Aerial manipulation in outdoor environments remains challenging due to the simultaneous requirements of reliable state estimation, stable aerial motion, and precise manipulation under external disturbances. In this work, we present a real-world outdoor aerial manipulation framework that integrates imitation learning, onboard LiDAR-inertial state estimation, and whole-body model predictive control. A Diffusion Policy is trained from manipulation demonstrations to generate desired end-effector motions from onboard observations. These learned commands are executed by a whole-body MPC that jointly coordinates the aerial platform and manipulator to realize the desired end-effector trajectory. To eliminate reliance on external motion-capture infrastructure, the platform employs onboard LiDAR-inertial odometry for state estimation during outdoor operation. We validate the complete framework on a physical aerial manipulator and demonstrate successful execution of outdoor manipulation tasks. The experimental results show that demonstration-driven manipulation policies can be effectively integrated with onboard state estimation and model-based whole-body control to enable aerial manipulation beyond controlled indoor environments.
Figures & tables
Fig. 1 : System Overview
Fig. 2 : Indoor peg-in-hole execution in the training environment, showing the initial configuration, target approach, and successful insertion. Insets show the onboard first person view used by the visuomotor policy.
Fig. 3 : Outdoor peg-in-hole execution in an unseen environment using the same learned diffusion policy, with onboard LiDAR-inertial odometry for state estimation. Insets show the onboard first person view used by the visuomotor policy.
Environment
Norm. EE Error
Max. Base Tilt [deg]
Indoor
0.102
3.32
Outdoor
0.104
1.84
TABLE I : End-effector tracking and base attitude performance in indoor and outdoor experiments.
Fig. 4 : End-effector position tracking performance in both inddor and outdoor environments. Solid lines denote the measured end-effector position, while dashed lines denote the policy-generated reference trajectory.