Autonomous UAV-Quadruped Docking in Complex Terrains via Active Posture Alignment and Constraint-Aware Control

TL;DR

An autonomous UAV-quadruped docking framework for GPS-denied environments that is validated in both simulation and real-world scenarios, successfully achieving docking on outdoor staircases higher than 17 cm and rough slopes steeper than 30 degrees.

Abstract

Autonomous docking between Unmanned Aerial Vehicles (UAVs) and ground robots is essential for heterogeneous systems, yet most existing approaches target wheeled platforms whose limited mobility constrains exploration in complex terrains. Quadruped robots offer superior adaptability but undergo frequent posture variations, making it difficult to provide a stable landing surface for UAVs. To address these challenges, we propose an autonomous UAV-quadruped docking framework for GPS-denied environments. On the quadruped side, a Hybrid Internal Model with Horizontal Alignment (HIM-HA), learned via deep reinforcement learning, actively stabilizes the torso to provide a level platform. On the UAV side, a three-phase strategy is adopted, consisting of long-range acquisition with a median-filtered YOLOv8 detector, close-range tracking with a constraint-aware controller that integrates a Nonsingular Fast Terminal Sliding Mode Controller (NFTSMC) and a logarithmic Barrier Function (BF) to guarantee finite-time error convergence under field-of-view (FOV) constraints, and terminal descent guided by a Safety Period (SP) mechanism that jointly verifies tracking accuracy and platform stability. The proposed framework is validated in both simulation and real-world scenarios, successfully achieving docking on outdoor staircases higher than 17 cm and rough slopes steeper than 30 degrees. Supplementary materials and videos are available at: https://uav-quadruped-docking.github.io.

Figures (8)

• Figure 1: Docking experiments in complex terrains. The top-left and top-right show tests on a $17\,\text{cm}$ outdoor stair where the quadruped actively adjusts its posture while ascending and descending. The bottom-left shows an indoor $15\,\text{cm}$ stair experiment, and the bottom-right presents docking on a $30^{\circ}$ outdoor rough slope, demonstrating the robustness of the proposed method.
• Figure 2: An overview of our UAV-Quadruped collaborative autonomous docking scheme. Both the UAV and the quadruped robot actively participate in the docking process, transferring data information between them. The UAV sends the docking request “docking signal” to the quadruped robot, and the quadruped robot actively aligns the posture after receiving the signal, and sends the IMU readings together with velocity estimation derived from the Estimator to the UAV.
• Figure 3: Illustration of the three docking phases: acquisition, tracking, and landing.
• Figure 4: Our experimental platform for UAV–quadruped docking, showing the quadrotor UAV equipped with onboard sensors and the Unitree quadruped robot with an AprilTag landing plate.
• Figure 5: HIM-HA Policy Testing in Different Terrains. We conducted tests on the HIM-HA method's capability to adjust the robot’s posture under various terrain conditions with different difficulty levels in simulation (Top: stair, Bottom: slope). At step 200, the robot receives a docking signal, requiring it to adjust its posture to a horizontal state. At step 300, the robot is instructed to restore its original posture. The vertical axis $g_{z}$ represents the projection of the normalized gravity vector onto the body-frame $z$-axis. Values closer to $-1$ indicate that the robot’s torso is more horizontally aligned with the ground.