Towards Non-Robocentric Dynamic Landing of Quadrotor UAVs

TL;DR

This paper addresses dynamic quadrotor landing without onboard exteroceptive sensors by shifting sensing and computation to ground stations. It introduces a relative state estimation module based on an IEKF on a pose manifold $\mathcal{M}=\mathbb{R}^3\times\mathrm{SO}(3)\times\mathbb{R}^3$ using LED markers, and a ground-based motion planner that leverages differential flatness to produce a minimum-jerk trajectory via Bézier basis with a visibility-safety corridor. The key innovations are the offboard sensing/control paradigm, IEKF-based robust pose estimation, and a QP-driven trajectory generation that respects FoV and dynamic constraints, demonstrated by successfully landing an $18\text{ cm} \times 18\text{ cm}$ quadrotor on a $43\text{ cm} \times 43\text{ cm}$ pad with detailed hardware/software disclosures. The results show comparable performance to conventional onboard methods, while reducing UAV payload and power demands, with potential implications for scalable UAV swarms and cost-effective dynamic landing in smart-city scenarios.

Abstract

In this work, we propose a dynamic landing solution without the need for onboard exteroceptive sensors and an expensive computation unit, where all localization and control modules are carried out on the ground in a non-inertial frame. Our system starts with a relative state estimator of the aerial robot from the perspective of the landing platform, where the state tracking of the UAV is done through a set of onboard LED markers and an on-ground camera; the state is expressed geometrically on manifold, and is returned by Iterated Extended Kalman filter (IEKF) algorithm. Subsequently, a motion planning module is developed to guide the landing process, formulating it as a minimum jerk trajectory by applying the differential flatness property. Considering visibility and dynamic constraints, the problem is solved using quadratic programming, and the final motion primitive is expressed through piecewise polynomials. Through a series of experiments, the applicability of this approach is validated by successfully landing 18 cm x 18 cm quadrotor on a 43 cm x 43 cm platform, exhibiting performance comparable to conventional methods. Finally, we provide comprehensive hardware and software details to the research community for future reference.

Figures (8)

• Figure 1: Transformation among inertial frame $\mathcal{N}$, non-inertial local target frame $\mathcal{N}$, and quadrotor body frame $\mathcal{B}$.
• Figure 2: A graphic presentation of the proposed VSFC $\mathcal{S}_L$, which consists two subsets, rear space $\mathcal{S}_{L_R}$ and touchdown space $\mathcal{S}_{L_T}$. Visualization package developed by liu2017planning is used here.
• Figure 3: The predefined finite state machine.
• Figure 4: Experimental setup for the proposed system.
• Figure 5: Relative state estimation. (a) Quadrotor moving in a block path in front of the camera. (b) Random maneuver in front of the camera. (c) Camera moving in block path, while quadrotor follows. (d) Camera moving in block path, while quadrotor flies in block path defined in the non-inertial frame $\mathcal{N}$.