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

98 PS | Alternatives to GNSS G NSS technologies are essential infrastructure that provide positioning, navigation and timing (PNT) services without which the unmanned and autonomous systems world as we have come to know it simply would not work (writes Peter Donaldson). They are not perfect though of course, so in May this year the US Government Accountability Office (GAO) published a review of the Pentagon’s efforts to develop complementary PNT technologies to provide more resilience. The GAO’s review provides a useful inventory of the available alternatives to GNSS signals, their strengths and weaknesses and state of development. These technologies are divided into sources of relative and absolute PNT information, the former with reference to an original known state after a period of time and the latter based on live measurements of the real world. Inertial sensors and clocks are relative PNT sources that are self-contained, but measurement errors add up over time and need resetting by GNSS or some other reference. The US Department of Defense (DoD) is therefore investigating new, more accurate and cheaper MEMS inertial sensors, along with atomic sensing techniques. The latter include a thermal beam atomic sensor that produces a stream of heated atoms and uses lasers to measure their motions in response to platform acceleration and rotation. While free from mechanical noise though, they are difficult and expensive to make, and are sensitive to temperature changes, for example. On the timing front, the DoD is investigating several options such as chip-scale atomic clocks, which are promising but currently 100 times less accurate after one day than GPS. Performance-enhancing efforts here are centred on better algorithms. The DoD is also researching optical clocks that promise GPS levels of timing accuracy. Like traditional atomic clocks, optical clocks use the natural oscillations of atoms as a pendulum, but use visible laser light rather than microwaves in the resonator to tune the clock more precisely to the atoms’ natural frequency. The best atomic clocks have an error of about 1 second in 100 million years, while optical clocks are expected to reduce that to 1 second in 15 billion years. Absolute positioning systems rely on sensors to identify distinct features in the environment to compare with stored maps. Novel celestial navigation techniques measure stars and sunlight reflected from satellites with reference to star maps and satellite orbit data, but they must be updated on a scale of days to months. An accuracy of 50 m is quoted, but the technique is susceptible to cloud cover. Variations in the Earth’s magnetic field can also be measured and compared with maps to find positions accurate to about 100 m. However, obtaining global magnetic maps is difficult, and magnetic ‘noise’ from the platform must be compensated for. Terrestrial image analysis is another absolute technique of interest. Here, algorithms analyse camera images to identify landmarks in geo-referenced databases and track changes in their relative orientation. This has demonstrated 10 m accuracy, but the technique is limited by weather and environments without fixed features, such as the open ocean. The DoD is also looking into the use of non-GPS radio signals, including purpose-built networks and ‘signals of opportunity’ such as commercial satellites, TV transmitters and so on. Using multiple frequencies makes this approach harder to jam or spoof, and the signals are often stronger than GNSS. None of these technologies is likely to be sufficient alone, so the DoD is pursuing an open architecture approach to their combined use. Now, here’s a thing “ ” August/September 2021 | Unmanned Systems Technology The error in the best atomic clocks is about 1 second in 100 million years, while optical clocks are expected to reduce that to 1 in 15 billion years

RkJQdWJsaXNoZXIy MjI2Mzk4