Autopilots | Focus operated. These are multirotors, fixedwing aircraft, VTOL-transitioning aircraft, attritable UAVs and USVs. To take a version of software designed especially for one of these vehicle types, and change it to control another one takes immense time and effort, not purely in development but also in the exhaustive simulated, laboratory and field tests necessary for validation. Additional sub-categories and nuances exist within each vehicle category, but they tend to boil down to small functions that developers routinely adjust for without too much difficulty. For instance, a company might design a fixed-wing UAV autopilot for systems to be launched by catapult and recovered via parachute landing, but then need to change 10% of the software code at the most to adapt it for UAV integrators who want to take off from and land on a runway. To design the autopilot correctly for each customer’s needs it is common to start with a request for information (RFI) on their platform, particularly its configuration, the layout of control surfaces, its operating envelope, and its modes of launch and recovery. Compiling these on a sheet of paper defines the customer’s functional requirements. The autopilot maker can then identify any potential risks that must be mitigated in the autopilot’s functions, how many units must be delivered (and hence how far its design must be tuned for automated manufacturing), the timeframe in which they must be delivered, and hence an estimate of the high-level design adaptation work to be done. Going deeper within those levels requires additional information to cater for vehiclespecific qualities. Ensuring controllability, for instance, mandates details on maximum take-off weight (MTOW), expected moments of inertia, the size of control surfaces and expected vibrational frequencies from subsystems. Customers’ plans on vehicle structures and subsystem mounting may be crucial for spotting conflicts or trade-offs; for instance; if high computational power is needed in a hermetically sealed fuselage, and then airflow cooling could become impossible. With such information, autopilot engineers can proceed to define appropriate parameters for the use-case, enter them into a hardware-in-the-loop simulator, and start testing and iterating towards an embodiment that will result in the vehicle behaving as desired. Trends likely to test such development processes and design loops in the years ahead include newer generations of uncrewed vehicles that move faster, with higher dynamics than current software control loops can handle. Autopilot designs with high levels of modularity and open architectures will be best placed to adapt for these operational requirements, as well as for regulatory dictums on how software modules and hardware components can achieve critical safety ratings. Towards bulk manufacturing Manufacturing and validating autopilots is not dissimilar to the mass production of most other high-end electronics, save for some specifics relating to quality control such that safety ratings are met for aviation authorities, according to pertinent standards such as DO-254 or EMAR 21, depending on the use-case. The process begins with sourcing components from distributors, with products authorised for use in the industry and optimised for bulk supply, with additional checks of component quality being a vital intermediate step. To link these components the autopilot maker needs a verified manufacturer of high-density, interconnect printed circuit boards (HDI PCBs) – not an easy find as such boards are extremely complex and not all manufacturers can deliver them to the quality needed for hitting the higher SAILs or other aviation standards. Manual visual inspections of the received boards can be painstaking, but also hugely valuable to ensure they have arrived clean and with good copper cladding, grounding, surface treatments and so on. Once satisfactory, the HDI PCBs can be put through high-speed production lines, with automated assembly and optical inspections. Such lines can be powered with standard manufacturing systems, but very tight tuning, control and integration can be critical to prevent imperfect placements or alignments of board-mount components on the PCBs, or anything else that would fall short of the targeted level of certifiability. The assembled and quality checked board modules can then be flashed with the basic software and firmware 93 Uncrewed Systems Technology | April/May 2024 Customising an autopilot design involves a lot of back and forth on the details, including vehicle configuration, and time and volume of delivery (Image courtesy of Airvolute)
RkJQdWJsaXNoZXIy MjI2Mzk4