Drones Guiding Drones: Cooperative Navigation of a Less-Equipped Micro Aerial Vehicle in Cluttered Environments

TL;DR

The paper addresses GNSS-denied navigation by enabling a minimal SUAV to operate through cluttered environments with the sensing, mapping, and planning burden offloaded to a more capable PUAV. The authors propose a cooperative, onboard framework where the PUAV builds a dense occupancy map $\mathcal{M}$ from 3D LiDAR, estimates relative pose to the SUAV, and jointly plans collision-free paths $\mathcal{P}_\mathrm{P}$ and $\mathcal{P}_\mathrm{S}$ so the SUAV can follow $\mathcal{P}_\mathrm{S}$ while the PUAV tracks to maximize line-of-sight via a LOS-aware guiding viewpoint. A unique relative-localization-aware planner, integrated with LiDAR-VIO fusion to estimate $\prescript{{S}}{{L}}{\mathbf{T}}$, enables robust guiding in GNSS-denied clutter. The approach is demonstrated in simulations and real-world experiments, including forest navigation and narrow-gap passages, with onboard computation and minimal communication bandwidth, illustrating the practical viability of heterogeneous UAV cooperation for tiny aerial platforms operating without external localization or computation.

Abstract

Reliable deployment of Unmanned Aerial Vehicles (UAVs) in cluttered unknown environments requires accurate sensors for Global Navigation Satellite System (GNSS)-denied localization and obstacle avoidance. Such a requirement limits the usage of cheap and micro-scale vehicles with constrained payload capacity if industrial-grade reliability and precision are required. This paper investigates the possibility of offloading the necessity to carry heavy sensors to another member of the UAV team while preserving the desired capability of the smaller robot intended for exploring narrow passages. A novel cooperative guidance framework offloading the sensing requirements from a minimalistic secondary UAV to a superior primary UAV is proposed. The primary UAV constructs a dense occupancy map of the environment and plans collision-free paths for both UAVs to ensure reaching the desired secondary UAV's goals even in areas not accessible by the bigger robot. The primary UAV guides the secondary UAV to follow the planned path while tracking the UAV using Light Detection and Ranging (LiDAR)-based relative localization. The proposed approach was verified in real-world experiments with a heterogeneous team of a 3D LiDAR-equipped primary UAV and a micro-scale camera-equipped secondary UAV moving autonomously through unknown cluttered GNSS-denied environments with the proposed framework running fully on board the UAVs.

Figures (9)

• Figure 1: The lidar-equipped primary UAV guides the secondary camera-equipped UAV through (a) narrow passages in an industrial complex, (b) a cluttered forest environment.
• Figure 2: The PUAV with body frame ${P}$ is localized in its local frame ${L}$, builds dense occupancy map $\mathcal{M}$, and plans collision-free paths for both UAV. The SUAV with body frame ${S}$ is localized in its local frame ${V}$. ${W}$ denotes the fixed world frame. All the reference frames are gravity-aligned. The PUAV periodically guides the SUAV to follow the planned path ${\prescript{{L}}{{}}{\mathcal{P}_\mathrm{S}}}$. Black dotted lines mark the line of sight between the PUAV position and the SUAV waypoints.
• Figure 3: The PUAV maps the surrounding environment using 3D lidar data, performs estimation of relative pose of the SUAV from lidar detections and received VIO data, and guides the SUAV to desired poses. The position control pipeline in the frame of each respective self-localization method is provided by the MRS UAV System bacaMRSUAVSystem2021.
• Figure 4: Example of guiding viewpoint selection. SUAV path (red line) consists of 3 waypoints passing through the door in the upper right corner of the map. The algorithm constructs regions from which the waypoints are visible (a), a safety buffer around the SUAV path (b), safe space w.r.t. obstacles for the PUAV (c), and the final intersection $\mathcal{I}_\mathrm{all}$ (d).
• Figure 5: The simulation with a gap of varying width and UAV trajectories from the test of the proposed guiding approach.