Unmanned Systems Technology 036

35 achieving higher accuracy and reliability over time amid interference. While those perceptions still hold true in some respects, recent innovations in hardware and software are blurring those lines. MEMS advances New techniques for the design, manufacturing, accuracy of calculations, calibration, testing and more have been developed for MEMS IMUs over the past few years. This has been a natural response to booming demand from autonomous vehicle manufacturers for higher performance in navigation and gimbal pointing, as well as savings in cost, size and weight. Image stabilisation in particular is driving MEMS systems developers to prioritise high data rates, high bandwidth and low latency in their data output, along with continuous minimisation of bias- and error-related parameters and wider dynamic measurement ranges in their sensing. For example, high-end MEMS navigation IMUs are now capable of measurement ranges of up to ±10 g of acceleration and ±475 º /s in angular rate. At the same time, accelerometers for a camera gimbal IMU might have a range upwards of ±36 g , and its gyroscopes might be designed for up to ±2000 º /s. Processor power improvements have increased measurable dynamic ranges of inertial data and enabled better layers of corrections for improving the speed and accuracy of IMUs, and their ongoing miniaturisation is motivating IMU manufacturers to package higher numbers of processors into their systems’ enclosures. That has driven down computing times and latencies for data outputs without increasing the sizes of products, enabling drop-in replacements. Also, a wide range of micro-sized structures are now used as piezoelectric sensing elements, with different manufacturers opting for different MEMS topologies – including vibrating rings, comb shapes, springs and ‘butterfly’ shapes – to achieve their design goals. Rather than standardising around one single ‘superior’ MEMS structure shape though, it has become increasingly common for high-end manufacturers to conduct internal r&d to characterise and optimise the MEMS topologies in their IMUs, thus enhancing their own base architectures and performance levels, rather than waiting for their suppliers to do it for them. Some components are still highly sought after for navigation-grade (and higher) IMUs. For example, closed-loop accelerometers are indispensable for their consistently high performance levels across wide sensing ranges. By contrast, open-loop accelerometers typically have to strike a trade-off between these characteristics owing to a lack of feedback source in their capacitive sensing element. In addition to hardware optimisations, software improvements have fostered ever-higher rejection of ‘noise’ (such as from shock, vibration and temperature), new lows in latencies, and generally smoother and smarter fusion of data from sources such as GNSS, Lidar, infrared or vision – largely delivered to end-users by over-the-air firmware updates. All this has resulted in high-end MEMS gyros capable of 0.002 º /s in angular random walk; 0.003-0.005 º /s was the benchmark just a few years ago. Given how many of these advances have come from regular development cycles and rigorous software engineering, rather than any landmark technological breakthrough, it is not unrealistic to expect such progress to continue. Beyond the numbers typically seen on datasheets, however, IMUs for defence- and industrial-grade UAVs must also be programmed with IMUs, gyros and accelerometers | Focus Unmanned Systems Technology | February/March 2021 IMUs require higher accuracy and reliability for payload integrations such as geo- referencing or tracking than navigation systems (Courtesy of Applanix) Developers of MEMS systems are prioritising high data rates, high bandwidth and low latency, along with wider sensing ranges