UST030

81 Flight The exact distance flown was 1.7 km in a straight line, at about 25 kph, from the backyard of the Thai restaurant to a selected drop-off point. Also an altitude of 50 m was selected to keep clear of buildings and other ground obstacles (the UAV was subsequently required to fly at that altitude, it having been specified in the permit application). As mentioned, the flight was largely autonomous, and therefore uneventful for the team. “For safety purposes, we included a 4G connection to a laptop running QGroundControl, so that we could monitor and take control of the UAV if needed, but that wasn’t an active part of the test,” Gising says. “Also, normally the GCS computes the RTK corrections stream; instead we built a standalone box hosting the RTK base, a battery and a 4G-connected Raspberry Pi that pipes the raw data to the drone, where the RTK corrections are computed and injected in the MAVLink stream between the DSS and the autopilot.” At the delivery waypoint, the UAV slowed to a hover, where Lundqvist was waiting and playing the part of the delivery recipient. There, he used his smartphone to access the web server (which would be replaced with an app or tracking web page in commercial operations) to confirm with the UAV that the area was clear and that it was safe for it to descend. “During the landing, if someone unexpectedly enters the landing zone, with one click he could execute a Return to Launch script to prevent accidental injury to that person,” Gising explains. “Those Land and Return buttons were the only two options built into the application at his end.” Upon executing the landing script, the Tarot 960 landed and autonomously commanded the gripper to release the food onto the ground, with the box already touching the ground at the same time as the UAV’s landing struts. After the drop-off, the UAV autonomously followed its track back to the original launch point for landing and mission completion, as would be expected during warehouse duties. “The exact application design depends on the infrastructure for whether the UAV is conducting warehouse-to-warehouse logistics or last-mile deliveries to consumers,” Lundqvist says. “We’ll probably design a ‘pre-launch’ app for the warehouse-based UAS technician, and a landing app for consumers, but warehouse-to-warehouse missions can be autonomous from end to end, with the planning and scheduling software handling UTM permissions and flight routing.” The return flight was faster than the delivery (at about 40 kph), owing to the absence of the payload box and its poor aerodynamics. Once at the launch site’s GNSS coordinates, Gising executed the landing script to confirm that the landing area was clear. A few subsystem variables were inspected – including the temperature of the motors and electronic speed controllers, and general visual checks for damage – and the UAV was taken back to the workshop. For the future Further automation of the UAV’s operation is expected as the warehouse project advances, potentially including automated battery-swap stations to reduce maintenance overheads. RISE anticipates that advances in UTM will reduce the scale of preparation needed for commercial UAV flights, especially through automation of air traffic routing and permissions distribution. They also intend to explore applications for swarming technologies for logistics, or other industries if investors request them first. “For the fully automated warehouse logistics side of the project, we will probably use either an EO/IR payload on the UAV combined with an Nvidia Jetson processor to confirm when flight routes and landing zones are free of people, or a set of fixed terrestrial cameras could transmit a signal to the UAV to let it know it’s clear to land,” Gising adds. With many routes to fully autonomous logistics available, RISE can be expected to release a highly optimised delivery solution to complement the level of integration it has achieved with this project. Unmanned Systems Technology | February/March 2020 UAV: Tarot Electro-permanent magnet payload gripper: NicaDrone Motors: T-Motor Electronic speed controllers: Hobbywing (XRotor) Batteries:  Tattu ADS-B receiver:  uAvionix Image processor: Nvidia Some key suppliers RISE’s Drone Safety System enabled quick access to a range of key scripts for testing, preparing and commanding the UAV throughout the operation