Unmanned Systems Technology 021 | Robot Aviation FX450 l Imaging Sensors focus l UAVs Insight l Liquid-Piston X-Mini l Riptide l Eurosatory 2018 show report l Zipline l Electric Motors focus l ASTS show report

32 systems can understand. The second, which he is implementing in the FX450, is to use a central mission computer that will process outputs into a common language. “That makes our life easier,” he says. “We don’t need to change the hardware or write the hardware abstraction all the time.” He adds that the computers in the payloads run operating systems such as Linux, so there are open source libraries that make translation simpler. Integrating comms The software side of integrating data links, Damodaran notes, is eased by their standardisation in Ethernet protocols. That leaves him to concentrate on configuring their RF parameters and optimising antenna installations to achieve the best line of sight to the receiver. A further challenge with antenna placement is avoiding interference between antennas, between the antennas and numerous subsystems, and among the subsystems themselves, which include transponders, telemetry links, video links and, in the near future, satcom. “We are using spectrum analysers to find the interference, then trying to create band gaps between the emitters, to separate them physically as much as possible and to shield them so they don’t bleed RF energy,” he says. Integration, testing and troubleshooting of the smaller FX20 over the past eight months or so have paved the way for the FX450. Damodaran notes, for example, that the data link is working well, having achieved a BLOS flight well beyond visual ranges with the FX20 just before our visit, with a plan to extend that to 30 km soon. He adds that this provides a degree of comfort when moving on to flight tests of the FX450. Test flying About 40 flight test hours have been logged and 100 test points achieved on previous prototypes. Four of them have been built, including the production- standard aircraft that is about to begin its flight test programme. The most critical aspects of any flight test programme are to be found in the standard envelope expansion process, Guy says, and include milestones such as the first flight, determination of stall speed and the initial flight under full autopilot control. Tests conducted on the early prototypes led to refining some configurations and re-sizing some components. Tests on the third and fourth prototypes will focus on regression testing to make sure changes that have been introduced have not brought any new problems with them, and will also cover BLOS and sensor testing. While the number of hours and test points under their belts so far is not large, the smaller FX20 makes use of similar, if not identical, hardware and software. That is particularly so in the autopilot, which is a simpler, single-channel version of the Micropilot in the FX450, and has served as a flying testbed to reduce development risk for the larger aircraft, Guy emphasises. Flight testing is the domain of flight controls engineer Dan Thilderkvist. His main task at the time of our visit was configuring the Micropilot for the FX450. Like many UAV autopilots, it is fairly generic, he says. It uses software- based proportional integral derivative (PID) controllers to fly a wide variety of vehicles, so it has to be tuned for individual aircraft types. The tuning procedure starts with a pilot flying the aircraft manually in full remote control mode and recording its behaviour to verify the accuracy of its sensors. The next step is to program the autopilot to reproduce that behaviour using the way the pilot flew it as a model and as a starting August/September 2018 | Unmanned Systems Technology Dossier | Robot Aviation FX450 The FX450 has about two months of flight testing ahead of it, assuming it will typically fly for three days a week (Courtesy of Robot Aviation)