Issue 39 Unmanned Systems Technology August/September 2021 Maritime Robotics Mariner l Simulation tools focus l MRS MR-10 and MR-20 l UAVs insight l HFE International GenPod l Exotec Skypod l Autopilots focus l Aquaai Mazu

84 assistance with far greater agility than many typical software companies. Although there are some differences in open-source licences nowadays, which set differing requirements for how extensively a developer must make their code available to customers, using an open-source autopilot can be highly advantageous for professional users who want to quickly incorporate mission-critical modules and interfaces into their vehicles. Naturally that can include solutions for integrating new cameras, IMUs, transponders and so on, including enabling diversity in comms networks, to enable the use of MAVLink, subsets of CAN and Ethernet across a single vehicle. Indeed, one of the major reasons behind the widespread use of ADS-B Out among UAV manufacturers for collision avoidance – despite having no regulatory requirement to do so – was the early adoption of the technology among the open-source community, which has produced a wealth of resources for quickly integrating ADS-B transponders with their autopilots. That also makes it much easier for manufacturers of payloads, batteries, transponders and other systems operated by the autopilot to implement their systems with the autopilot developer’s source code. By contrast, a proprietary autopilot manufacturer would need much more ‘back and forth’ to ensure sufficient sharing of interfacing information. However, the open-source community also increasingly provides higher forms of application-critical software. These include new and robust flight logic behaviours for flying amid skyscrapers or airports, new forms of data processing or analytics, solutions for cloud connectivity, and modules for AI or machine learning to advance autonomous systems in a particular direction. And as commercial unmanned systems look set to take on tasks in congested urban streets and airspaces, fixes for edge-case issues such as rapid object detection and avoidance, landing or integrating high-power electric motors and motor controllers can be sourced quicker than ever. Developments in flight control architectures oriented around unconventional airframes are particularly prominent in the open-source community, with resources for tiltrotors, tail-sitters, and so on that can be rapidly programmed and prototyped now being available. Even in the defence world, open- source systems are gathering attention. While that might seem counter-intuitive given the military’s perpetual need for secrecy about the technology it uses, the ability to quickly crowdsource solutions to problems relating to GNSS jamming detection, cyber security or other threats is increasingly viewed as more important than worrying about open access to such solutions. The logic here is that if points of weakness in software can be fixed faster than they can be exploited, there’s no need to be concerned about the damage that may be caused. To enable their communities to develop and share solutions confidently, open- source autopilot manufacturers have to supply a stable, accessible and familiar platform upon which independent programmers can develop their fixes and features, often using a well-established microcontroller at the core of their system, such as an STM32 or similar. While open-source autopilot software can typically be downloaded, installed and run on very low-power architectures, providers continue to develop their August/September 2021 | Unmanned Systems Technology Many major organisations are eyeing open-source autopilot architectures with interest, for their modularity and the agility with which they can be upgraded and improved (Courtesy of Auterion) Open-source autopilot software can quickly be updated for operating in urban canyons, advanced data analytics, performing VTOL transitioning and more (Courtesy of CubePilot)