a 3D Light Detection and Ranging (LiDAR) and a monocular camera. A key contribution is its cross-session fusion strategy, which integrates visual and geometric information collected from multiple agents during routine baseline surveys to improve localization consistency and map completeness. The system employs LiDAR-based odometry for motion tracking and image-based feature matching via a single red-green-blue (RGB) camera to correct drift and improve accuracy. Unlike visual-inertial systems, Cross-Fusion maintains a simple sensor setup and avoids the complexity of stereo or global shutter configurations. Experimental results demonstrate that Cross-Fusion achieves localization accuracy comparable to GPS-based methods and performs reliably in challenging feature-sparse environments.
Cong Hoang Quach, Chi Thanh Vo, Dong LT. Tran +3
1 – School of Electrical and Data Engineering, University of Technology Sydney, New South Wales, Australia · 2 – Institute of Aerospace Engineering and Technology, Duy Tan University, Da Nang, Vietnam · 5 – Department of Artificial Intelligence and Robotics, Sejong University, Seoul, South Korea +2