Druid Drone—A portable unmanned aerial vehicle with a multifunctional manipulator for forest canopy and mistletoe research and management

The forest canopy, as a biodiversity hotspot with many wildlife habitats, remains a difficult site to access for researchers and forest managers. Here, we present a novel, small, unmanned aerial vehicle, called the Druid Drone (DD), equipped with a multifunctional manipulator designed for tree crown management and biodiversity research. Its use is demonstrated by studies and the control of the widespread, obligatory hemiparasitic European mistletoe (Viscum album L.). The DD provides integrated pest management by canopy surveying, spraying and sampling of leaves/branches for further studies. The pilot and operator, using first person view goggles, handle the lightweight backpack‐sized drone (DJI Mavic 2 Pro, Air 2, Phantom 3 or custom Rotorama quadcopter) equipped with a manipulator and five replaceable 3D‐printed functional modules. These include the peripheral high‐resolution Observer camera for close, non‐invasive inspection of tree crowns, the entomological Collector unit for arthropod trapping, the forceps arm Sampler for soft plant tissue collection, the terminal Secator saw for cutting harder tree tissues such as twigs and the Green Doctor precision sprayer, which disperses growth regulators or other compounds for targeted tree and mistletoe management. This method offers a potential solution for safe, precise and minimally invasive research on forest canopy biodiversity in different forest types across the globe, tree health care and mistletoe management as compared to the manual work of arborists or the use of canopy cranes, forest helicopters and firearms.

special cameras and/or manipulators, whose application in plant biology has progressed quite rapidly. Thus, aerial photography on UAV platforms has been employed for invasive species tracking (Cruzan et al., 2016), field phenotyping, studies of tree flowering (López-Granados et al., 2019), analysis of tree mortality (Khokthong et al., 2019) and tree species identification in forests (Kentsch et al., 2021). Drone-based light detection and ranging (LiDAR) scanning and multispectral imagery have facilitated the monitoring of tropical forest diversity (De Almeida et al., 2021) and mapping of plant functional traits in the high Arctic (Thomson et al., 2021).
Tree phenology, distribution, coverage and individual crown delineation as well as infection status and biotic interactions can be remotely evaluated using sUAV-borne imagery (Gu & Congalton, 2021).
Various plant sampling drone platforms have been proposed, such as the Flying Tree Top Sampler (Käslin et al., 2018), DeLeaves' Tree sampler (Charron et al., 2020), the assisted canopy sampling architecture developed by La Vigne et al. (2021) and the iterated hedge trimmer-based 3D-printed canopy sampling UAS by Krisanski et al. (2022), though they all have low manoeuvrability, are heavy and non-precise. Moreover, UAVs with special manipulators can perform complicated tasks such as artificial plant pollination (Abutalipov et al., 2016). Recently developed, agile and highly dynamic Grasper drones also have potential in forest research (Roderick et al., 2021).
Here, we present a novel, multifunctional sUAS, called the Druid Drone (DD) for precision forestry based on commercially available lightweight UAS platforms. It is equipped with a manipulator with five 3Dprinted interchangeable functional units for close observation, efficient sampling and pest spraying within a tree crown, which were assessed in Czechia and Ukraine. Among the five specific units of DD are the Observer for close, non-invasive inspection of the canopy, empty bird nests, mistletoe and other objects; the Collector for arboreal arthropod trapping; the Sampler for soft tree tissue harvesting; the Secator for harder tree tissue cutting; and the Green Doctor for targeted remote canopy spraying with growth regulators. Overall, our joint approach offers a user-friendly, precise and safe solution for tree care, plant pathology surveys and biodiversity research in tree crowns. It is particularly helpful for ecological studies in hardly accessible sites such as the upper layers of forests, the periphery of tree crowns and rocky slopes.

| Commercial drone platforms and their specifications
Three common, off-the-shelf drones from the A3 open class category manufactured by DJI (China), namely Mavic 2 Pro, Mavic Air 2 and Phantom 3 Standard, as well as Rotorama Spectre custom drones (Rotorama, Czechia), were chosen as stable and widely available manipulator platforms (Figure 1; Methods S1; Table S1).

| 3D printing of the manipulator base
All functional units were connected by one or two carbon or aluminium tubes (outer diameter: 8 mm; inner diameter: 6-7 mm; lengths: 33, 40 and 50 cm, masses: 9-13.5 g. The length of the Observer and

| Operation mode and vision-based navigation
The DD must be operated by at least two people: a pilot controlling flying of the DD in the visual line of sight (VLOS) and an operator ( Figure 3), manipulating the functional units using an updated remote control transmitter. In a more complex environment, we prefer using EVLOS (Extended Visual Line of Sight), when the operator is standing closer to the tree and communicates with the pilot in order to direct DD flight and manipulations. However, BVLOS (Beyond Visual Line of Sight) mode of operation, especially within the tree canopy, is risky.

