Inside Unmanned Systems provides actionable business intelligence to decision-makers and influencers operating within the global UAS community. Features include analysis of key technologies, policy/regulatory developments and new product design.
Issue link: https://insideunmanned.epubxp.com/i/720234
56 unmanned systems inside August/September 2016 ugust/Septe August/September 2016 August/September 2016 be August/September 2016 0 August/September 2016 6 August/September 2016 AIR UAS RELIABILITY f light tests were conducted with prior knowl- edge of the time of occurrence of the faults. In the approach taken at the UMN UAV Lab, a fault-tolerant control law is available onboard the f light computer. This fault-toler- ant control law is not engaged until the fault diagnosis module raises a f lag on a detected fault. A full implementation of the system would involve a switching algorithm that dis- engages the nominal control law and engages the fault-tolerant control law after a fault is detected. However, in the f light tests at UMN, the faults were pre-programmed into the test. The nominal controller is a classical loop-at-a- time design: Separate feedback control loops are used to control the longitudinal and later- al-directional motion of the aircraft. Because a def lection of the left elevator induces both pitching and rolling moments, the classical ap- proach can no longer be followed for designing the fault-tolerant controller. Instead, multi- variable feedback control theory is leveraged to enable f light with a single control surface. In particular, two techniques (linear quadratic Gaussian and H-infinity) are applied to design the fault-tolerant controller. 5 The control laws are implemented onboard the Goldy Flight Control System (FCS). Goldy FCS is an open source f light computer devel- oped at the University of Minnesota. Version 1.0 of the Goldy FCS as shown in Figure 5 was used for the f light demonstration. The f light demonstrations include the air- craft performing straight & wings-level and banked turn maneuvers. Figure 6 shows f light test data from when the modified Ultra Stick 120 performed a left banked turn maneuver. In this maneuver, the aircraft is commanded to roll such that its port side wing drops 5 de- grees below wings-level. The desired state of the aircraft is shown pictorially in Figure 6. The horizontal axis shows time and the verti- cal axis shows the actual roll angle of the air- craft. The red, blue and green plots are data recorded on three separate f lights. The black dashed line is the desired roll angle of the air- craft, which is set to -5 degrees three seconds into the f light. After the fault is injected, the controller is reconfigured to f ly with only one control surface. After about 15 seconds have elapsed, the aircraft is close to the desired state.In the best-case scenario, the actual roll angle of the aircraft is within a ±5 degrees band of the desired roll angle. The flight tests highlight the potential for ap- plying robust control theory in designing fault- tolerant controllers. In particular, advanced Figure 6. Actual and Commanded Roll Angles After a Simulated Servo Failure. Results from fl ight tests wherein the modifi ed Ultra Stick 120 is fl own with only a single operating elevator (E1 in Figure 4). All other control surfaces were inoperable. Figure 7. Fault-Tolerant (Analytically Redundant) Navigation and Attitude Determination System.