Issue 55 Uncrewed Systems Technology Apr/May 2024 Sellafield’s UAV equipment l Applied EV Blanc Robot l Battery tech l Robotican’s Goshawk l UGVs l UAVHE RW1 rotary l Roboat UVD l Autopilots l Arkeocean UVD l UMEX 2024 l CycloTech UVD

94 Focus | Autopilots for vehicle control, navigation and other functions, before being assembled into complete modules with enclosures, cooling fans, ruggedisation and so on, per the design specifics of the product. Following this could be rounds of testing, from visual checks to automated test procedures of all physical interfaces and separate functionalities, to random or particular selections of units for deeper quality checks, such as shock, vibration, EMI and thermal chamber-test regimens, defined and well-established across various military and aerospace standards, as well as flight tests and simulator tests. Many manufacturers find testing throughput to be a production bottleneck, to the point that some are circumventing it by building test rigs that hold scores of autopilot modules at a time, all connected to a measurement device or simulator, such that batches upwards of 50 units can be subjected to intense shock, cold, heat and other extreme conditions at once in controlled and repeatable conditions. Standards in software The evolution of standards to which software programming must adhere is a subject that autopilot manufacturers pay close attention to, affecting guidelines for the MAVLink protocol that is central to many autopilots, as well as for payload buses, compatible board configurations and application-specific protocols, such as military protocols for the integration and connection of UAV autopilots, subsystems, payloads, ground control stations and so on. For instance, the new RAS-A protocol, for use in systems for the US Department of Defense, has been designed to improve over MAVLink in vital ways, such as better handling of system IDs across network layers through point-to-point communications. Autopilots are also being programmed for more application-specific integrations, such as new cameras prized by industrial mapping and photogrammetry customers. Software standards for interfacing with them are being established to smooth the work needed for end-users to be able to plug in such cameras and still see their data in their user interface. The same can be said for components such as data links and ECUs, as each customer may prefer a different solution. Hence, an API and drivers for each of them are key to swapping and configuring them in a straightforward manner within a unified interface. Coding software patches and features for such functions is a vital step towards minimising the degree of training needed for new uncrewed systems technicians and operators. While many tools are available for automating code production, and ensuring software follows regulatory and mission requirements, good procedures and well-trained programmers remain key to robust software with multiple features and minimal error. Other assets are also vital for reliable, smooth running. High-end compilers, for instance, can be expensive, but tend to be updated frequently to prevent bugs or insufficiencies that can hamper the throughput and quality of autopilot code. On top of this are other useful tools for optimising the back end of software, such as code analysis solutions and cyclomatic complexity verification systems, which can gauge the complexity of a batch of code and indicate places where five lines of code could do the job of 15. Future of flight (and ground and sea) control In addition to certification standards continuing to evolve and define the requirements of autopilot design, engineering, manufacturing and testing, ongoing economic concerns such as labour shortages and the need to cut operating costs are likely to compound with defence services’ hunger for autonomous force multipliers. This would result in a gradual transition towards fleet- and swarm-oriented autopilot configurations, with each autopilot not only designed for mass production at reduced cost, but also to function intelligently and intuitively enough to take much of the complexity of operating (or monitoring) uncrewed systems out of operators’ hands, decreasing the training and number of personnel required per uncrewed vehicle. The emergence of standards on fleet management and swarming is inevitable, although they may take time to be formed. In lieu of those, one can expect next-generation autopilots to leverage a variety of technologies crucial to April/May 2024 | Uncrewed Systems Technology Test rigs like this one can enable high-throughput batch testing of autopilot; testing rounds can otherwise form a bottleneck in autopilot production (Image courtesy of CubePilot)