| Prototype validation
The DD prototype with all manipulators was tested under field conditions in Czechia (Olomouc, 'Holický les') and Ukraine (the

| RE SULTS
Testing the chosen drone platforms under controlled conditions (see Methods S3) revealed that they could lift almost half their mass as a payload. For example, the DJI Mavic 2 Pro can lift 400 g, with a maximum operating time exceeding 10 min (defined by a low battery warning at 25% of the battery charge). Based on this, we measured the flight duration of the selected drone platforms with different payloads (0, 150, 300 and 600 g), hovering them at a distance of at least 1 m above the ground to eliminate deviation caused by ground effect, an aerodynamic phenomenon, which reduces the thrust requirement close to the ground (Table 1). The selected payloads correspond to the masses of individual units.
The mass of the Observer and Collector are ~102 g and ~92 g, respectively, while they can be increased up to 300 g when using other types of camera or the block with sticky tape, which are heavier than those used in our prototype. In turn, the masses of the Sampler and Secator are ~473 g and 491 g, respectively, while they can be increased to 600 g in case of using heavier sampling and cutting units. As for the Green Doctor, the mass of the empty unit is ~550 g, and that of the unit filled with water-dissolved growth regulators is ~650 g.

F I G U R E 2
Block scheme of the Green Doctor unit. Indications: Blocks: blue-electronic devices, orange-power blocks, green-mechanical parts, yellow-optional video transmitters. Lines: blue-commands transmission, orange-electric power demand, green-mechanical connection or liquid/air flow, yellow-video signal from camera. Considering the reduction in the drone's maximum flight time with increasing payload (Figure 4), all DJI drones can still fly for more than 7 min with a 600-g payload, sufficient to perform tree research and care operations without battery change.
At the same time, increased payloads affect the drone's motor temperature, which may reach the overheating values in the DJI Mavic 2 Pro and especially DJI Mavic Air 2 platforms bearing maximum payloads. However, motor temperature is less responsive in the other platforms tested (Results S4). The field validation of the DD performance with mistletoe case studies is discussed in Results S5. Based on these results, a drone-based multilevel mistletoe management strategy was proposed ( Figure 5).
This six-step well-balanced approach was developed for regular surveys of the European mistletoe (Viscum album subsp. album) infestations and can be utilized in a variety of tasks related to mistletoe sampling, control, general tree care and biodiversity research in tree canopies based on Results S5-S9.

| Applicability of the DD interchangeable functional units
The combination of several tools on one manipulator represents an all-in-one approach for accurate canopy sampling and observation, rapid phytopathological surveys of trees and tree management, e.g. by targeted spraying with growth regulators. This multifunctional drone is a valuable instrument for specialists in a variety of fields, such as botany, dendrology, mycology, microbiology, phytopathology, entomology, ecology and landscape management. The undamaged, mostly intact material freshly collected directly from a tree without soil cross-contamination is well suited for further processing for in vitro cultivation, microscopy, DNA or RNA extraction, production of microbial cultures and research for potential biocontrol agents. DD-based targeted spraying with contact herbicides, using the Green Doctor, has exhibited promising results for the eradication of mistletoe in an economical and environmentally friendly manner in highly infested tree stands (Varga et al., 2014).

| Benefits of the DD
The DD is a compact, lightweight, publicly available, non-industrial UAV that does not require a specially prepared runway and can perform vertical takeoffs and landings in complicated terrain as compared to other sampling platforms (Charron et al., 2020;Käslin et al., 2018;La Vigne et al., 2021). It has zero fuel consumption owing to rechargeable batteries and offers a cost-and effortefficient mode of canopy sampling and spraying compared with tree climbing techniques and/or automatic mobile aerial platforms (Barker & Pinard, 2001). The DD complements the work of arborists considerably because it operates on thin peripheral branches that are otherwise inaccessible. Furthermore, it can be used in complicated landscapes, such as trees in tropical forests or trees growing on rocky slopes, which cannot be reached by mobile aerial platforms.

| Limitations of the DD
The key limitations are country-specific regulations, obligatory pilot certification and the ban on drone flights in city areas without special permits. Despite good manoeuvrability in the tree crown periphery, the DD is more vulnerable to mechanical obstacles due to disabled sensors; therefore, special pilot training is required. If the DD gets stuck in branches, it must be released either manually by a long telescopic rod or using another drone with a rope and hook.
The operability of the DD remains limited under insufficient visibility and non-favourable weather conditions such as wind (more than 10 m.s -1 ), rain, snow, mist, fog, etc. Hindered GPS signal by mechanical/electromagnetic obstacles or a lack of GPS satellites affects the geo-positioning of the DD and limits its operational capability. Both The language was professionally corrected by Editage.