Issue 39 Unmanned Systems Technology August/September 2021 Maritime Robotics Mariner l Simulation tools focus l MRS MR-10 and MR-20 l UAVs insight l HFE International GenPod l Exotec Skypod l Autopilots focus l Aquaai Mazu

54 Insight | UAVs Heavy lift The advantages of finding and incorporating the ideal flight controller cannot be denied. To that end, Kansas- based Flint Hills Solutions (an unmanned aerial services company) has partnered with UAV Navigation for detailed integration and support of the latter’s Vector autopilot, among other avionics products. The use of UAV Navigation’s Mil- qualified avionics systems is aimed at optimising the flight safety and precision of Flint Hills Solutions’ FH-900 helicopter UAV in applications such as heavy package delivery missions, infrastructure inspections, agricultural spraying and construction project surveys. The all-electric FH-900 helicopter has a MTOW of 67 lb (30.4 kg) and an empty weight of 20 lb, meaning a payload capacity of around 21.4 kg. Its payload bay has an open architecture for installing and swapping survey sensors, chemical dispensers, and package protection and dropping mechanisms. It can fly for up to an hour between recharges, and has a top airspeed of 60 mph (96 kph), with BVLOS being a key goal of the collaboration with UAV Navigation. An FAA waiver can be obtained for end-users requesting BVLOS operations of the craft. Warehouse inventory management San Francisco-based Ware has developed a pallet rack inspection solution around the Skydio 2 quadrotor to enable autonomous inventory management, with an eye on remedying the shortage of labour availability for such duties. “With so many huge warehouses concentrated in industrial zones, it’s hard to get hold of employees willing to inspect the inventory in these incredibly tall pallet racks hour after hour, making it almost impossible for warehouse managers to know where everything is,” says Ian Smith, CEO of Ware (echoing the sentiments of Exotec – see page 72). This generates considerable human error, creating the risk that hundreds of thousands of dollars’ worth of stock could go missing and perish. UAVs however can fly up, down and across racks quickly, removing the risk of employees falling from a scissor lift, say, or of them being poached by a rival warehouse. The only danger comes from navigating indoors through confined spaces without GNSS for localisation. “That’s the major reason we’ve partnered with Skydio – their UAS is 30- 40 times faster than humans at tracking inventory throughout warehouses,” Smith explains. “One Skydio 2 craft autonomously takes pictures of thousands of pallet locations per day; we then process those and use the data to accurately update customers’ warehouse management systems [WMSs].” Selecting the Skydio 2 came down not only to its picture quality and rapid data-gathering capability but also its indoor vision-based localisation, which Smith and his team saw as being the biggest hurdle to successful UAS-based warehouse management operations. The Skydio 2 integrates six 4K cameras, each with a 200 º FoV to give it a 360 º sensing capability at high resolution, as well as a gimballed 4K 60 fps 12.3 MP camera on its front for a more detailed image and video capture in industrial operations. An Nvidia Tegra TX2 embedded computing board with a 256- core GPU is installed for rapidly processing and fusing this data, to generate real-time 3D models of its surroundings, thereby enabling autonomous indoor navigation and localisation. Its 3D world model is updated with upwards of a million new point measurements per second, ensuring obstacles can be avoided in all three axes. Nine custom deep learning networks are embedded on the TX2 processor for tracking up to 10 objects (such as people or vehicles) at a time. The Skydio 2 can fly for up to 23 minutes between charges, and lands autonomously using its downward- facing 4K camera. “The TX2 would be overkill for what a lot of UAVs are trying to do these August/September 2021 | Unmanned Systems Technology Ware is using the Skydio 2’s intelligence and vision-based navigation to conduct autonomous warehouse management (Courtesy of Ware)