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

point for further tuning and refinement. Measured parameters such as speeds, roll rates and bank angles are programmed into the autopilot as limits. It is then told how much of the actuators’ authority it can use to achieve the desired behaviour. Thilderkvist says the autopilot contains a hierarchy of PID controllers that can directly drive control surface deflections, command pitch according to a desired climb rate, or command a climb rate based on a desired altitude. Likewise in roll, commanding the control surface to a specified angle of deflection, a roll command based on a desired heading or a heading command based on a desired track are all possible. “All of that requires a lot of tuning based on how your aircraft behaves, on its physics,” he says. “That’s the main part of my flight control job – getting the aircraft to fly well with that number of controllers in it, fly straight, turn left and right, climb and descend. It gets more complicated though when you want it to land.” Starting with the pilot’s landing, based on how he uses throttle and pitch, Thilderkvist first replicates it as closely as possible, then challenges it to see if the autopilot can achieve a shorter, safer and more accurate landing. He changes a few parameters at a time then repeats the procedure. The system’s redundancy won’t be tested in the air by deliberately failing any of the channels, as Guy is keen to avoid what he calls “test bleeding”. It could be done, he says, by pulling pins out of connectors, cutting wires or putting a switch in the line, but he considers that forcing it to switch to the next channel in such a way in flight is too risky. “That’s something that really needs to be done on the bench.” Guy reckons that the FX450 has about two months of flight testing ahead of it, assuming that it will typically fly for three days a week, weather issues and aircraft snags permitting. “We’ll go out in a mid-weight band first where we’re pretty sure it will fly right,” he says. “Then we’ll burn down to a light weight, and save maximum gross weight tests for last.” When the team is happy with the way the FX450 flies, the plan is to go up to the Andoya Space Center at the northern tip of Norway, with which Robot Aviation has a partnership. Guy likens Andoya, from which rockets are fired and UAVs flown, to a mix of several US test centres, such as Edwards Air Force Base, NASA Dryden (now Armstrong) and White Sands missile range. “Probably in the winter time or early next spring we’ll deploy up there,” he says. “That’s where you can really push the button and fly a 20-hour mission on full auto. I don’t feel I need that for a production decision, but it’s nice to do an end-to-end check like that.”