Thông tin tài nguyên

Steps to build a DIY low-cost fixed-wing drone for biodiversity conservation

pmc-PMC8363011.pdf · Phiên bản 1 · 17 trang

18
Bản gốc
Trang 8 · Đoạn 19Vị trí nguồn

Step 7: Performance tests Performance tests were carried out for proving the drone’s ability to cover an area of up to 1 km2 with a minimum spatial resolution of 3 cm / px with only one battery. The drone flight tests were performed in July 2018, 15:00–18:00 h local time in different areas within the north- ern region of Maranhão state, Brazil. We performed 16 test flights (four conducted in manual mode and 12 in auto mode) to assess flight autonomy, communication range and resolution of aerial images. Before flight missions we followed safety procedures regarding the operator, drone stability and the protection of others involved (See S1 Checklist). During the four flights in manual mode, we tested the aerodynamics, control surfaces (aile- rons, flap, rudder and elevator) and engine. All manual flight manoeuvres were performed in VLOS (Visual Line of Sight), <500 m from GCS, and the take-off and landing were performed manually. The 12 flights in auto mode aimed to check the autopilot, pre-programmed flights, the telemetry RFD 900, GNSS and the other external sensors (compact RGB camera and 3DR Digital Airspeed). All auto mode flights were performed in VLOS or EVLOS (Extended Visual Line of Sight). For EVLOS flights to test RFD 900 long-range telemetry, it was necessary to have a second pilot with a second RC connected (binding) to the drone’s transmitter. In these flights, this pilot followed all trajectories of the drone beyond the visual line of sight of the main pilot, by moving parallel to the drone’s trajectory by car. For this, we strategically choose an open field

Trang 8 · Đoạn 20Vị trí nguồn

adjacent to the road outside an urban area that allowed the second pilot to travel by car as well as to watch the drone during the entire flight path. In direct communication with the main pilot, the second pilot was able to manoeuvre the drone with the second RC in case of any eventual problem. This type of logistics on EVLOS flight tests was necessary both to avoid the loss of the drone due to possible connection and failsafe failures and to comply with local civil aviation legislation. Although the second RC signal was always within the range of 500 m throughout the trajectory of the EVLOS test flights, we disabled the Failsafe action in QGroundControl in case of RC loss signal, to avoid the automatic return to home of the drone in long-range flights and we enabled the Failsafe action in case of long-range telemetry signal loss. We opted for the execution of “Return mode” action in situations of telemetry RFD 900 loss for more than 10s (See Item 9 in S3 Text). For comparison with other DIY drones, we performed a simple transect flight and law- nmower flight pattern simulating methodologies employed in studies using DIY drones [9, 17]. For transect flights, within the QGroundControl, we programmed a "Corridor Scan" flight pattern consisting of a straight-line flight with a maximum length and telemetry distance of 20 km. The drone was programmed to fly at a constant altitude of 100 m (AGL) and at a speed of 15 m/s. The "Corridor Scan" flight pattern was performed twice within the same area and with

Trang 8 · Đoạn 21Vị trí nguồn

the same parameters (See flight-plan-corridor-scan in S1 Plan). In these flights, we mainly tested the flight range and the maximum telemetry range (Table 1). For the lawnmower flight we programmed a "Circular Survey" flight pattern covering 10 ha (lat: -2.524484˚ / long: -44.208837˚) at 100, 75, 50 and 25 m AGL (See flight-plan-lawnmower in S1 Plan) which was performed in VLOS in order to identify the best flight altitude for dis- tinct objectives (monitoring of anthropic activities, vegetation analysis, fauna and flora identi- fication). The camera was triggered automatically using the “Survey Mode” (See Item 10 in S3 Text) and based on a predefined flight plan to produce at least 70% overlap and side lap among each image. We performed two test flights for each altitude in the same area applying the same parameters, totalling eight flights (Table 1). For the last two test flights, we programmed a lawnmower flight with a grid pattern cover- ing around 1 km2 in a specific area inside the “Area de Protec¸ão Ambiental do Itapiraco´” (lat: PLOS ONE Building a DIY low-cost fixed-wing drone for biodiversity conservation PLOS ONE | https://doi.org/10.1371/journal.pone.0255559 August 13, 2021 8 / 17