The design of the spore capture device was primarily driven by the requirement to open and close a Petri dish remotely, to maintain a seal on the dish when samples are not being collected and to allow for easy swapping of Petri dishes between collection profiles, all while minimizing any impacts on the UA’s flight performance. The design is easily transportable and readily installed with basic tools, facilitating deployment during field campaigns in remote areas.
In anticipation of future deployment, a series of exploratory experiments were conducted to test the deployment of the capture device while mounted to the UA and to gain insight into what parameters impact the capturing of spores.
Apparatus design and hosting
A commercial off-the‐shelf (COTS) quadcopter-style UA (DJI Inspire II) was used to host the spore collection device. The platform lends itself well to the mission requirements with a payload capacity (710 g) that well exceeds the mass of the capture device (145 g), and offers augmented flight stability, ease of use, the ability to hover, and the ability to execute a pre-defined flight plan autonomously. To simplify development, minimize the size of the apparatus, and keep sample collection straightforward, the collection device was designed to be attached to the UA’s retractable landing gear.
Importantly, the design exploits the UA’s ability to cycle the landing gear remotely to open and close the Petri dish, thus removing the requirement for any sort of ancillary servo and associated command and control link. Using this design strategy, the remote pilot can open the Petri dish by raising the landing gear and securely close the Petri dish by lowering the gear. This strategy ensures that spore samples are not collected until the desired portion of the flight plan, and that contamination of the collection medium does not occur during launch or recovery of the UA. To protect the Petri dish during its opening and closing, each side of the dish is secured to the respective portion of the mounted apparatus by Velcro.
The computer-aided design (CAD) drawing of the apparatus is shown in Fig. 1a. Figure 1b shows the apparatus mounted with the gear fully extended and the Petri dish fully closed. The lefthand portion of the apparatus was attached to the UA by reusing screw holes for the UA’s camera gimbal. These two screws securely mount one side of the apparatus while minimizing any additional material weight and simplifying assembly of the device in the field (Fig. 1c). The attachment of the other side of the apparatus to the main gear was accomplished using a flange with four readily accessible screw holes (Fig. 1a). This second attachment, likewise, made for a simple and lightweight connection point to the UA. With weight being the main driver behind most aircraft design considerations, the collection device was printed with standard polylactic acid (PLA) filament and a honeycomb infill to reduce the overall weight. The fully configured UA is shown ready for takeoff in Fig. 1d.
Spore capture trials
Study Area
The experiment was undertaken in an agriculture field located within a humid subtropical climate zone (Daytona Beach, Florida, USA). Figure 2 shows the 31-hectare field where the investigation took place. The field is bound by a combination of woods, composed of both palm and hardwood trees, to the west and south, additional agricultural plots to the north, and wetlands to the east. The topography of the area is generally flat and an interstate highway runs north-south adjacent to the eastern edge of the operational area. The field is used for cattle grazing and is largely composed of pasture grass and was dry over the operational period.
Mission Planning
A suite of 13 flight plans were created to investigate how exposure time, orientation of the Petri dish to the prevailing wind, altitude, and medium hardness impacts the number of spores captured. All 13 flights were flown back-to-back over the period of three hours, hence allowing for a quasi steady state assumption to be made for the ambient environment. The baseline flight plans consisted of horizontal transects defined by either length or time depending on the parameter of interest for the given flight. Flight plans were constructed so that the UA’s nose always remained orthogonal to the flight path vector of the UA and with the agar side of the Petri dish aligned with the flight path vector. This effectively resulted in the UA flying sideways and the agar side of the Petri dish always facing the direction of travel.
Two control petri dishes accompanied each flight. First, another UA was flown in a stationary hover adjacent to the dynamic flight plan of the first UA. The hovering UA hosted a similar Petri dish, with the same open and shut times, as the dynamic UA. Second, an additional Petri dish was set up at ground level, adjacent to the operational area. This Petri dish was housed in an open-top box and placed on a tabletop elevated 1 m above the ground. The test parameters are shown in Fig. 3. A full test matrix with details of each UA’s flight plan and the objective for each flight is available in the Supplementary Material Table S1.
The data collection flights required pre-planned missions for precise control of the aircraft during the flight. These missions were built and executed in the DJI Ground Station Pro (GSPro) iOS application. DJI GSPro allows for the creation of pre-planned GPS waypoint flights, which are autonomously executed on-site. GSPro also allows for control of the aircraft’s flight speed, orientation, direction, rotation speed, and altitude. The use of this software ensured a more consistent approach across all data collection flights.
Sampling Flight Execution
The spore capture device accommodated Petri dishes with a 90 mm diameter and 15 mm depth. Dichloran Rose-Bengal Chloramphenicol (DRBC) was used as a collection and growth medium in the petri dishes, as it is commonly used for impaction air samplers and is effective at selectively isolating airborne fungal spores without bacterial growth [Mentese et al. 2017]. The loading of the Petri dish onto the UA was undertaken carefully so that the dish was never opened and exposed to the ambient environment before data collection began.
After loading the Petri dish, pre-launch checklists were conducted and the UA was launched on the pre-programmed mission. The landing gear legs remained extended throughout ascent to the mission’s first waypoint. Once the UA arrived at the first mission waypoint, it autonomously rotated to place the nose orthogonal to the ensuing flight path vector and retracted its gear, hence opening the Petri dish. During the data collection phase, the UA translated autonomously between the two previously defined waypoints. At the conclusion of the data collection mission, the landing gear was extended, hence closing the Petri dish. Following recovery of the UA, the Petri dish was promptly removed from the UA and sealed with tape.
Colony counting
Petri dishes were kept sealed and incubated in a temperature-controlled environment of 23 ⁰C for 82 hours. Colony-forming units (CFUs) were quantified by counting the number of discrete fungal colonies on each dish after the incubation period.
Statistical analysis
Petri dish colony counts for each experiment were fit to separate generalized linear mixed models for each experiment. Models assumed a Poisson distribution with a log link. All models also included an intercept offset random effect that grouped concurrent samples (i.e., paired ground control, hovering UA control, and dynamic UA flight) to account for shared variability due to environmental conditions or other outside factors (see Equation S1 Supplementary Material). An interaction effect (flight x treatment) was included to test for differences in treatment effects between the dynamic flight and controls. Analyses were conducted with the statistical software R [R Core Team, 2023]. A full description of statistical packages and outputs is given in the Supplementary Material.